Nondestructive Characterization and Imaging of Wood

Springer Series in Wood Science
Editor: T. E. Timell
Springer-Verlag Berlin Heidelberg GmbH

Springer Series in Wood Science
Editor: T. E. Timell
L. w. Roberts/p. B. Gahan/R. Aloni
Vascular Differentiation and Plant Growth Regulators (1988)
C. Skaar
Wood-Water Relations (1988)
J. M. Harris
Spiral Grain and Wave Phenomena in Wood Formation (1989)
B. J. Zobel/J. P. van Buijtenen
Wood Variation (1989)
P. Hakkila
Utilization of Residual Forest Biomass (1989)
J. W. Rowe (Ed.)
Natural Products of Woody Plants (1989)
K.-E. L. Eriksson/R. A. Blanchette/P. Ander
Microbial and Enzymatic Degradation of Wood and Wood Components (1990)
R. A. Blanchette/A. R. Biggs (Eds.)
Defense Mechanisms of Woody Plants Againts Fungi (1992)
S. Y. Lin/C. W. Dence (Eds.)
Methods in Lignin Chemistry (1992)
G. Torgovnikov
Dielectric Porperties of Wood and Wood-Based Materials (1993)
F. H. Schweingruber
Trees and Wood in Dendrochronology (1993)
P. R. Larson
The Vascular Cambium: Development and Structure (1994)
M.-S. Ilvessalo-Pfiiffli
Fiber Atlas: Identification of Papermaking Fibers (1995)
B. J. Zobel/J. B. Jett
Genetics of Wood Production (1995)
C. Matteck/H. Kubler
Wood - The Internal Optimization of Trees (1995)
T. Higuchi
Biochemistry and Molecular Biology of Wood (1997)
B. J. Zobel/J. R. Sprague
Juvenile Wood in Forest Trees (1998)
E. Sji.istri.im/R. Alt!n (Eds.)
Analytical Methods in Wood Chemistry, Pulping, and Papermaking (1999)
R. B. Keey/T. A. G. Langrish/J. C. F. Walker
Kiln-Drying of Lumber (2000)
S. Carlquist
Comparative Wood Anatomy, 2"d ed. (2001)
M. T. Tyree/M. H. Zimmermann
Xylem Structure and the Ascent of Sap, 2nd ed. (2002)
T. Koshijima/T. Watanabe
Association Between Lignin and Carbohydrates in Wood and Other Plant Tissues (2003)
V. Bucur
Nondestructive Characterization and Imaging of Wood (2003)
Voiehita Bueur
Nondestructive
Characterization and
Imaging of Wood
With 201 Figures, 24 of them also in color and 49 Tables
i Springer
Prof. VOICRITA BUCUR Series Editor:
Institut National de la Recherche T. E. TIMELL
Agronomique State University of New York
Centre de Recherches Forestieres de College of Environment Science and
Nancy Forestry
Laboratoire d'Etudes et Recherches Syracuse, NY 13210, USA
sur le Materiau Bois
54280 Champenoux
France
Cover: Transverse seetion of Pinus lambertiana wood. Courtesy of Dr. Cari de Zeeuw, SUNY eollege
of Environmental Scienee and Forestry, Syracuse, New York
ISSN 1431-8563
ISBN 978-3-642-07860-6
Library of Congress Cataloging-in-Publication Data
Bucur, Voichita.
Nondestructive characterization and imaging of wood / V. Bucur.
p. cm. - (Springer series in wood science; 760)
Includes bibliographical referenees and index.
ISBN 978-3-642-07860-6 ISBN 978-3-662-08986-6 (eBook)
DOI 10.1007/978-3-662-08986-6
1. Wood-Testing. 2. Non-destructive testing. 1. Title. II. Series.

TA420.B92 2003 620.1'27-dc21
This work is subject to copyright. AII rights reserved, whether the whole or part of the material is con-
cerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this pub-
lication or parts thereof is permitted only under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be obtained from
Springer-Verlag Berlin Heidelberg GmbH .
Violations are liable for prosecution under the German Copyright Law.
© Springer-Verlag Berlin Heidelberg 2003
Originally published by Springer-Verlag Berlin Heidelberg New York in 2003
Softcover reprint of the hardcover lst edition 2003
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant pro-
tective laws and regulations and therefore free for general use.
Product liability: The publisher cannot guarantee the accuracy of any information about dosage and
application thereof contained in this book. In every individual case the user must check such
information by consulting the relevant literature.
Typesetting: SNP Best-set Typesetter Ltd., Hong Kong

Cover Design: Design & Production, Heidelberg
31/3150 - 5 4 3 2 1 0- Printed on acid-free paper
Hommage a la memoire
de mes parents qui m'ont appris
apenser et arefLechir avec
enthousiasme.
Foreword
This book on the Nondestructive Characterization and Imaging of Wood by

Professor Voichita Bucur is truly the most outstanding reference on the subject
ever written. Since the origins of mankind, wood has played a key role in the
history of humans and other living creatures, ranging from provision of life
from trees giving air, heat, light, and food to nourish their bodies to structures
to protect them from the elements. Wood has also played a key role in one of
the world's primary religions.
Nondestructive diagnostics methods have long found application in medi-
cal practice for examination of the human body in order to detect life-
threatening abnormalities and permit diagnosis to extend life. Nondestructive
testing has been used for many years to insure the safety of machinery, air-
craft, railroads, tunnels, buildings and many other structures.
Therefore, it is timely for a treatise, like the present one, to be written
describing how wood can be characterized without employing destructive
test methods. Since wood is so valuable to mankind, it is important to know
the latest methods to nondestructively characterize wood for all practical
applications.
Starting with a historical review of nondestructive evaluation of wood,
the author proceeds with a systematic description of the nondestructive
techniques used to image the structural elements of wood. The author has
organized the book according to the wavelength of the interrogating nonde-
structive characterization technique used to create images of the internal
structure. The book proceeds from high resolution computed tomography
using ionizing X-ray and gamma radiation, through thermography, thermal
wave imaging, microwaves, ultrasonic waves and nuclear magnetic resonance.
Although theoretical approaches are presented, emphasis is placed on practi-
cal experimental applications which are extremely useful in many fields of
human endeavor.
Professor Bucur is an internationally known expert on the mechanical prop-
erties of wood and nondestructive techniques for the characterization of
wood. Not only are the techniques she describes scientifically important, but
often the images she provides are beautiful works of art.
Baltimore, December 2002

Professor Robert E. Green Jr.
Johns Hopkins University, USA, Center for Nondestructive Evaluation
Preface
In writing this book, I have attempted to provide a comprehensive account of

the principles, practice and achievements of imaging of the internal structure
of wood by various techniques developed and used during the last decades of
the twentieth century. This book is written from the point of view of one who
sees the development of these techniques principally as procedures for non-
destructive evaluation. I hope that it will be of particular interest to those who
are concerned with the study of wood as an engineering material, either for
fundamental research or practical application. The technical level and scope
are such that it should be of interest to scientists and students. Wherever pos-
sible, the use of complicated mathematics was avoided in favor of physical rea-
soning. The book can serve as a guide to the literature on the subject and cites
more than 500 references.
At the present time, two fields claim predominant attention: the imaging of the
internal structure of wood with methods based on the propagation of electro-
magnetic waves, elastic waves, and heat; and imaging with nuclear particle radi-
ation-based methods. For this reason, the book is divided into two main parts.
The first part is presented in five chapters and deals with the imaging of wood
structure via wave propagation. These chapters are presented in order of increas-
ing wavelength, and hence resolution of the wave phenomena used for imaging.
X-ray computed tomography (10- 12 to 1O-Io m wavelength, 1 nm resolution) is
followed by thermal wave imaging (10-6 to 1O-4 m wavelength, 1 mm resolution),
microwave radiation (10--4 to 10-1 m wavelength), ultrasonic, and nuclear mag-
netic resonance methods (10-3 to 10-1 m wavelength, 1 mm resolution).
The second part of the book is devoted to neutron imaging of wood struc-
ture. The task of writing this has been difficult in attempting to convey some
sense of the relative importance of the various contributions to this rapidly
developing subject.
First of all, I am very much indebted to Professor Timell for the opportu-
nity to write this book for the Springer Series in Wood Science.
I am very much indebted to the following colleagues and friends for reading
the manuscript and making comments for the improvement of the compre-
hension of the text: Dr Gunter S. Bauer, Germany, Professor Frank Beall,
Berkeley, USA, Dr. Ion Paul Beldie, Berlin, Germany, Dr. Harald Berndt, Oakland,
USA, Professor Gerd Busse, Stuttgart, Germany, Dr. Giorgio Catena, Rome, Italy,
Dr. Laurent Chrusciel, Nancy, France, Professor Colin Gough, Birmingham,
UK, Professor Adolf Habermehl, Marburg, Germany, Professor Phong M. Luong,
Paris-Palaiseau, France, Dr. Frederic Mothe, Nancy, France, Professor Peter
X Preface
Niemz, Zurich, Switzerland, Dr. Patrick Rasolofosaon, Paris-Malmaison, France,

Dr. Robert Ross, Madison, USA, Professor Christen Skaar, Blacksburg, USA.
The bibliography was prepared with the kind cooperation of various librar-
ians in France: Marie Annick Bruthiaux, University Henri Poincare, Nancy,
Marie Jeanne Lionnet, curator of the library of "Ecole Nationale des Eaux et
Foret de Nancy", David Gasparotto, librarian at the same institution, and
Angelica Onteniente from the library of the "Institut National de la Recherche
Agronomique" in Versailles. The difficult task of carefully checking all biblio-
graphic references was accomplished by David Gasparotto, whom I wish to
acknowledge. Thanks are also due to my colleagues Simone Garros, Andre
Perrin, Pierre Gelhaye, Christian Herbe, Claude Houssement and Etienne Farre
for their continuing help and technical assistance for more than 20 years, as
well as to Corrine Courtheoux and Yvonne Sapirstein for every day help and
assistance during the writing of this book.
I wish to acknowledge my Ph.D. students present in the laboratory during
the writing of this book: Nadia Mouchot, Adrian Hapca, Saeed Kazemi, Roberto
Martinis for their generous cooperation and assistance with handling modern
computer techniques for writing the text and for the illustrations. Their com-
ments and criticism were very stimulating. The final electronic version of the
manuscript was accomplished with the very enthusiastic help of Adrian Hapca.
I am also very grateful to my brother-in-law, Constantin Spandonide, for his
significant contribution in scanning and printing the numerous figures of this
book. Thanks are due to my sister, Despina Spandonide, and to my nephew,
Bruno Spandonide, for continuous and particularly enthusiastic encourage-
ment during the writing of this book.
I am indebted to many individuals and organizations for permission to
reproduce figures and tables. In each case, the sources are indicated in the text
and in the figure legend. I am especially conscious of my indebtedness to my
colleagues and friends from all over the world who followed the writing of this
book with interest.
I wish to thank Professor R. Green, Jr., Baltimore, USA, for his enthusiastic
support of my project - in writing this book and for agreeing to write the Fore-
word of this book - from the beginning of this project, in autumn 1999, when
we were at a conference in Brazil.
Last, but not least, I wish to thank Springer-Verlag, the publisher of this
book, for producing the printed volume, and the following French institutions
for sponsoring the color figures: CUGN (Communaute Urbaine du Grand
Nancy); INRA (Institut National de la Recherche Agronomique); Forests and
Natural Environment Department, Communication Department and Michelle
Cussenot; Forest Research Center in Nancy, Director Dr. Gilbert Aussenac; and
the Wood Research Laboratory (UMR 1093) associated with University Henri
Poincare, Nancy I, Director Professor Xavier Deglise.
Nancy, France, December 2002
VOICHITA BUCUR
Contents
1 Introduction ....................................... . 1
1.1 Brief Historical Review of Nondestructive Evaluation
of Wood and Aim of the Book ......................... . 1
1.2 General Concepts of Nondestructive Testing of Wood ...... . 2
1.3 Classification of Nondestructive Techniques for Wood
Quality Assessment .................................. . 4
1.4 Imaging of the Internal Structure of Wood ............... . 9
1.5 Summary and Outline of the Book ..................... . 11
2 Ionizing Radiation Computed Tomography .............. . 13

2.1 Introduction ....................................... . 13
2.2 Basic Phenomena ................................... . 15
2.2.1 X-Ray Propagation in Solids ........................... . 17
2.2.2 Attenuation and Profile of Inspected Solids ............... . 20
2.2.3 Reconstitutive Algorithms ............................ . 24
2.2.4 Treatment of Images ................................. . 25
2.3 Equipment for Imaging Techniques ..................... . 28
2.3.1 Description of the Equipment ......................... . 29
2.3.1.1 Fixed Equipment .................................... . 29
2.3.1.2 Portable Equipment ................................. . 32
2.3.2 Factors Affecting the Quality of the Image ............... . 32
2.3.2.1 Beam Path ......................................... . 33
2.3.2.2 Spatial and Contrast Resolution . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.3.2.3 Anisotropic Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3.2.4 Beam Hardening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.4.1 Examination and Inspection of Trees and Poles ............ 41
2.4.1.1 Growth Rate Assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.4.1.2 Moisture Content of Trees ............................. 49
2.4.1.3 Pollution Effects on Trees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.4.2 Wood Quality Assessment ............................. 52
2.4.2.1 Inspection of Logs for Lumber and Veneer ................ 54
2.4.2.2 Inspection of Poles ................................... 60
2.4.2.3 Inspection of Lumber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.3 Wood Technology and Wood Processing . . . . . . . . . . . . . . . . . . 63
2.4.3.1 Control of Lumber Drying . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
XII Contents
2.4.3.2 Control of Wood -Based Composites .................... . 67

2.4.4 Dendroarcheology, Dendrochronology,
Dendroclimatology .................................. . 69
2.5 Summary .......................................... . 70
2.6 Annexes ........................................... . 72
3 Thermal Imaging ................................... . 75

3.1 Introduction ....................................... . 75
3.2 Basic Aspects ....................................... . 78
3.3 Equipment for Thermal Imaging ....................... . 83
3.4 Applications ....................................... . 89
3.4.1 Imaging of Defects in Trees ........................... . 90
3.4.2 Timber Imaging .................................... . 91
3.4.2.1 Imaging of Knots ................................... . 92
3.4.2.2 Imaging of the Slope of Grain ......................... . 96
3.4.2.3 Imaging of Wood Density ............................. . 98
3.4.2.4 Imaging of Moisture Content Distribution ............... . 101
3.4.2.5 Imaging of Wood Rupture Phenomena .................. . 106
3.4.3 Imaging of Delaminations in Wood-Based Composites ..... . 110
3.4.3.1 Detection of Adhesion Defects in Laminated Wood
Composites ......................................... 113
3.4.3.2 Detection of Subsurface Defects Under a Veneer
Lamina ............................................ 117
3.4.3.3 Detection of Defects in Particleboards. . . . . . . . . . . . . . . . . . . . 117
3.4.4 Imaging of Defects in Different Types of Lumber Joints ..... . 120
3.5 Summary .......................................... . 121
4 Microwave Imaging ................................. . 125

4.1 Introduction ....................................... . 125
4.2 Basic Aspects ....................................... . 126
4.2.1 Effect of Experimental Factors ......................... . 132
4.2.2 Effect of the Physical Properties of Wood ................ . 137
4.2.2.1 Effect of Moisture Content ............................ . 141
4.2.2.2 Effect of Anisotropy ................................. . 142
4.2.2.3 Effect of Density .................................... . 145
4.2.2.4 Effect of Chemical Constituents ........................ . 146
4.3 Equipment for Dielectric Measurements and for Microwave
Imaging Technique .................................. . 148
4.3.1 Equipment for Laboratory Measurements of Dielectric
Constants .......................................... . 149
4.3.2 Equipment for Online Imaging of Wood Structure ......... . 153
4.4 Applications ....................................... . 158
4.4.1 Microwaves for Inspection of Forests .................... . 159
Contents XIII
4.4.2 Microwaves for Internal Inspection of Logs ............... . 164

4.4.3 Microwaves for Mechanical Grading of Lumber ........... . 169
4.4.4 Microwaves for Inspection of Wood-Based Composites ..... . 170
4.5 Summary .......................................... . 177
5 Ultrasonic Imaging .................................. . 181

5.1 Introduction ....................................... . 181
5.2 Basic Aspects ....................................... . 185
5.3 Equipment for Ultrasonic Imaging ..................... . 189
5.3.1 Equipment for Contact Scanning ....................... . 191
5.3.2 Equipment for Noncontact Scanning .................... . 194
5.4 Applications ....................................... . 200
5.4.1 Imaging of the Internal Structure of Standing Trees ........ . 202
5.4.2 Imaging of Lumber Structure .......................... . 208
5.4.3 Imaging of Defects in Wood-Based Composites ........... . 208
5.4.4 Imaging of Defects in Wooden Poles .................... . 212
5.5 Summary .......................................... . 213
6 Nuclear Magnetic Resonance .......................... . 215

6.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
6.2 Basic Aspects of NMR Imaging ......................... 216
6.3 The NMR Imaging Technique .......................... 237
6.3.1 Techniques for Imaging ............................... 237
6.3.2 Algorithms.... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
6.3.3 Deduction of Relaxation Times from Measurements. . . . . . . . . 242
6.4 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
6.4.1 Inspection of Living Trees, Logs and Lumber .............. 245
6.4.1.1 Water Content in Living Trees .......................... 246
6.4.2 Growth Rate and other Structural Features
in Logs and Lumber .................................. 250
6.4.2.1 Structural Features at the Macroscopic Level . . . . . . . . . . . . . . . 250
6.4.2.2 Structural Features at the Microscopic Level . . . . . . . . . . . . . . . 252
6.4.2.3 Spatial Distribution of Chemical Components
in Cell Walls ........................................ 256
6.4.3 NMR Imaging in Wood Processing ...................... 259
6.4.3.1 Moisture Content in Lumber During Drying. . . . . . . . . . . . . . . 259
6.4.3.2 Quality Control of Wood-Based Composites. . . . . . . . . . . . . . . 268
6.4.3.3 Control of Adhesion .................................. 270
6.4.3.4 Control of the Impregnation Process . . . . . . . . . . . . . . . . . . . . . 272
6.4.3.5 Examination of Archeological Wood ..................... 273
6.4.4 Further Applications for Measurements
of Elastic Constants .................................. 276
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
XIV Contents
7 Neutron Imaging 281

7.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
7.2 Basic Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
7.3 Equipment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
7.4 Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
7.4.1 Imaging of Water Distribution in Trees. . . . . . . . . . . . . . . . . . . 285
7.4.1.1 Water Distribution in Trees of Different Species ............ 285
7.4.1.2 Water Distribution in Trees of the Same Species. . . . . . . . . . . . 287
7.4.2 Imaging of Moisture Content in Lumber During
Drying............................................. 288
7.4.2.1 Short Drying Time ................................... 289
7.4.2.2 Long Drying Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
7.4.2.3 Imaging of Water Absorption. . . . . . . . . . . . . . . . . . . . . . . . . . . 292
7.4.3 Imaging of Moisture Distribution in Structural Elements .... 293
7.5 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
References ................................. . . . . . . . . . . . . . . . 299
Subject Index ............................................. 323
List of Notations ..................... . . . . . . . . . . . . . . . . . . . . . . 335
Color Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337

1 Introduction
1.1 Brief Historical Review of Nondestructive Evaluation

of Wood and Aim of the Book
Nondestructive evaluation of the physical properties of wood has its origin in

the need to solve practical problems without destruction of the integrity of
the object under inspection. It is generally accepted that the terms "nonde-
structive evaluation of wood properties" and "nondestructive testing of wood
properties" can be used interchangeably. Beall (1996), Ross and Pellerin (1991)
and Bodig (1994) selected the term "nondestructive evaluation" as more attrac-
tive since "it represents the process by which selected physical properties of
a material is being assessed without damage or alteration to its end-use
capabilities:'
The earliest nondestructive evaluation of wood was visual inspection,
largely used for the selection of timber used as load-bearing members for
specific applications. Even today, this method is extensively used for the
grading of wood products for lumber, poles, plywood, etc. and for the detec-
tion of biological degradation of these products.
The development of scientific nondestructive methods became possible in
the early 20th century with the development of the theory of elasticity and of
the instrumentation for the measurement of wood properties. The interest of
scientists was initially focused on the determination of the modulus of elas-
ticity by static methods (Horig 1935; Kollmann 1951). Later, the use of acoustic
vibrations for dynamic methods to determine elastic constants of wood were
published in Europe (Barducci and Pasqualini 1948; Hearmon 1948, Kollmann
and Krech 1960) in North America (Jayne 1955; James 1959), and in Japan
(Fukada et al. 1956).
The enthusiasm for the development of X-ray techniques for evaluation of
the internal structure of wood in the 1960s had a beneficial influence on the
development of micro densitometry in different laboratories all over the world
(Polge 1978). The X-ray diffraction technique was developed in Japan for
the study of the crystallinity of cellulose in wood (Fukada 1965; Fukada et al.
1956).
Pioneering results of dynamic tests using vibrational methods were
reported in the early 1960s by Hoyle (1961), Senft et al. (1962), and Pellerin
(1965) for the nondestructive testing of structural lumber. The development of
ultrasonic techniques for elastic characterization of wood was promoted in the
2 Introduction
United Kingdom by Hearmon (1965). Since the late 1970s, the activity of lab-
oratories all over the world on nondestructive testing of wood has been stim-
ulated by symposia organized in the USA at Washington State University by
Pellerin and coworkers in collaboration with the Forest Products Laboratory,
Madison (Ross and Pellerin 1991,1994) and since 1996 alternatively in Europe
and in the USA.
Since 1970, reference books have been published in the field of mechanical
characterization of wood (Jayne 1972; Bodig and Jayne 1982), dielectric prop-
erties of wood (Torgovnikov 1993), and acoustical properties of wood (Bucur
1995). Schniewind (1981) was the editor of the first encyclopedia of wood
science. The publication of an especially important series of books in the
Springer Series in Wood Science was initiated in 1983. The last 20 years of the
20th century have been characterized by an extraordinary development of dif-
ferent methodologies for nondestructive evaluation of wood products, which
has succeeded in the imaging of the internal structure of wood at both the
macroscopic and microscopic levels.
Due to the extensive literature on this subject, this book will provide an
overview of wood structure imaging techniques and the corresponding basic
concepts related to the nondestructive characterization of this material that
permitted the development of the modern imaging procedures. These out-
standing modern wood structure imaging techniques are a logical extension
of existing nondestructive methods developed previously and has benefited
from their breadth.
1.2 General Concepts of Nondestructive Testing of Wood
To promote the efficient use of wood materials in the future, three major areas
need to be addressed:
- development of nondestructive techniques for the evaluation of different
properties such as: physical, mechanical, chemical, aesthetic, etc.,
- improvement of natural qualities of wood through the modification of prop-
erties with different treatments, and
- creating new products using wood as a major raw material, corresponding
to the requirements of a modern society.
The development of nondestructive techniques has as it principal purpose
to reduce the uncertainty of wood products characteristics as influenced by
wood's biological nature. Despite the great attention given to quality control
in the development of manufacturing processes for glue laminated timber,
laminated veneer lumber or plywood, for particleboard and other wood-based
composites, interfaces are still the weakest link in the performance of these
products. To ensure interfacial integrity it is important to develop methods for
General Concepts of Nondestructive Testing of Wood 3
nondestructive inspection of wood-based composites and their components

(solid wood and adhesives).
The interfacial discontinuities, delaminations, cracks, porosity or density
variations may be detected by nondestructive techniques such as acoustic,
microwave, thermal, radiographic, or classical static methods. These methods
can help in the understanding of material behavior under different environ-
mental conditions, but difficulties still remain for detection and for quantita-
tive description of structural discontinuities and defects. It is important to
relate the nondestructive measurements to the mechanical properties of wood
and wood-based composites. The basis of such relations is the dependence of
the interfacial strength, on one or more mechanical characteristics related to
the nature and morphology of defects. There is also a need for the develop-
ment of nondestructive techniques in industry. The development of these tech-
niques will lead to intelligent manufacturing processes for wood products,
resulting in processes that will identify defects without characterizing "good
material" as defective.
The second approach is the quality assessment of improved wood products
from different chemical or mechanical treatments, which requires the devel-
opment of nondestructive techniques appropriate to each material. Analytical
assessment techniques will be used to gain an understanding of failure mech-
anisms in wood composites. The combination of physical testing and model-
ing should then yield an improved understanding of the factors that determine
the quality of products.
The third approach is to create new products using wood as a major raw
material, corresponding to the requirements of modern society. The market
for wood-based composites is a growing one worldwide, with new production
plants coming on line. Probably "biomimetics" will be a source of inspiration
for designing new composites. Progress can be expected from a "better micro-
architecture and from the mastering of smaller and smaller scales for the basic
constituents, and their spatial arrangement" (Roux 1998). In many applica-
tions, the nonlinear behavior of solid wood and of its composites must be con-
sidered. The sensitivity of new nondestructive evaluation techniques to defects
provides an opportunity to control the statistical properties of structures.
Given the hierarchical structure of wood it is obvious that one should seek
multiscale characterization tools. The use of multiscale approaches and the
prediction of the behavior of very complex systems through probabilistic
approaches and small-scale measurements must be emphasized. The problem
of selecting the most relevant scale for the study of the properties of the equiv-
alent medium has to be addressed for all applications.
The importance of the microstructure must be underlined because of its
basic role in nondestructive evaluation of the properties of materials. New
materials are currently "designed with a view towards developing micro-
structural conditions that more effectively allow the material to meet applica-
tion specific structural requirements" (Frantziskonis and Blodgett 1998). The
4 Introduction
mechanics of heterogeneous media require the definition of the representative

elementary volume (Bourbie et al.1987; Chelidze et al.I998). This volume must
be larger than the size of the elementary heterogeneity and, in the case of wood,
larger than the width of the annual ring, fiber length, etc (Fig. 1.1). Under
this assumption at different scales, the sample can be considered quasi-
homogeneous. This approach can be applied to any physical property and the
principle of physical analogies can be used. However, experimental studies
as well as theories have confirmed the dependence of the properties of
heterogeneous media on the scale of observation and size of the system. It is
reasonable to expect that in the near future the new approach to the mechan-
ics of heterogeneous materials will be that of fractal mechanics (Sahimi and
Arbabi 1993). It is also generally accepted that the control of mechanical prop-
erties of new wood-based composites such as glued laminated timber and ori-
ented strandboard, can be performed using probabilistic design (Castera
1998). At times, the deterministic approach may be better than the probabilis-
tic one, because it is difficult to find a general correlation between micro- and
macro-scale behavior of wood-based composites. A possible explanation of
this can be found in the multiplicity of the factors influencing the phenomena
studied.
Today, the technologies for wood-based composites are based on quality
control with nondestructive evaluation techniques, which play an increasing,
important role in adapting the market to the change of timber resources. In
the future, it will be important to develop new nondestructive techniques and
devices for quality control of new wood composites produced with a more
diverse raw material supply. As noted by Youngquist and Hamilton (1999), this
is the challenge for the 21st century.
1.3 Classification of Nondestructive Techniques for Wood

Quality Assessment
The characterization of wood properties is critical for the understanding of

material behavior and performance under operating conditions. Tailoring the
properties of new wood-based composites is essential for advanced product
design. The composites of the future will have to be made from such new
resources as underutilized species, recycled wood, and municipal waste, which
will be a mixture of wood, paper, plastic, and agricultural residues (Maloney
1992; Peterson 1993; Greaves 1998; Bowyer 2000).
The need to characterize wood-based composites for a myriad of applica-
tions has spurred the development of many new methods and instruments. An
ideal characterization tool would provide data about the properties that are
related to micro and macro structure without destructive sectioning. Such data
can only be obtained using nondestructive methodologies. Nondestructive
evaluation of wood and wood-based composites enables the determination of
01
J J \
10
(")
0;-
t-1U!l1--t
'"'"S;
5 . T ()
A I O.3A ~
&W , .. 1 o·
::s
,,' o
...,
o ~ ~
::s
C>.
group of trees tree (m) annual ring (COl) cells (mm) Icell wa ll (J.lm)1 fibril (nm)l,cellulosic crystal ...'"~
'~"
~.
megascopic scale mesoscopic scalelmacroscopic scale microscopic scale submicroscopic scale
~
g..
::s
.Eo
Fig.I.I. Hierarchical structure of wood. (Data from Bucur 1995, with permission)
''"'""
...C'
~o
C>.
/::)
'~"
~
i!;"
'"
'"
8'"
a'"
V1
6 Introduction
material parameters at nearly any point, line, surface or volume element of

interest, and at nearly any state during the life of the material. The large
number of potential methods for nondestructive evaluation of wood requires
a synergism of many scientific and engineering disciplines. Beall (1996) sum-
marized the nondestructive evaluation opportunities, and the needs for wood
products, as can be seen from Table 1.1. As noted, the development and the
implementation of nondestructive techniques depend on comprehensive
monitoring, and on integrated and joint-venture solutions. "Comprehensive
monitoring implies multiple sensing methods to determine a fuller spectrum
of properties that could be part of an in-process technology or one in
service."Having in mind the multiple sensing methods for the determination
of one parameter", Sobue (1993) proposed the classification of nondestructive
methods (Table 1.2) as a function of the physical property of wood and the
necessity for application (moisture content, density, stress rating, detection of
knots, grain orientation, etc.).
In the case of industrial wood-based composite products, the design and
production have always been considered the preeminent engineering chal-
lenges, with inspection and defect detection relegated to a subsidiary role.
However, today there is an increasing awareness of the potential for in situ
monitoring of processes by nondestructive methods with the objective of sig-
nificant improvement of predictability of the quality of the products while
optimizing utilization of raw material.
Table 1.1. Nondestructive evaluation opportunities and the needs in the wood products. (Beall
1996, with permission)
Material properties Preprocess Process Product Field
Defects
Surface defects X X X X
Internal defects X X
Basic properties
Grain orientation X X
Density X X X
Moisture content X X X X
Strength/stiffness X X
Permeability/diffusion X X
Surface roughness X X
Geometry
Profile, shape X X X
Thickness X X
Bonding quality
Composites X X
Laminate materials X X X
Fasteners X X
X indicates the need for nondestructive characterization

Classification of Nondestructive Techniques for Wood Quality Assessment 7
Table 1.2. Classification of nondestructive methods for solid wood as a function of its physical
properties. (Sobue 1993; with perrmission)
Wood parameter Properties measured Techniques
Moisture content Dielectric properties Impedance DC and high frequency,

microwave, infrared, NMR
Thermal properties Infrared
Ionizing rays absorption X-rays; gamma rays
Elastic properties Frequency resonance method,
ultrasonic method
Density Weight Gravimetric method
Ionizing rays X-ray micro densitometry; attenuation
of X, beta and gamma rays
Dielectric properties Microwave technique
Mechanical properties Pilodyne
Fiber direction Elastic properties Ultrasonic velocity
Dielectric properties Microwaves
Light scattering Laser, visible light
Thermal properties Infrared
Knots detection Optical properties CCD camera, colorimetry,
Elastic properties Ultrasonic CT
Ionizing radiation X-ray CT, gamma ray CT
Thermal properties Infrared thermography, laser
thermography
Electrical properties Microwave, NMR
Decay detection Elastic properties Vibration, ultrasonic
Mechanical properties Penetration of pilodyn
Dielectric properties Microwave
Ionizing radiation X-ray CT
Mechanical capacity Static properties Bending stress grading, proof-loader
Dynamic properties Vibration and ultrasonic methods
Acoustic properties Acoustic emission
This book is an attempt to comprehensively review numerous aspects of

imaging and nondestructive characterization of wood products, ranging from
the tree scale to the most complex wood composite, and hence to provide a
convenient point of reference. For this purpose, the various nondestructive
methods can be classified according to the characteristic scale (i.e., frequency,
energy or wavelength) of the radiation that interacts with wood specimens
(Fig. 1.2).
The synthesis of a wide range of phenomena involving electric and mag-
netic fields is provided by the electromagnetic theory developed by Maxwell.
The position of the spectral lines can be characterized by their frequency, 1],
wavelength, A, and wavenumber, k, which are defined by Eq. (1.1):
C = 1]A = 1] 2II!k (1.1)
8 Introduction
~l Fig. 1.2. Characteristics of the electromagnetic

5 waves. (Data from Swanson and Hailey 1987; cour-
1
(m)
tesy of Forest Products Society)
X-rays, gamma rays
tltraViolet rays
JViSible light
]
infrared
microwaves
ultrasonic
waves
nuclear
magnetic +-_--'
resonance
where c is the speed of propagation of the radiation.

The position of spectral lines of radiation can be characterized by their
wavelength, as is shown in Fig. 1.2. The various radiation and therefore types
of nondestructive methods can be classified according to the frequency of the
radiation involved. This is the criterion chosen for the description of different
wood imaging techniques in this book.
As can be seen, the highest transmitted energy, corresponding to the small-
est wavelength (l0-12 m) is obtained with X-ray methods. In the opposite posi-
tion we have acoustic and radio waves, for which the wavelengths are in the
range of 1 m. The interaction of X-rays with wood is accompanied by changes
in the energy of the electrons. The visible and ultraviolet regions are related to
transitions of the valence electrons, and the microwave and infrared regions
are related to molecular rotational and vibrational energy changes. The fre-
quency characteristics of the NMR region are low and in the range of 1 to
500 MHz and associated with transitions between energy levels corresponding
to the magnetic states of atomic nuclei.
Wood is a natural composite that has an hierarchic structure, which is het-
erogeneous and anisotropic. Anisotropy and heterogeneity are not absolute
characteristics, but are relative to a given physical property and to the scale
length of the corresponding physical phenomenon, characterized by a wave-
length (Bucur and Rasolofosaon 1998). Having in mind the hierarchical archi-
Imaging of the Internal Structure of Wood 9
Table 1.3. Scale of wood structural characteristic and the required resolution of imaging tech-
niques expressed by the wavelength
Imaging method with Wavelength (m) Scale of wood structure
X-ray 10- 12 ••• 10-9 Submicroscopic, microscopic and macroscopic

structure
Visible light 10-6 Micro and macroscopic structure
Infrared 10-5 ••• 10-6 Macroscopic structure
Microwave 10-3 ••• 10- 1 Macroscopic structure
Ultrasound 10-2 ••• 103 Mesoscopic (tree), macroscopic and microscopic
structure
Radio wave 10 3 ••• 10 5 Gigascopic (forests) and megascopic (group of
trees) scale
tecture of wood, the selection of the most relevant technique for structure
imaging is directly related to the resolution required by the selected method
and consequently to the wavelength, as can be seen from Table 1.3.
The general understanding of the interaction between the electromagnetic
waves and the wood material under inspection must be based on an accurate
description of the phenomena, with an increasing degree of sophistication of
the theoretical models. Consequently, the nondestructive evaluation of the
properties of the media becomes an inverse problem.
The development of a corresponding theory that describes the interaction
between the electromagnetic waves and the material requires three principal
steps:
- observation of the phenomena to provide guidance for the theoretical
approach;
- integration of the theory to the specific problem of observed phenomena;
- conduct of experiments that verify the first and the second steps.
The interaction between the second and the third step generates a model
that can be validated by the experimental data of the first step. The second half
of the 20th century has shown the development of a new branch of wood
physics related to the nondestructive evaluation of its properties.
1.4 Imaging of the Internal Structure of Wood
The nondestructive imaging of wood structure is based on different methods

utilizing typical wavelengths of elastic waves, penetrating radiation, light, elec-
tric and magnetic fields, etc. Computed tomography was developed on the
mathematical basis established by Radon (1917). He demonstrated that from a
complete set of projections of relevant physical variables it is possible to recon-
struct the image of an object. The image can be reconstructed by mapping of
10 Introduction
different measured parameters or can be obtained by a tomographic tech-

nique, using algorithms and advanced computational procedures for data col-
lection, image reconstruction, and display. Analytical methods, based on
Fourier transform or algebraic iterative methods, using substantial projection
data, are used for tomographic reconstruction. In the future, tomographic tech-
niques are expected to become of general use.
The nondestructive techniques presented in this book for wood imaging are
ionizing radiation (X-rays and gamma rays) having a resolution of lO-9 m,
thermal waves with a resolution of 10-3 m, microwaves with resolutions from
lO-4 to lO-1 m, and ultrasonic waves with a resolution of lO-3 to lO-2 m, or more,
depending on frequency, for imaging wood structure at the macroscopic scale.
The nuclear magnetic resonance window is superimposed on the ultrasonic
window.
X-ray tomographic imaging can be produced with fixed or mobile equip-
ment with one source and one detector acting in parallel by translation or with
an array of transducers located around the sample.
High resolution ultrasonic tomographic techniques were developed as a
complement to X-ray tomographic imaging. The ultrasonic waves propagating
in wood are related to its elastic properties. The waves interact with the
microstructure, and their characteristics are modified as they travel through
the material due to reflection, scattering and attenuation. The detected signal
is displayed, processed, and interpreted in terms of the internal structure of
wood based on its relations to the input wave. Recent technological advances
allow new capabilities for measurement of elastic properties and, consequently,
the production of high-resolution ultrasonic images.
The microwave frequency region is bounded by the radio wave region
(upper bound) and by the infrared region (lowest bound). The nature of the
interaction of microwaves with a wood structure suggests a certain similarity
with ultrasonic waves. An advantage of the high velocity microwave propaga-
tion coupled with the non contacting feature of the transducers is that
it permits a rapid inspection limited only by mechanical considerations.
Therefore, amplitude and phase measurements of waves transmitted or
reflected by the specimens will contain information about the structure of
internal flaws, inhomogeneities, moisture content of logs, lumber and wood-
based composites.
Microwave imaging of vegetative material, including leaves, stems and
trunks at various moisture contents and temperature conditions is possible
over a wide range of frequencies. The probes can be either scatterometers,
radar, or reflectometers and can operate in the proximity of the medium as
open-ended coaxial lines, cavity resonators or antennas.
In wood science, thermal imaging is a relatively new field for ascertaining
the integrity of subsurface structure. The scanning infrared imaging technique
appears to be a very promising remote sensing method for thermographic
inspection of trees, solid wood, and composites in situ. The NMR imaging
method is totally nondestructive and noninvasive for wood. The availability
Summary and Outline of the Book 11
in the future of low-cost devices will contribute to the widespread use of this
method. Neutron imaging is one of the most recent nondestructive techniques
developed for wood. High-resolution imaging (151lm) of water distribution
in wood can be obtained with small, clear specimens and with structural
elements.
1.5 Summary and Outline of the Book
The purpose of this introductory chapter, which is concerned with the nonde-
structive characterization and imaging of wood, is to provide a brief overview
of the material that will be discussed in the subsequent chapters. With com-
puted tomography two- and three-dimensional images of the internal struc-
ture of objects can be obtained without physical sectioning. Tomograms
obtained with different techniques, ranging from ionizing radiation (X-rays
and gamma rays) to ultrasonics and nuclear magnetic resonance and nuclear
imaging, provide excellent means of obtaining information about the internal
structure of wood.
The criterion selected for the sequence of the chapters is the wavelength of
the physical radiation, which is closely related to the scale of observation used
for wood characterization and imaging. This means that the book will begin
with X-ray and gamma ray imaging techniques, corresponding to the smallest
wavelength and consequently to the highest resolution, followed by thermal
imaging, microwave imaging, and ultrasonic and nuclear magnetic resonance
imaging methods. The last chapter is devoted to neutron imaging.
The book offers an in-depth review of the state of the art of the use of non-
destructive evaluation techniques as applied to the imaging of wood. The the-
oretical approach will be outlined, and emphasis will be placed on practical
applications. For the readers who would like to have more detailed informa-
tion about any particular technique, an extensive bibliography with a subject
index concludes the book.
2 Ionizing Radiation Computed Tomography
2.1 Introduction
Computed tomography scanning with ionizing radiation provides three

dimensional information about the internal inhomogeneous structure of the
specimen under test in a nondestructive, noninvasive and rapid manner. This
technique is a "cornerstone of materials characterization" (Kinney et al. 1988).
The aim of computed tomography is to create a precise map of the internal
inhomogeneities of the sample. This map is obtained through the determina-
tion of the attenuation coefficient of an ionizing radiation (X-ray, y-ray) in a
single thin layer of the sample. The tomogram shows this layer in a perpen-
dicular position to the main axis of the object, while conventional X-ray tech-
nique, called Rontgen densitometry, produces an image parallel to the object
under inspection. A conventional radiographic image is produced by sample
translation and offers a rapid scan of the specimen under test, and, at the same
time, locates the defect zones in one plane.
Tomographic images produced with ionizing radiation (X-rays and y-rays)
are called tomograms of slices or CT images and are obtained from the trans-
lation and rotation of the source and detectors around the specimen. The slices
produce spatial information (in 3D) able to discern zones of low attenuation
contrast. First-generation tomograms have been obtained with one source and
detector acting in parallel, by translation. The sample was rotated by a 1°-step
angle and the whole image was obtained for 1800 of collected data. The second-
generation tomograms were also obtained by translation, using an array of
detectors that made simultaneous measurements through different angles
during a single traverse inspection. The sample was rotated by the array beam
angle. The third-generation tomograms are produced by a fanning movement,
with a scanner provided with many detectors located on an arc focused at the
X-ray source. The fourth generation of scanners also has a fan system of detec-
tion, and the detection array is located on a circle that surrounds the source
and the sample.
In inhomogeneous media like wood, the attenuation coefficient depends
on both the quantum energy of the ionizing radiation and the chemical
composition of the sample. The electronic signals corresponding to images
taken on wood specimens with X-rays in the Compton energy range enable
the precise recording of mass density variations, clearly distinguished from
the effects of high atomic number constituents. A characteristic signature
of the specimen inspected with X-rays can be obtained (Polge 1966; Mothe
et al. 1998).
Because the high quantum energy of radiation and the low atomic numbers
of chemical constituents of wood, the attenuation phenomenon is caused
mainly by the Compton effect and consequently is proportional to the mass
density of the wood. It is generally accepted that the density of the cell wall in
oven dry wood is constant (I500kg/m3). Therefore, the density variations
observed on tomographic images are due to the distribution of various
anatomic structural elements of the specimen under inspection and to the
water content in the cell walls and in lumina.
The tomograms are obtained by calculation, using a sophisticated computer
program (Kak and Slaney 1988; Habermehl and Ridder 1994; Grundberg et al.
1996; Schmolt et a1.1999) that involves a complex technology. Each zone of the
specimen having the same density is evaluated statistically and is represented
by one color, or by a gray scale. More commonly, the red and the violet indi-
cated low wood density and yellow high density.
This chapter discusses the ability of X-ray or gamma rays to provide a non-
destructive method to identify and evaluate the internal structure of wood and
wood-based composites. X-ray or gamma ray imaging is a rapid tool able to
provide visual and analyzable internal structure of solid wood and of wood-
based composites. The ability of this method to quantify spatially the density
variations in solid wood, the presence of internal defects and inhomogeneities
in lumber or in wood-based composites, the variation of moisture content in
wood specimens, the distribution of water inside a trunk, etc. explains its
important increasing implementation in wood science and technology.
During the last 20 years, Scandinavian countries and Germany in Europe,
Canada, Japan, Australia, New Zealand and the United States have put stress
on the understanding of the basic aspects of X-ray computed tomography in
view of the development of specific wood, industrial scanners for routine
quality control of wood products (Benson-Cooper et al. 1982; Habermehl
1982a,b; Mc Millin 1982; Cown and Clement 1983, Asplund and Johanson 1984;
Qnoe et al.1984; Funt and Bryant 1985; Lindgren 1985; Harley and Morris 1988;
Davis et al. 1989; Wagner et al. 1989a,b; Rinn 1991; Habermehl and Ridder
1992a,b, 1993,1995, 1998; Grundberg et al. 1995; Schmoldt 1996; Sugimori and
Lam 1999; Qja et al. 2000).
The advantages of ionizing radiation computed tomography are numerous
when compared with conventional radiography. Firstly, this technique elimi-
nates the intermediate steps involving photographic film and optical densito-
metry and is able to make data available in real time. In process control and
in manufacturing situations, the density feedback is very important for tech-
nological competitiveness. Secondly, an important advantage of direct scan-
ning technique is the improvement of the calibration procedure, because the
mass attenuation coefficient can be determined directly, using a scintillation
detector with standard radiation pulse shaping and counting equipment, which
allows the user to select the energy range counted (Fig. 2.1). Thirdly, a large
Basic Phenomena 15
Fig.2.l. X-ray device for computed tomography on a tree. (Habermehl, pefs. comm., with
permission)
volume of material can be inspected quickly and implementation of scanning

technology in sawmills and other factories using wood materials will have an
important payback for wood processors.
2.2 Basic Phenomena
The basic phenomena involved in X-ray reconstitutive tomography is illus-

trated in Fig. 2.2. The translation and rotation of a radiation beam, defined by
collimators, are scanned across the sample. After each linear scan, the source-
detector system rotates through an angle around an axis perpendicular to the
chosen plane. The sequence continues for a complete rotation of 180°. For each
beam path, the detector quantifies the attenuation coefficient of the radiation
beam through the sample, also called the "object".
As noted by Liu et al. (1988), the term "attenuation coefficient of X-rays"
used in this chapter refers to "total decreases in X-ray intensity, regardless of
whether absorption or scattering have been the interaction mechanisms in the
--"
rotation
"" "
~ translation
a)
slice
volume elements "voxels"

containing Nx square pixels of thickness w
b) obj ect
y SD = X-ray path
r =ray of the projection
8 =angle of proj ection
c)
Fig. 2.2. Basic aspects involved in X-ray reconstitutive tomography. a Rotation of source-detec-
tor system around the specimen (Lindgren 1991b, with permission). b Volumetric element (voxel)
in a slice (Lindgren 1991b, with permission). c Beam path geometry between the source and the
detector. (Gilboy and Forster 1982, courtesy of Academic Press)
Basic Phenomena 17
transmission equipment". The term "absorption coefficient", often used in the

literature must be avoided. From a physical point of view, X-ray or gamma ray
attenuation corresponds to the removal of quanta from the incident beam by
whatever process reduces its intensity. Ionizing radiation computed tomogra-
phy is a nondestructive radiation method that allows the conversion of atten-
uation coefficients into density data and then into images.
2.2.1 X-Ray Propagation in Solids
As described in reference books (Herman 1980; Kak and Slaney 1988) and arti-
cles (Polge 1966; Hagglund and Lindgreen 1985; Lindgren 1985, 1991a,b; Fio-
ravanti and Ricci 1991; Habermehl and Ridder 1992 a,b; Lindgren et al. 1992),
in the case of monoenergetic photons such as y-rays or highly filtered X-rays,
with rectilinear propagation in solids, the relationship between incident and
transmitted ray intensities, the attenuation coefficient and the thickness of the
sample, is:
1= 10 e-I.lI (2.1)
where
I = intensity of the transmitted ray beam, passing through the sample
10 = intensity of the incident ray beam, passing through air only
/l = attenuation coefficient
t = thickness of sample
As ionizing radiation traverses the matter it is absorbed exponentially in
accordance with the equation:
1/1 0 = e 41t (2.2)
The attenuation coefficient /l depends on the mass density of the sample

p (kg/m3) and on the mass attenuation coefficient /l' (m2/kg or cm2/g as noted
more often in the literature).
/l = /l'p (2.3)
The density is then
p = /ll/l' (2.3')
The density (Eq. 2.3') of the inspected sample is the ratio between the linear
attenuation coefficient /l and, /l', the mass attenuation coefficient, and can be
determined experimentally by measuring the difference between the intensity
of the incident beam and of the absorbed beam.
The Eq. (2.1) can be rewritten as:
1/10= e 41Pt (2.4)

and the corresponding logarithm is:

(In I -In Io)/t =Il'p = 11 (2.5)
From Eq. (2.5) the attenuation coefficient 11 may be calculated, as demon-
strated by Olson and Arganbright (1981).
The mass attenuation coefficient of wood 11', has been determined by several
methods, using X-rays or gamma rays, and reported in the literature by many
authors, for specimens of different thicknesses and with different energy levels
of the radiation source, as can be see from Table 2.1. The mass attenuation coef-
ficient is a basic parameter of wood. Moschler and Dougal (1988) noted the
value 11' = 0.185cm2/g, Malan and Marais (1992) using gamma rays, reported
values for 11' ranging between 0.1858 cm2/g for air dry wood and 0.1797 cm2/g
for oven dry wood, measured on 32 wood species having a gravimetric density
ranging between 156 (balsa, Ochroma pyramidalis) and 1194kg/m3 (tropical
boer-bean, Schotia sp.). The experimental value of 11' depends on the energy
level of the source, on the moisture content of the wood and on the wood
species. Theoretical values reported by Tiitta et al. (1996) is 11' = 0.192 cm2/g at
0% moisture content and 59.5keV and, is 11' = 0.198cm2/g at 100% moisture
content.
Table2.1. Wood mass attenuation coefficient /1' (cm'/g) as found in the literature
Coefficient /1' Energy source Moisture content Reference Differences

(cm'/g) (% ) (% )
0.191 Am24! 8 Moschler and Dougal (l988)

0.185 Am 241 0 Moschler and Dougal (1988)
0.189 Am24 ! 0, Particleboards Ranta and May (l978) 2
0.191 Am24! 0, Pinus elliotti Feraz (1976) 3
0.21-0.28 Am'4! 0, 17 Species Ferraz (l976) 12-34
0.192 0.060 Mev 0 Olson and Aganbright (1981) 3.6
18.22 Fe55 8 Moschler and Dougal (1988)
17.72 Fe 55 0 Moschler and Dougal 1988
17.70 0.006 Mev 0% Moschler and Dougal (1988) 0.1
0.185 Am'4I 8, 31 Species Malan and Marais (1992) 0
0.176 Am'4I 8, Oak Malan and Marais (l992)
0.0726 Ce!37 8, Oak Malan and Marais (1992)
0.190 Am'4I 8, Rose gum Malan and Marais (1992)
0.082 Ce!37 8, Rose gum Malan and Marais (1992)
0.198 Am'4! 8, Stinkwood Malan and Marais (l992)
0.081 Ce 137 8, Stinkwood Malan and Marais (l992)
0.192 59.5keV 0, Theoretic Tiitta et al. (l996)
23.42 5.9keV 100, Theoretic Tiitta et al. (1996)
The differences were calculated considering the data reported by Moschler and Marais as
reference
Basic Phenomena 19
The attenuation coefficient in inhomogeneous solids is not constant and

depends on the local coordinates x, y of the pixel. Consequently, the intensity
of the ray beam is:
1= 10 e - fSD~(X,Y)d' (2.6)
where SD are the ray paths and x, y the coordinates of the chosen slice (Fig.
2.2c). The integral is taken along the beam path SD, that is the distance between
source and detector.
The ratio
1/1 0 = e - fSDf(x,Y)d' (2.7)
and the corresponding logarithm, noted p, is called the ray sum of the
function f(x, y):
P -In
-
r
1/10-- - JSDf(x,y)d, (2.8)
and is proportional to the average linear attenuation coefficient along the total
beam path.
The tomograms are reconstructed with the measuring projection data
f(x, y). A set of ray sums p(r, e) at a given angle e defines a projection that is
used for the reconstructive radiographic image in polar coordinates. For a
point (x, y) the thickness, w, of the pixel is:
w = y cose+ x sine (2.9)
A projection at the angle e is called the set of ray sums with e constant, for
a range of r. The function f(x, y) in Cartesian coordinates or f(r, e) in polar
coordinates are the basic data for the reconstruction of the image, which the-
oretically requires an infinite number of ray sums. Practically only a finite
number of ray sums are recorded. The image reconstruction is based on a
square array containing N x N square pixels of thickness w. Each projection
consists of S measurements at spacing w.
The spatial resolution of the image is determined by the beam width, which
is related to the spacing between two neighboring rays in each projection. For
medical scanners used also for wood scanning, the calculated X-ray linear
attenuation coefficient in each voxel is normalized within the computer by the
corresponding linear attenuation coefficient for water according to Eq. (2.10).
This normalized value is referred to as the CT number, or Hounsfield number.
The unit is the Hounsfield, noted [HJ. The CT number is calculated as:
CTnumber = (/-Ls - /-Lw) x 1000 (2.10)

/-Lw
where /-Lw is the linear attenuation coefficient in water and /-Ls is the attenua-
tion coefficient for the x voxel in the cross-sectional slice of the tested mater-
ial at an average photon energy of 73keV, a typical value used for medical
scanners.
2.2.2 Attenuation and Profile of the Inspected Solids
The attenuation values calculated for each pixel are translated into a gray scale
or a color scale and transformed into a picture of the cross section of the
sample on a TV screen. For medical scanners, the CT number ranges between
-1000 for air to 0 for water and to +1000 for human bone. When working in
the optimum range, the medical scanners exhibit a typical photon limited noise
level of 0.5% in pixels of 1 mm2 with l-cm slice thickness. The image is often
reconstructed with 512 x 512 pixels and saved on floppy discs. In this way,
it is possible to use different digital image techniques such as geometrical
transformations followed by subtraction of the image to measure the dynamic
processes such as wood drying as a function of time or wood impregnation
with preservative substances as a function of time or of thickness of the
sample.
The relationship between wood density and the attenuation coefficient was
established by Lindgren (1985) for pine (Pinus sylvestris) on a Philips 210 CT
medical scanner and is expressed by Eq. (2.11) as:
Jl = 0.93 Po -1001 (2.11)
where Po is the oven dry density (kg/m3) and the attenuation Jl is expressed in
[H]. The corresponding correlation coefficient is r = 0.99 (Fig. 2.3). The rela-
tionship between the moisture content and the attenuation coefficient for
Po = 430kg/m3 and for Po = 540kg/m3 was also established as can be seen from
Fig. 2.4, using rectangular samples of 14 x 59 x 116mm and 30 x 97 x 250mm.
The accuracy was 1.5% for density measurements and 2% for moisture content.
It is well known that the attenuation measurements depend on different factors
such as: the energy level of the source, the moisture content, the geometry of
the specimen and the general noise observed on the tomograms.
H
r = 0.99
-520
-560
-600
-640
360 400 460 480 520
density kg/m3
Fig. 2.3. Relationship between oven-dry density and attenuation coefficients in Pinus sylvestris
sapwood (Lindgren 1985, with permission)
Basic Phenomena 21
Legend
j= 540 kglm3
H
CD H=3.ISa-506 r=0.99
@ H=4.71a-554 r=0.99
j = 430 kglm3
-100 <1l H=2.97a-609 r=0.99
® H=3.96a-646 r=0.99
§
~
::I -200
ii
~
-300
-400
-500
-600 .,:::;_...L.._...L..._-L-_--l_---I_---'L--_L-_-'--_...1-_...L..._-L-_--l_ _L..-

20 40 60 80 100 120
moisture content (%)
Fig. 2.4. Relationship between moisture content and attenuation for oven-dry wood at two con-
stant densities of 430kg/m3 in Pinus sylvestis. (Lindgren 1985, with permission)
Based on the data existing in the literature, Gierlik and Dzbenski (1996)
studied the influence of moisture content on the mass attenuation coefficient
for four levels of moisture content of wood, namely 0, 9.5, 18.5 and 26.8%. A
nonsignificant decreasing relationship was obtained.
The theoretical problems of the noise on the image were discussed by Gilboy
(1984) for the conventional translate rotate scanning geometry, using the fol-
lowing characteristics: D, diameter of the sample; N, pixel size; w, the width of
the pixel; S, ray sum; t, spacing between rays, usually the same as w; M, total
number of projections; n, finite photon count per ray sum; k, a constant
depending of the type of the filter used; T, total run time; /1, linear attenuation
coefficient; A', the effective activity.
The noise propagated to the image is estimated as a variation of the atten-
uation coefficient /1, and can be expressed by Eq. (2.12):
(2.12)
The relative accuracy of the measured attenuation coefficient is given in

Eq. (2.13) as:
.1/1//1 = 4I1 kN2.(D/t TA,)I/2 .[sinhC/lD)/ C/lD)3t2 (2.13)

The first part of this equation, 4I1 kW. {D/tTA')1I2, depends on the geomet-
rical factors, the total run time and the source intensity. The second term of
Eq. (2.13), [sinh{J.1D)/{J.1D)3]1I2, is called "sensitivity factor:' and is accuracy
dependent. The plot of the sensitivity factor versus the diameter of the sample
to be inspected goes through a minimum and can indicate the choice of
optimum photon energy required for the best image.
Lindgren (1991a) identified the noise on five specimens of Pinus sylvestris
using a medical scanner. The noise in each pixel was ±4 CT numbers and a
variation with ±1 CT number corresponded to a ±2 kg/m3 variation in dry
wood density and ±6kg/m3 variation in wet wood density. No difference in
noise depending on annual ring width was observed as long as the same direc-
tion of rotation was used to obtain the images. The accuracy of the noise needs
to be established each time measurements are performed on the same sample,
for example, in studies of moisture content variation over time.
The image quality is dependent on spatial and density resolution and an
improvement in one can only be achieved by a reduction in the other. For a
given time of measurement, the use of a high intensity source (such as a gamma
source) reduces the compromise between these two parameters. Gamma
sources possess this important practical advantage and also have a constant
photon energy over space and time. X-ray tubes produce an energy spectrum
that is variable with time, due to supply fluctuations or to tube degradation.
They also induce variations of the take-off angle for fan beam measurements.
Beam hardening is affected by all these effects. Indeed, the effective energy of
the radiation is a function of many parameters related to a specific scanner,
and the corresponding resulting image is scanner dependent. Gilboy (l984)
noted "the lack of precision in the X-beam energy becomes a greater problem
when working in the photoelectric region where the strong energy dependence
magnifies the effect of small energy changes:'
Lindgren (l991a) noted that the total attenuation coefficient of a material J.1
is the sum of the attenuation coefficient du to the photo-electric effect, J.1P' and
of the attenuation coefficient due to the Compton scattering effect of photons,
J.1" as can be seen from Eq. (2.14).
(2.14)
where J.1 P and J.1c depend on the density of the material being tested, electron
density, effective atomic number and photon energy. Consequently, it is possi-
ble to calculate the attenuation coefficient for any material if the chemical com-
position and density are known.
The attenuation coefficients of principal wood constituents are: 0.2634 for
cellulose, 0.2655 for hemicellulose and 0.2608 for lignin. Lindgren (given cita-
tion) calculated the attenuation coefficient of wood and the corresponding CT
numbers for different volumes of cellulose, hemicelluloses and lignin and for
the oven-dry density Po = 500kg/m\ as shown in Table 2.2.
The decrease in the attenuation coefficient is related to the increasing lignin
volume and the decreasing cellulose volume. A more detailed analysis consid-
Basic Phenomena 23
Table 2.2. The influence of chemical constituents of wood on the attenuation coefficient and CT
numbers, calculated for an oven-dry density of 0.5 g/cm3. (Lindgren 1991a, with permission)
Cellulose (%) Hemicellulose (%) Lignin (%) Attenuation coefficient CT number
50 25 25 0.08775 -527
25 25 50 0.08753 -528
50 40 10 0.08799 -526
10 10 80 0.087l5 -530
Table 2.3. Theoretical mass attenuation coefficients of wood calculated for two levels of photon
energy and for the following mass fraction of elements: 50% C, 44% 0,5.7% H, 0.3% N (Tiitta
et al. 1996) (lOth International Symposium on Nondestructive Testing of Wood, 2002, Press Poly-
techniques et Universitaires Romandes, p. 189. Reproduced with permission of the editor. All
rights reserved)
Material Moisture content Mass attenuation Coefficient (cm2/g)

(%) 5.9keV 59.5keV
Wood 0 22.61 0.192

5 22.69 0.193
10 22.76 0.193
20 22.88 0.194
30 22.98 0.195
50 23.15 0.196
100 23.42 0.198
Water 24.23 0.205
ering the mass fraction of C, 0, Hand N was proposed by Tiitta et al. (1996).
They calculated the mass attenuation coefficient of wood for two levels of
photon energy and for moisture content ranging from 0 to 100%., as can be
seen from Table 2.3. The mass fraction of the elements was: 50% C,44% 0,5.7%
H, 0.3% N. From this theoretical approach, it was confirmed that the photon
energy level is a very important factor for the definition of mass attenuation
in wood. For example, at 0% moisture content, the mass attenuation coefficient
is 22.61 cm2/g at 5.9keV and 0.192cm2/g at 59.5keV. It was also noted that at
the same level of photon energy the mass attenuation coefficient increases
slightly with moisture content. For example, at 59.5keV and 10% moisture
content, the mass attenuation coefficient is 0.193cm2/g and at 100% moisture
content, it is 0.198 cm2/g. A possible explanation of this statement is the small
difference between the mass attenuation coefficient in water (0.205 cm2/g) and
in oven-dry wood (0.192cm 2/g).
The linear regression equations established between the attenuation coeffi-
cients expressed in CT numbers and the measured density of oven dry wood
and of green wood are given in Table 2.4. The proposed equations are calibra-
Table 2.4. Regression relationships between wood density and CT numbers for different
moisture contents. (Lindgren 1991a).{with permission)
Variables Regressions Density CT numbers

Standard deviation Standard deviation
po and CT po = 1053 + 1.052 CT; R = 0.99 ±2.7kg/m3 ±4

pu and CT pu = 0.993 CT + 1015; R = 0.99 ±6.7kg/m3 ±13.4
tion equations. They must be established for each scanner used for laboratory
or industrial measurements.
2.2.3 Reconstitutive Algorithms
As noted in reference books and articles (Herman 1980; Kak and Slaney 1988;
Gilboy and Foster 1982; Gilboy 1984; Fioravanti and Ricci 1991; Schmoldt 1996;
Bahy11998; Coles 1999), image reconstruction methods are:
- analytical methods, based on Fourier transforms and related techniques
including filtered back-projection (convolution). Data obtained from con-
volution can be processed immediately as collected and the virtual image
built by projections is available instantly after scanning. Finite Fourier trans-
form (FFT) is periodical in the time domain, requires an infinite number of
projections and gives a picture in the frequency domain in the Cartesian
coordinate system. In practical situations, the number of projections is not
infinite as it should be. The Radon transformation in the image reconstruc-
tion process overcomes this aspect. Convolution of the density function
using a filter function realizes the filtering process of the transformed data,
for a selected angle and position. A huge amount of computation is required
with parallel projections or with fan beam projections. The effect of an
increasing number of projections on image quality has been reported and
seems to be limited to 45 projections, as noted by Gilboy (1984) for a spec-
imen in light alloy casting. When data rates are low, this method can give a
useful saving in measurement time.
- algebraic (iterative) methods need a very large amount of data for image
reconstruction, and better results are obtained from relatively few
projections.
Each attenuation value used for the image reconstruction is associated with
a specific cross-sectional pixel size and finite slice thickness (creating the
voxel). Pixel squared size is between 0.9 and 3 mm. Pixel resolution depends on
the source-detector distance and on the number of detectors. Slice thickness
can vary from 1.5 to 20mm (Schmoldt 1996) and limits the sensitivity and res-
Basic Phenomena 25
olution of scanning of heterogeneous samples. Indeed, a better coverage of the

specimen is achieved with higher scanning frequencies, but the process
requires a long exposure time to ionizing radiation and can reduce the life of
the X-ray tube. The run time can be reduced by the use of a wider scanning
beam, but this choice induces a loss in spatial resolution.
2.2.4 Treatment of Images
The images reconstructed by back projection of a filtered projection require

attenuation values at any point in the image. These data can be numerically
presented and statistically interpreted as density variations along any line seg-
ments across the image. Histograms of the attenuation values within any region
can be plotted, and line integrals can be obtained (Funt and Bryant 1985).
Groups of pixels can be then interpolated to give a degree of image smooth-
ing. Faust et al. (1996) show a histogram of the image (Fig. 2.5) and threshold
selection for several gray levels for knots and cracks, in a red oak log as: gl is
the lowest gray level corresponding to pixels from air, cracks, large voids, parts
of earlywood; g2 is the gray level corresponding to pixels from decay, early-
wood, and a few portions of latewood; g3 is the gray-level corresponding to
pixels from latewood and a few portions of knots; g4 is the gray level corre-
sponding to pixels from knots and some portions of latewood.
The scanning parameters are influenced by different factors, such as: wood
species, size of the specimen (diameter of log or thickness of sample), level of
contrast in density change necessary to discriminate various defects, defect
size, end use of scan information, with large-scale or fine-scale details, and the
speed of scanning required for the production context.
The huge amount of data must then be condensed, and only those internal
features important for the subsequent processing must be identified automat-
ically (Schmoldt 1996; Schmolt et al. 1997) with a machine vision using artifi-
cial neural networks. Javadpour et al. (1996) developed a specific artificial
neural network for the identification and classification of knottiness of Irish
Sitka spruce. The study was concerned with the optimal orientation of the log,
from its modeled images as a function of internal defects, to yield a high
quality lumber. Major difficulties are related to the high moisture content of
green logs that render obscure the recognition of the knot pattern from sur-
rounding clear wood. The knot is simulated as a branch of a right elliptic cone
and is defined by several geometric parameters related to its shape and posi-
tion in the stem. The model provided geometric parameters and density vari-
ations as a function of the age of the annual ring, within the trunk and internal
multiple branches and, effectively simulates the profiles obtained by tomo-
graphic techniques. Using this model, it was possible, for a given stem, to obtain
any number of internal transverse and longitudinal profiles of densities. The
annual ring widths and the percentage of earlywood and latewood in each ring
1000
<I>
0;
'"
'0.
'-
800
0
600
400
200
0
0 40 80 120 160
b) gray level
Fig.2.S. Histogram of the image of knots and cracks in a red oak log (Faust et al. 1996). a CT
image of red oak section; b histogram of the image and threshold selection. (10th International
Symposium on Nondestructive Testing of Wood 2002 Press, Poly techniques et Universitaires
Romandes, p. 203. Reproduced with permission of the editor. All rights reserved)
were incorporated into the model. This was a very original application of this
model.
Faust et al. (1996) presented a flowchart for a system (Fig. 2.6) to detect knots
and cracks in red oak, black walnut and hard maple, with 3D analysis of the
position, orientation and shape of these features. The basis of this analysis is
the density. The system first processes each single image and then relates
processed results of each CT image across neighboring CT images to pinpoint
defects. The most current algorithms link annual ring structure in an edge-
detected image. Locating the pith is an important step in identification of
Basic Phenomena 27
initial image number i=O
read ith image file

.... _- - ......... --_ ................ - ........ -- .. __ • _ _ _ .. _ . . . . . . . . . . . . . . w _ _ . . . . _ _ _ .. _ _ _ _ . . . . . . . . . . . . _ _ . . . . . . . . . ..
area-based threshold selection pith locating
selective smoothing
segmentation of smoothed image

pith [i], area,
--------segmentatioli ------------------ orientation,
shape limit value
yes defect-like
~--+I region list
extraction of defect-like regions

r--------,
defect-like
region list
correlation for each group
correlation across image slices
Fig. 2.6. Flow chart for the detection of cracks, knots, earlywood, and latewood in a hardwood
(Faust et al. 1996) (lOth International Symposium on Nondestructive Testing of Wood 2002, Press
Polytechniques et Universitaires Romandes, p. 202.Reproduced with permission of the editor. All
rights reserved)
Table 2.5. Precision of pith location in three hardwood logs using

the algorithm proposed by Faust et al. (1996) with the precision
within 5, 10 and 20 pixels. (lOth International Symposium on
Nondestructive Testing of Wood, 2002. Press Polytechniques et
Universitaires Romandes, p. 205. Reproduced with permission of
the editor. All rights reserved)
Species Images with Pith location At precision

5 pixels (0/0) 10 pixels 20 pixels
Maple 54.5 77.3 99.0

Oak 60.5 83.0 98.9
Walnut 83.2 97.8 100.0
growth rings and the longitudinal axes of knots and cracks that normally pass
through the pith. Table 2.5 shows the percentage of images with pith location
precision with different numbers of pixels ranging from 5 to 20. The best loca-
tion was obtained with 20 pixels for walnut, a semi-ring porous species.
2.3 Equipment for Imaging Techniques
In 1979, the works of Cormak in South Africa and Hounsfield in England were
recognized with the award of the Nobel Prize for Medicine for the develop-
ment of ionizing radiation computed tomography. Cormak (1963) manually
scanned collimated beams of 60Co gamma rays across aluminum and wood
phantoms and detected the transmitted photons with a GM counter. He recon-
structed the image of the scanned sample from one-dimensional projections
viewed at different angles in terms of a map of linear attenuation coefficients,
using a mathematical model he developed. Hounsfield (1972), using a much
more intense X-ray source, produced images of a human head in a run time of
several minutes, using a one-line minicomputer. This was the starting point for
the development of X-ray computed tomography for medical use. It is also
notable that today the unit used for X-ray imaging with medical scanners is
the Hounsfield [HJ. The Hounsfield scale for medical scanners is between -
1000, corresponding to attenuation in air, and +1000, corresponding to atten-
uation in human bone, and is 0 for attenuation in water. This scale is linear
between 1000 and 0, and its accuracy is ±2 H units.
Medical scanners were used in the 1980s for nondestructive inspection of
different nonmedical objects, and new devices were developed for specific pur-
poses. Medical scanners were first used for imaging wood specimens because
the physical properties of wood do not differ greatly from that of the human
body. The energy normally available on a medical scanner is about 70 ke V, with
soft X-rays. Presently, in wood science and technology, four generations of
scanners are being used. A detailed description of each type of scanner will be
given in the next chapter.
The development of the nonmedical scanners, first experimental and later
commercial, started in the mid-1970s (Reimers et al. 1984). A wide range of
applications were emphasized, ranging from an investigation of the foaming
behavior of various coals at high temperature, concrete specimens, cracks and
delaminations in metallic structures, defects in ceramics, fluid distributions
inside porous simulated rock cores, wood structures and trees, thick plastics
to a maximum 400mm, crack detection in welded steel bridges, internal fea-
tures in jet and automobile engines, fluid flow measurements in nuclear engi-
neering, to a complete inspection of the Trident C-4 rocket motor which is
nearly 2 m in diameter. However, the cost of the systems is a real barrier, and
the choice of testing methods must be made chiefly on economic grounds. The
industrial systems in the wood industry require high energy X-ray sources,
accommodation for a wide range of specimen sizes and densities and opera-
tion in industrial environments with a speed imposed by effective wood
processing.
The major system components of an X-ray scanner are: X-ray generation
system, fan beam or parallel beam X-ray collimator, system for data collection
and analysis, image display software, mechanical gantry with possibility of
Equipment for Imaging Techniques 29
rotation or translation or of simultaneous translation and rotation of the

sample under inspection.
2.3.1 Description of the Equipment
Fixed or portable equipment has been used for different purposes in wood
science and technology. If the test object can be brought to the scanner, which
is the case of logs and boards, fixed equipment is recommended. In this case,
an X-ray source has commonly been used. The portable equipment was
designed to be used for in situ inspection of trees, poles and building elements.
Gamma rays, which are mono energetic, were used in this case because they
have many advantages in the wide range of energies available, portability and
cost. In spite of these superior characteristics, gamma ray sources are of
low intensity, which is 104 times less than X-ray tube sources (Gilboy 1984).
Malan and Maris (1992) have summarized the use of gamma rays for wood
densitometry.
2.3.1.1 Fixed Equipment
Four types of scanner systems, referred to as "generations" have been devel-

oped, as can be seen in Fig. 2.7. Each type of the architecture of the scanner
has advantages in terms of speed and artifact generation, which make it ben-
eficial for specific applications.
The first-generation scanner (Fig. 2.7a) is a single-detector translate-rotate
system with a single X-ray source detector. The reconstructed 2D tomograms
are based on the attenuation coefficient measurements obtained when the
specimen was rotated through 180°. The source-detector pair move in unison
past the sample. The system is relatively inexpensive, simple in data collection
and requires simple algorithms for tomogram reconstruction. Any size of spec-
imen can be inspected. Despite its advantages, this scanner is speed limited and
is used only for laboratory measurements.
The second-generation scanner (Fig. 2.7b) is a translate rotate system with
an array of detectors, which makes simultaneous measurements through dif-
ferent angles in a single traverse. The geometry of the fan system is simple,
data collection and reconstruction algorithms are all simple. Unlimited size
samples can be inspected. This system is very slow because of the time needed
for the mechanical operations with the sample and for the collection of
unneeded data at the beginning and at the end of each traverse.
The third-generation (Fig. 2.7c) scanner rotate-only system has a detector
array in an arc focused at the X-ray source. The data are collected in a fanning
movement and for 180° data the sample is rotated 180°. The advantages of this
system are: simultaneous collection of data through the entire sample for each
X-ra source detector
a)
"
detectors
X-ray source
c)
Fig. 2.7. Scanner system with detectors perpendicular to the axis of rotation (Schmoldt 1996).
a First generation, a single detector, translate-rotate system (fan system). b Second generation,
multidetectors, translate-rotate system. c Third generation, rotate only.
view, simple rotation of the sample, with a continuous mechanical motion,

short scan times because of simultaneous collection of data. The limitations of
the system are related to the single slice data collection, the diameter of the
sample and number of the detectors, and high cost because of a large number
of detectors required for large objects.
The fourth-generation scanner is the most rapid, where only the source
moves (Fig. 2.7d). The data collection is based on a fan system, with detectors
Fig. 2.7. d Fourth generation, stationary detector, rotate only (with permission)
located on a circle that surrounds the X-ray source and the sample. The main
advantage is the simultaneous data collection through the entire sample for
each view, resulting in fast scans. Due to the very high cost, the system is rarely
used.
To avoid the limitations of all four generations of scanners, an alternative
was proposed with a scanner operating tangentially as can be seen from Fig.
2.8. The main advantage of this system is its ability to collect data for the entire
volume of the object simultaneously from many cross-sectional slices. Most
images are generated using data from a single detector. The speed is faster than
that of the third- and fourth-generation systems. The mechanical motion
system is very simple. The limitation of this system for the moment is the lack
of a fast reconstruction algorithm.
The capability of the tangential system to generate images of any planar or
curved surface within the object was demonstrated in laboratory work by
Lindegaard -Andersen et al. (1990) for a block of pine wood of 60 x 60 x 51 mm.
The distribution and shape of the annual rings were clearly identified. The
detector array is placed parallel to the axis of rotation of the sample and per-
pendicular to the cross section. A fan-shaped X-ray beam is formed by the
source and the detector array. The number and spacing of detectors determine
how many tomograms can be collected simultaneously.
"' .
Fig. 2.S. Tangential scanning system. (Schmoldt et al. 1999, with permission)
2.3.1.2 Portable Equipment
Portable equipment for computed tomography requires an array of detectors.

The X-ray tubes rotate in a plane perpendicular to the longitudinal axis of the
sample. Usually a thin fan beam is intercepted by a detector array. The X-ray
beam has a finite thickness that defines the "slice" plane or cross section that
is cut through the sample. After a complete rotation of the X-ray tube and
detectors, the computer receives the transmitted X-ray intensity for each
narrow beam passed through the sample. From data obtained for all paths
through the sample, at various angles of rotation, the computer calculates the
average of the X-ray linear attenuation coefficient, 11, for each small volume
element (voxel) within the slice.
2.3.2 Factors Affecting the Quality of the Image
Qualitative and quantitative information can be obtained from the X-ray com-
puted tomography images. Qualitative applications require only visual assess-
ment of the images and include evaluations of internal flaws and orientation
of the samples. Quantitative applications of X-ray imaging are obtained in
Table 2.6. Optimal X-ray energy, combined effect and transmission probability for softwood
species (1 mm thick) (Olson et al. 1988, courtesy of Wood Fiber Science)
Species Energy Low High Density Combined Transmission

(keV) density density range effect probability
(kg/m') (kg/m') (m'/kg) (maxdP)
Balsam fir 5.621 240 760 0.52 2220 0.40

Grand fir 5.531 250 680 0.43 2330 0.35
Sitka spruce 5.379 220 650 0.43 2520 0.38
Jack pine 5.266 240 580 0.34 2600 0.31
Shore pine 5.690 310 720 0.41 2060 0.30
Ponderosa pine 5.423 270 620 0.35 2380 0.30
Red pine 5.472 270 650 0.38 2310 0.31
Scots pine 5.127 230 520 0.29 2810 0.29
addition to qualitative ones and are related mainly to the measurement of the
density, moisture distribution and porosity of the wood. To obtain accurate and
reliable data from X-ray imaging, the scanning must be carried out in a very
consistent manner as was demonstrated by Lindgren (1985, 1991a,b), Liu et al.
(1988), Moschler and Dougal (1988), Olson et al. (1988), Fioravanti and Ricci
(1991) and others.
2.3.2.1 Beam Path
X-rays travel in a straight line through materials. The monoenergetic beam will
radiate perpendicularly on a piece of thin wood with uniform thickness and
heterogeneous density structure, consisting of four major elements (R, C, N,
0) and various ash ions (Ca, K, Mg, P, Si) present in minor quantities
(0.15-0.37% based on oven dry weight). Two main parameters (Table 2.6) are
related to the beam path as defined by Olson et al. (1988): the range of trans-
mission probability and the combined effect. The range of transmission prob-
ability (AP) is defined as "the radiation resolution of a given wood X-ray
experiment, in reference to the separability of transmission probabilities
resulting from density variations in the radial direction of wood:' The com-
bined effect C is a global factor taking into consideration the combination of
X-ray energy, wood specimen thickness and heterogeneities induced by the
anatomic structure. Consequently, the maximum radiation resolution is deter-
mined by density variations in wood specimens, X-ray energy (keV), X-ray
intensity (rnA), spatial and contrast resolution, display contrast, and beam
hardening. The quality of the tomographic image is directly related to the X-
ray energy. Because X-ray attenuation is highly dependent on the energy of the
X-ray used, it is always necessary to provide this information for laboratory or
field measurements. Theoretical studies published by Olson et al. (1988) have
1.00
g 0.80 LEGEND
:g
.n
0
....
p. pine heartwood
=
0
'in
0.60 t= 1.0 mm
·s H = heighest
en
en L = lowest
§ 0.40
p
0.20
0.00
I 2 5 10 20 50 100
photon energy (keY)
Fig. 2.9. Transmission probability and X-ray energy in pine heartwood of 1 mm thickness (Olson
et al. 1988). H Highest value; L lowest value. (Courtesy of Wood Fiber Science)
demonstrated that accurate wood density measurements can be achieved only

if the radiation used is in the optimal range. Values of optimal X-ray energy
for eight species are given in Table 2.6 and are in the range of 5.127 and 5.69
ke V for specimens of 1-mm thickness. The transmission probabilities under
various combinations of radiation parameters are between 0.29 and 0040. The
transmission probability increases with increasing X-ray energy and decreases
with increasing density of the sample. The relationship between maximum
transmission probability and X-ray energy is shown in Fig. 2.9. The maximum
transmission probability corresponds to the maximum value of LlP, identified
at the peak of about 6keV. The use of a monoenergetic beam is recommended
because the radiation resolution reaches its maximum when the polyenergetic
X-ray is converted to photons of the same energy.
X-ray attenuation in wood densitometric analysis is influenced by the quan-
tity of oxygen and carbon present in the specimen. The influence of photon
energy on mass attenuation coefficients of C, 0, Nand H and on spruce heart-
wood is shown in Fig. 2.10. It is recommended to choose the photon energy in
the linear part of the diagram and to avoid the abrupt zones corresponding to
the presence of different elements in wood.
Lindgren (1991a) theoretically obtained an attenuation coefficient of
500 kg/m3 for various contents of cellulose, hemicelluloses, lignin and CaC0 3
as can be seen from Table 2.7. Because of the low content of Ca «1 %), this ion
has no significant effect on the attenuation coefficient and CT numbers. A
similar situation was observed for other ions (Mg, Mn, K, etc.) present in wood
in very small quantities.
spruce heartwood
10- 1
10- 1
photon energy (keV)
Fig. 2.10. The influence of photon energy on mono atomic attenuation coefficients for C, 0, H, N
and for spruce heartwood. (Liu et al.1988, courtesy of Wood Fiber Science, with permission)
Table 2.7. The influence of CaCO, on the attenuation coefficient of wood and on CT numbers.
(Lindgren 1991a, with permission)
Cellulose (%) Hemicellulose (%) Lignin (%) CaCO, (%) flwood CT number
45 30 25 o 0.0878 -527
45 29.5 25 0.5 0.0882 -525
45 29 25 0.0886 -523
2.3.2.2 Spatial and Contrast Resolution
Spatial resolution is the ability of the system to distinguish two or more closely
spaced high contrast details of the internal structure of the specimen under
test. The spatial resolution of any imaging system is controlled by the accuracy
of ray paths through the sample and is related to the following factors: opera-
tive size of the source and detector, distance between the source/detector
and the sample, number of independent projections produced by the system,
ability to separate overlapping features information, signal-to-noise ratio
at small source and detector apertures (usually <0.25 )lm), and the cost of
computation.
The contrast resolution is the ability of the scanning system to distinguish

small differences in contrast in relation to the homogeneous contrast of the
background and is directly related to the linear attenuation coefficient. The
smallest difference in density of the sample that can be identified with a given
system depends on the contrast resolution of the scanning system. A low-
contrast detail of the structure can be identified if it is large, whereas small
details can be identified if they have a large contrast relative to the homoge-
neous background. This is due to the partial volume effect. The smaller details
of the structure must be of significant attenuation contrast to affect the voxel
average. Lindgren (I991a) studied the relationship between wood density and
the X-ray attenuation coefficients and CT numbers with a medical scanner and
specified that the minimum volume required must exceed 0.005 x 1 x 1.5 mm.
For each scanning system, this relationship between the density and CT
number must be determined precisely for the correct interpretation of data.
The range of CT numbers displayed in a given image is referred to as the
image "window". This range is mapped and is represented by a discrete number
of gray levels to create the image. A pixel (picture element) is the bidimensional
representation of a voxel (volume element) on which two dimensions corre-
spond to the pixel dimension, and the third dimension corresponds to
the thickness of a slice (I-IOmm). The field of view (FOV) is equal to the
maximum dimension of the scanned specimen. As noted previously, on the
image each pixel represents the average value of the attenuation coefficient of
the volumetric unit. For correct reproducibility of the image, the technical
parameters to be defined are: dimension of the FOV, thickness of the slice, time
of scanning and the algorithm of reconstruction. For calibration, FOV is
usually taken between 3 and 50 cm. For image reconstruction FOY is related to
the pixel size as can be seen from Table 2.8. The minimum FOY size of 30 mm
is related to the representative size of the anatomic structure of wood. The
selection of the thickness of the slice is also a function of the size of the pixel
and the voxel as well as the intensity of the beam. The thickness of the slice
gives the volumetric resolution of the image as shown in Table 2.9. The value
attributed to the voxel corresponds to the average value of all anatomic ele-
ments from the voxel. Important errors can occur when the densitometric
analysis is performed at the limit of an annual ring, that contains latewood and
earlywood. For this reason it is recommended to select the minimum dimen-
sion of the pixel with the thickness of 1 mm, if possible. The selection of the
intensity of the beam is a function of the slice thickness and of the volumet-
Table 2.8. The size of the pixel for the array of 512 x 512 as a function of the size of the FOY.
(Fioravanti and Ricci 1991, with permission Holzforschung)
FOV (mm) 500 400 300 200 100 50 30

PIX (mm) 0.976 0.781 0.586 0.390 0.195 0.097 0.058
(pix = FOV/512)
ric resolution. For wood, the intensity can be higher than (130 rnA) for a scan-
ning time of 3 s. Fioravanti and Ricci (1991) published an example of the matrix
of density (Fig. 2.11) with all the measured values for all voxels, given in a
matrix of 11 lines x 5 columns. A specific voxel can be identified with the recon-
struction matrix of 512 x 512 elements. The visualization with points is a
concise representation of all experimental data, making possible the observa-
tion of the isodensity curves inside a selected annual ring (Fig. 2.12). The pre-
cision of the measurement of the attenuation coefficient, using X-ray computed
tomography, is ±0.5%, compared with conventional radiographic techniques
for which the precision is ±3%.
TabIe2.9. Relationship between the voxel volume and the slice thickness (Fioravanti and Ricci
1991) (with permission)
Pixel (mm) Voxel Volume (mm 3 )

Slice thickness 1 mm Slice thickness 5 mm Slice thickness 10mm
0.586 0.343 1.717 3.434

0.390 0.152 0.760 1.524
0.195 0.038 0.190 0.380
0.097 0.009 0.047 0.094
0.058 0.003 0.017 0.033
Fig. 2.11. The attenuation values superimposed on the tomographic image of annual rings
pattern. (Fioravanti and Ricci 1991, with permission)
Fig.2.12. Isodensity zones in an annual ring. (Fioravanti and Ricci 1991, with permission)
2.3.2.3 Anisotropic Direction
Theoretically, density is a scalar and not a vector and a priori is not influenced
by the anisotropic direction of wood. Nevertheless, because wood is a natural
composite, laboratory studies have reported the influence of the anisotropic
direction on the measurement of attenuation coefficient (Moschler and
Winistorfer 1990; Fioravanti and Ricci 1991; Malan and Marais 1992;
Karsulovic et al. 1999). For precise densitometric measurements, it is
recommended to orient the irradiation direction of the specimen with the
longitudinal anisotropic direction of the wood. In this case, the number of
photons absorbed by the cell wall is minimal. If scanning along radial or tan-
gential directions is required, it is obvious that the scanning inspection must
be performed parallel to the selected anisotropic direction. The attenuation
coefficient depends on the thickness of the cell wall. Errors in attenuation mea-
surements are created if the specimen is presented in an inclined position
versus the incident radiation, because the experimental wall thickness is
greater than the theoretical.
Table 2.10 gives the effect of anisotropic direction on the linear attenuation
coefficient measured on 141 wood specimens {Malan and Marais 1992}. The
longitudinal values compared with those in radial and tangential directions are
the lowest values. No significant differences were observed between the radial
and the tangential coefficients. The linear relationships between the density
and the linear attenuation coefficients in radial and tangential directions for
Pinus radiata are given in Fig. 2.13. The mean value of the attenuation coeffi-
cient in the radial direction was 4% greater than in the tangential direction
but the difference is not significant at the 95% confidence level. Moschler and
Table2.10. The effect of anisotropic direction on the linear attenuation coefficient. (Malan and
Marais 1992, with permission Holzforschung)
Species Source Mean linear Attenuation Coefficient (cm- 1)

Radial Tangential Longitudinal
Coniferous 137Cesium 0.0406 0.0400 0.0392

Hardwoods 137Cesium 0.0567 0.0571 0.0571
Coniferous 241 Americium 0.0986 0.0997 0.0943
Hardwoods 241 Americium 0.1390 0.1386 0.1389
0.14...-----------------------,
0.13 radial
irradiation
S 0.12 tangential
-2 irradiation
::, 0.11
...l
:::1. 0.10
0.09 .j. _ _ _
knot
~-/_-..r
0.08 H-HH-+-t--H,-t-t-t-H,-t-t-++t-t-t-++-t-HH-+-t-H,-t-t-+-t-t-t-t-+-H
o 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
point measured on Y axis (em)
Fig. 2.13. Relationship between the density of Pinus radiata wood and linear attenuation coeffi-
cients in radial and tangential anisotropic directions. (Karsukovic et a1. 1999, courtesy of Forest
Product J)
Winistorfer (1990) emphasize the necessity of the alignment of the aperture

with the anisotropic directions. Misalignment induces more measurement
errors. Routine laboratory scanning for densitometric purposes is always per-
formed in the radial direction and requires good alignment between the
growth rings and the aperture. For wood species with large cell lumina, the
porosity can also be determined by scanning in the radial direction (irradia-
tion in the longitudinal direction) with a convenient size of the aperture. To
avoid any underestimation of density and all the experimental artifacts, the
authors proposed that the attenuation coefficient be measured in the tangen-
tial direction. To our knowledge, until today, no data have been reported in the
literature related to the micro densitometric analysis of growth rings in the tan-
gential direction because of the difficulty in observing the limit of the annual
rings.
2.3.2.4 Beam Hardening
From a theoretical point of view, the X-ray attenuation coefficient is calculated

for a monochromatic source having a single wavelength or energy. However,
Table 2.11. CT numbers of different materials. (Hattori and Kanagawa 1985; courtesy of The
Japan Wood Research Society)
Material Density (kg/m3) CT number surrounding CT number surrounding

with granular sugar with air
Polyprophylene 905 -120 -87

Polyethylene 953 -71 -39
Nylon 1129 71 127
Aeryl 1189 130 185
Polycarbonate 1191 97 139
Polyacetat 1405 342 418
Polyvinylchloride 1401 1315 1622
the scanner produces polychromatic X-rays. The attenuation of a polychro-

matic wave is not constant for all energies and is higher for the lower energy
components of the X-ray beam (effect of beam hardening). This effect can sig-
nificantly decrease the accuracy of quantitative results. However, it can be
diminished by surrounding the sample with different compensatory materials,
such as paraffin wax, granular sugar or acrylics, for small, clear specimens of
wood (2 x 2 x 1 cm) as noted by Hattori and Kanagawa (1985). Beam harden-
ing can also be compensated by reconstruction algorithms, as noted by Vinegar
(1986). Calibration of scanners with attenuation standards is an essential step
for calculation or comparisons of X-ray attenuation data and for enhancing of
the accuracy of experimental measurements. As for classic radiographic analy-
sis, calibration of the scanning system must be performed every day before
starting the experiments to identify and characterize the drift of the X-ray
source and the detectors. The reference material should exhibit characteristics
close to those of the material to be analyzed. Correction must be made prior
to calculation of density. Hattori and Kanagawa (1985) proposed the following
materials as standard attenuators for wood specimens: polypropylene, poly-
ethylene, nylon, acrylics, polycarbonate, polyacetal, and polyvinyl chloride. The
CT numbers of these materials surrounded with air or with granular sugar are
given in Table 2.11. CT numbers surrounded by air are inferior to those sur-
rounded by granular sugar.
2.4 Applications
The field of applications of computed tomography is very large, including

building inspection, wood preservation, stability of wooden construction ele-
ments, preservation of monuments, new constructions, pole inspection, inspec-
tion of wooden toys, tree examination and inspection, arboriculture, growth
rate assessment, wood quality assessment, forest sciences, wood technology,
wood biology, dendrochronology, dendroarcheology, climatology, geomor-
Applications 41
phology, phytopathology, tropical trees, plantation trees, and wood products

(Rinn 1991). The choice of methods for specific inspections must be made on
economic grounds. Using the great potential of this nondestructive method, it
is possible to study the spatial distribution of matter on a scale from meter to
micron in a very wide number of applications.
2.4.1 Examination and Inspection of Trees and Poles
X-ray computed tomography opened a new era for the nondestructive exami-
nation and inspection of trees and poles. The quantitative determination of
decay, knots, checks, heartwood formation, ruptures, wet core, spreading of dry
decayed tissue and other major defects as well as the distribution of moisture
inside the stem of a tree is now possible. Fundamental differences among the
images obtained for different species are easy to recognize.
Pioneering research results related to the development of portable instru-
ments, suitable for outdoor use were published by several authors (Habermehl
1982a,b; Onoe et a1. 1984; Habermehl and Ridder 1992a,b,1993, 1994, 1996;
Niemz et a1. 1998a,b). A description of these devices will be presented in the
following pages.
For the examination of forest trees or of park and street trees, Habermehl
and Ridder (1998) proposed two types of portable instruments, one with a par-
allel beam system (Fig. 2.14) and another with a fan beam system (Fig. 2.15),
equipped with a y-ray source. The source and detectors are sustained by the
mechanical device shown in Fig. 2.16.
l:?
o
'0
E laptop computer
Fig.2.14. Portable apparatus for the examination of tree with a parallel beam system.
(Habermehl and Ridder 1998, with permission)
fan vith 3D detector
measuring plane
within the tree
motor for rotation
bearing ring
fixed ring with 4 legs
source of radiation
Fig. 2.15. Portable apparatus for the examination of trees with a fan beam system. (Habermehl
and Ridder 1998, with permission)
Fig. 2.16. Mechanical device for the sustaining of the source and detectors of the portable appa-
ratus for the examination of trees. (Habermehl and Ridder 1992b, with permission)
The size and location of the defects that can be observed on a tomographed
cross section are cavities and decay of the trunk, mechanical injuries, fissures,
holes, zones destroyed by fungi, areas of lowered density with less moisture
than in healthy zones, the presence of metallic inclusions in the trunk, like nails
Applications 43
and bullets, zones of perfectly healthy wood, and the difference between
sapwood and heartwood in some species because of the difference in their
moisture content. The devices are very effective for the examination of park
and street trees. The decision of the environmental authority to conserve trees
or not after examination of tomographic images can be objective, and the
safety of the traffic on the streets can be assured. For forest trees, the questions
answered by the portable apparatus are related to the effect of liming
and fertilization on the sapwood to heartwood ratio, to the water pathways
above the cross section of the trunk, to the existence of frost cracks, and to the
existence of wet cores and other specific internal structures that can affect
the variation of the attenuation coefficient of the radiation and consequently
of the density.
The source of radiation is 137 cesium having a half-life of 30 years and emit-
ting y-rays of a quantum energy of 662keV. The parallel beam system is
equipped with a source of radiation of 13 GBq and allows the translation and
rotation of detectors and only translation for the source. The detector system
is consists of three detectors with an angle of 3.75° between them. One trans-
lational displacement produces three simultaneous attenuation profiles at the
same time. A total of 48 profiles are recorded in 16 rotational positions. The
smallest diameter of the tree to be measured is 12 cm with a counting rate of
19,000 counts/s and the largest diameter of the tree is 72cm. The tomograms
reconstructed on the computer are performed in less than 2 min for one profile.
The reconstruction process is based on the analysis of a matrix that contains
values of the attenuation coefficient (cm-I ) of the examined slice. In the case
of a tree, the reference value was considered to be 83 units for water and 0 for
air. (Hounsfield units in medicine are -1000 for air and zero for water.) The
spatial contrast for the standard adjustment is 10 mm for the full width at half-
maximum (FWHM) of the signal. For trees with diameters up to 48 cm, the res-
olution must be 7.5 mm FWHM of the signal. The contrast resolution obtained
was 2% and was related to the water attenuation.
The fan beam system is a more complex system and has 30 detectors. The
source is 185GBq and 8000 counts/s, mounted on a metallic ring with a 100cm
diameter. The complete data set produces 150 profiles, measured with a rota-
tion of the source-detector system with a step of 2.4°. The fan system has the
same spatial and contrast resolution as the parallel system. The algorithm
reconstruction for the PC time is 4 min for one profile. For larger trees the total
examination time is about 1 h. The tomograms can be coded by color or by a
gray scale, as can be seen in Figs. 2.17 and 2.18.
2.4.1.1 Growth Rate Assessment
Using computed tomography Onoe et al. (1984) were the pioneers in the
development of an automatic method for ring width measurement and ring
counting of several species, including red pine, black pine, Douglas fir,
a)
24
30
--
36
42
48
--_.. _---
54
60
66
72
78
84
90
96 24
30
------
36
42
48
54
60
66
72 ----
78
84
90
%
Fig.2.17. Tomogram on a color scale of a transverse section of a healthy (a) and of a decayed (b)
Norway spruce tree. (Habermehl and Ridder 1996, with permission)
Fig.2.18. Tomograms on a gray scale (Habermehl and Ridder 1996). Legend: a) a lime tree with
internal decay and big holes b) an ash tree, showing a metallic inclusion (with a very high atten-
uation) (with permission)
Japanese cypress, birch, horse chestnut and maple. The details of the annual
ring structure can be seen using high spatial resolution in the image recon-
struction with computed tomography. A few tenths of millimeter resolution
can be obtained with well-collimated, narrow X-ray beams and dense sam-
pling. The device designed and used by Onoe et al.(l984) was portable, with
an X-ray tube of 40-120keV and three NaI scintillation counters. A fan beam
algorithm was used for reconstruction, and 1200 samples of 16-bit projection
data were taken at 2° intervals. The reconstruction images show very clearly
Applications 45
the annual rings in Douglas fir as well as the heart wood central zone, having
decay (Fig. 2.19).
Figure 2.20 gives the profile and the width of automatically measured rings.
Firstly, the pith was determined by contour tracing of an inner ring. Secondly,
several diameters were drawn through the pith. Thresholding after differenti-
ation yields the number and the width of the rings. Hauffe and Mahler (2000)
studied the possibility of evaluation of log quality using only the width of
annual rings measured with a scanner on spruce logs from southern Germany.
The technique was successful for 80% of the logs. Rinn (1991) proposed a
mobile tomograph with 12 sensors for scanning growth rings in trees and
poles.
Fig. 2.19. Reconstructed images of the cross section of Douglas-fir pole compared with a photo-
graphic image of the same section (Onoe et al. 1984). a The reconstruction image of the wrap-
ping pattern of the annual ring around a branch. b The reconstructed image of the heart decay
in the central zone. c Photographic image of a. d Photographic image of b. (Reprinted from
Nuclear Instruments and Methods in Physics Research, vol. 221, 1984,213-220,2001, with per-
mission from Elsevier Science)
c
--I-HlIIIHHHt-Hfff-lI-fIH1HIIHHHI-lUllIlIIIHtItHHlftlH1HHnUfHHHHlI----
Fig. 2.20. Profile and width automatically measured annual rings (Onoe et aL 1984)_ (Reprinted
from Nuclear Instruments and Methods in Physics Research, voL 221, 1984,213-220,2001, with
permission from Elsevier Science)
Another aspect of the examination of growth rate of trees where the appli-
cation of X-ray computed tomography is very successful is related to the devel-
opment of microtomography for 3D anatomic structure examination, as a very
important step in the further development of wood microdensotometry in
general. Davis et al.(l991) reported very well-defined images of transverse
structures of wood of Pinus sylvestris and Eucaliptus delegatensis, as can be
seen in Fig. 2.21. Details of the structure of the annual growth rings are clearly
visible, and tracheids, rays and resin canals can be observed on the Pinus
sylvestris specimen. For Eucaliptus delegatensis, a collapse check in the high-
density zone is observed. The device for X-ray microtomography of wood was
attached to an X-ray diffractometer. The schematic representation of this
device for X-ray microtomography is shown in Fig. 2.22. The technical char-
acteristics of this X-ray microtomograph are as follows:
- X-ray energy: 8 keY;
- beam characteristics: two beam collimation arrangements, one cylindrical
with 20 Jlm and another rectangular of 5 Jlm width and 100 Jlm height;
- translation movement 255 stepwise, followed by 1.50 rotation repeated 120
times for 1800 rotation of the specimen;
- reconstruction algorithm using a summation-convolution back-projection
method;
- data collection time: 7 h for 20-Jlm scan step;
- computation time for image reconstruction: 5 min;
Applications 47
Fig.2.21. Reconstructed images of the anatomic structure of Pinus sylvestris (Davis et al. 1991).
a 5-l.Lm spatial resolution of the transversal section (0.9 x 0.6mm) of the annual ring with early-
wood and latewood zones. b 5-l.Lm spatial resolution of the transverse section (0.9 x O.5mm) in
which the highest density is observed in the middle lamella between the cells. The images were
obtained with the first generation scanner (8keV, X-rays from the Ka peak of a eu target tube).
(with permission)
- pixel gray is zero (black) for the air and 255 (white region) for highest
density in latewood;
- X-ray attenuation coefficient is defined with 5% accuracy;
- spatial resolution 111m;
- specimens section 2 x 2 mm of Pinus sylvestris and Eucaliptus delegatensis;
- annual ring width: 0.4 mm.
.,..
ex>
RS232C
r-,
------·1 Link r - - - - - - - , ~i
interface controller ~ Micro-Computer! ), 0'
8.
(MC68701) System Graphic Display t::.
::;I
11v------/ .
k I and (JQ
r-----~I . i I
{l Hard disk I Ii ,Image Processing ~
!!;.
o·
::;I
stepper III----- n
X-Ray generator o
motor drivers .g
RS232C Link ;;.
(!)
0-
S'
y~ Ni filter o
3
(JQ
H r ~
X-Ra;tube I ~
I! (Cu ~/S\~ nl ' JI ~
Target) r n ~ n: object: .. jdetector Na(TI)~';l Data Acquisition
. U U\~~~/ ~ uI 1~lsystem
C1 C2 ~ C 3 '~- - - - - - - - - '
\ translate - rotate stages
Fig. 2.22. X ray microtomographic device. (Davis et al. 1991, with permission)
Applications 49
Improvement of a miniature scanner allows (Sasov and van Dyck 1998) to

produce images of biological specimens obtained on a X-ray microtomograph
coupled with a corresponding microscope. The device fits in any laboratory
environment and enables views of the structure in 3D, under ambient condi-
tions, without any need for specimen preparation.
Badel (1999) produced 2D images of oak at a microscopic scale using an X-
ray device with 70 kV and 95 ~A. Vansteenkiste (2000) studied the ability of
automatic analysis of X-ray images obtained in 2D for the detection of the
annual ring limit, fiber and vessel zones and rays of oak.
2.4.1.2 Moisture Content in Trees
The moisture content of living trees is related to the viability of trees and is
dependent on many factors such as climatic conditions, the moment of the day
when the test of water flow is made, the sanitary conditions, etc. The forest
decline observed today seems to be induced by the disturbance in water trans-
port in trees.
The portable computer tomograph proposed by Habermehl and Ridder
(1992a,b) and used by Szendrodi et al. (1994) permits mapping the distribu-
tion of water inside the trunk and observation of the mechanism of water
transport in xylem. Raschi et al. (1995) and Tognetti et al. (1996) utilized a
portable apparatus equipped with a parallel beam system and with a source of
radiation of 13 GBq for measurement of the water content in the stems of
sessile oak (Quercus petraea) and in Turkey oak (Quercus ceris) (Fig. 2.23) . For
the portable computer used in this experiment, the CT numbers are on a scale
corresponding to 83 units for pure water and 0 for the density of pure air, on
a grid of 75 X 75 pixels for Quercus ceris. and of 63 x 63 pixels Quercus petraea
with a reconstruction interval of 3-4 mm. The tomograms for Quercus ceris
displayed green and blue colors, indicating a high density in the stem. The
sapwood had a high density (green color) and the heartwood had slightly lower
density (dark blue color). In Quercus petraea, the sapwood (blue and green
rings) is well hydrated. The embolized zone is represented by the violet ring.
The heartwood of highest density is represented by a dark blue color. These
species are characterized by different susceptibilities to drought; the first one
is a mesophytic type species while the second is xerophytic. The trees origi-
nated from a forest in Hungary at latitude 47°N, longitude 20° and altitude
320-340 m, with an annual precipitation mean of 600 mm. The trees from this
forest had suffered a prolonged period of drought for many years before the
experiments, which were performed in the spring. Characteristic anatomical
structures of these species are schematically given in Fig. 2.24, where large
vessels can be seen in the earlywood. The main function of large vessels is to
increase the efficiency of sap flow. In this tissue, cavitation risk is always
present. When the large vessels, malfunction, the sap flow is supplied by the
a) b)
2". 00
,. ...
,..aD 2 •. 00
JO .OO
• . 00
-
11.00
Fig. 2.23. Tomograms of Quercus petraea and Quercus cerris (Tognetti et al. 1996). a Quercus
".00
cerris; b Quercus petraea. N, E, S, Ware the cardinal points; 1,2,3,4 are the position of measured
--
.....
N."
points The density scale in CT numbers is represented by different colors (i.e., CT = 30 corre-
sponds to dry condition and CT = 96 corresponds to high moisture content). (with permission)
o
o
I ,
Quercus ceris
(b)
o
Quercus pelraea
Fig. 2.24. Anatomic structure of a Quercus petraea and b Quercus ceris. (Jacquiot et al. 1973, with
permission)
Applications 51
smaller latewood vessels and fiber-tracheids, which remain functional for years
(Zimmermann 1983).
In both species, a narrow ring of low density surrounded the high-density
ring which indicates that the outermost vessels buffer the sap flow rate (Granier
et al. 1994). The sapwood area is influenced in different proportions by the sap
flux as can be seen from the different colors displayed on the tomogram. The
relative uniformity of the density in Quercus ceris compared with Quercus
petraea could partially explain the better drought resistance of Quercus ceris.
In Quercus petraea a high proportion of red and violet zones on the images
indicate a relatively low wood density. The active sapwood appears in pale
green and blue zones corresponding to 90 and to 72 units, respectively. The
tree examined was not very healthy because of a prolonged dry period.
However, it seems that the high capacity of water storage in the sessile oak
vessels in the narrow external rings play an important role in tree survival
during the periods of low soil water availability in the summer. The attenua-
tion coefficient of the y radiation of well-hydrated tissue is twice that in dry
wood and a high correlation between the density of sessile oak sapwood
and its water content was noted by Weibe (1992). It was also observed that
the north-facing quadrant of the trees had a higher density. Tognetti et al.
(1996) confirmed the previous statements of Wiede (1991) for sessile oak
sapwood, namely, that in living trees the density is highly correlated with water
content.
The key point in the interpretation of tomograms is related to the calibra-
tion of the pictures. For this purpose Beres and Fenyvesi (1994) as well as
Raschi et al. (1995), Tognetti et al. (1996) suggested the use of several methods
(ultrasonic acoustic emission, thermoelectric method, etc.) together with a
computed tomography method. For quantitative interpretation of tomograms,
the data deduced with these methods must agree with anatomic studies as well
as with micro densitometric research.
2.4.1.3 Pollution Effects on Trees
Studies related to the pollution effect on trees (Pinus sylvestris) induced by

ammonia produced by a large cattle farm were reported by Katzel et al. (1997).
Twenty-eight 73-year-old trees were investigated in situ, using mobile com-
puted tomography. The nitrogen produced by the cattle farm and added to the
soil over more than 20 years induced changes in the energetic, carbohydrate,
amino acid and metabolisms of the tree needles up to a distance of 1 km from
the source. The wood quality of the trees was modified because of long-term
changes in metabolism. The ratio of sapwood to hardwood was modified, as
can be seen from Table 2.12 and Fig. 2.25.
The tree cross stem area in the highly polluted zone is larger than that in
the minimum polluted one. The corresponding ratio between the sapwood
and heartwood areas is 1.5 for the maximum polluted area and 3.2 for the
Table 2.12. The influence of ammonia stress pollution of Pinus sylvestris on the sapwood/
hardwood ratio and the moisture content of the trees. (Kiitzel et al. 1997, with permission)
Site Distance Pollution Cross stem Sapwood Hardwood Moisture Content CT number
(m) (area em') (%) Ratio Sapwood Hardwood Ratio
1 200 Maximum 624 53.8 1.5 81 50 1.62

2 280 745 57.8 1.5 81 51 1.59
2900 Minimum 513 77.0 3.2 86 53 1.62
minimum polluted area. The diminishing of the sapwood area induced a reduc-
tion of sap flow in the tree and indirectly a reduction of water supply in the
needles. The increasing heartwood area combined with the diminishing
sapwood zone in the polluted tree can be understood as an accelerating aging
process of the tree. The moisture content of both sapwood and heartwood
zones in the polluted area was much reduced when compared with the unpol-
luted area. The survival prognosis of tree in the polluted zone is reduced. The
results of this research demonstrated that computed tomography with a
portable device is a very useful tool for nondestructive diagnosis of living trees.
2.4.2 Wood Quality Assessment
During the last 20 years, there has been a growing interest in the development
of automatic methods for quality assessment of logs and lumber and in the
determination of optimal edging and cutting procedures in the lumber indus-
try (Birkeland 1985; Wagner and Taylor 1985).
The first commercial scanner developed in 1973 was relatively slow, requir-
ing 4 min per image. The rotation of a significant mass around a heavy object
like a log limited the rotation speed and increased the resulting scanning time.
Recent technological developments permit the production of appropriate
industrial devices able to detect small structural details. An equipment
resolution of about lOmm (or 1 cm3 in volume) is of interest for the nonde-
structive control procedure. Wagner et al. (1989 a,b) described the first
utilization of an ultra-fast computed tomographic medical scanner with
34 images/s for the detection of defects within hardwood logs. The fast scan-
ning time was achieved by eliminating all moving parts of the device, using a
fan detection system and a reconstruction matrix of 128 x 128 for a water
oak log with a diameter of 38 cm and a length of 3.6 m. The tomograms were
validated by visual comparison with real images of log cross sections, each
8mm thick.
Chiorescu and Gronlund (2000) noted that the recent technological
advancements in X-ray scanning techniques have a good potential for on-line
implementation in softwood sawmills. It has been shown that the sawing
Applications 53
24.00
30.00
36.00
42.00
48.00 --
54.00
60.00
66.00
d)
72.00
78.00
84.00
90.00
96.00
24.00
30 . 00
36.00
42.00
48.00
--
54.00
60.00
66.00
72.00
78.00
84.00
Fig. 2.25. The development of sapwood in Scots pine induced by ammonia pollution, compared
90.00
with a normal tree at 100 em stem height. Trees are located in a forest from the district Torgelow,
Germany (Katzel et al. 1997, with permission). A Tomogram for tree no. 1, with 624-cm' cross
96.00
section located at site 1, 200m from the farm (maximum pollution zone). B Tree no. 8, with 745-
em' cross section located at site 2,280 m from the farm. C Tree no. 2 at site 3, with 513-cm' cross
section located 2900m from the farm (minimum pollution zone). D Tree no. 8 at site 3 located
2900m from the farm (minimum pollution zone). The density scale in CT numbers is represented
by different colors, CT = 30 corresponds to dry conditions and CT = 96 corresponds to a high
moisture content. The tomograms were taken in September 1993 and 1994
process can be considerably improved by the utilization of software based on

X-ray scanning data deduced from the scanning of 600 Scots pine (Pinus
sylvestris) saw logs in Sweden (Grondlund et al. 1994). For hardwood logs, the
problems are even more complex because of the structural complexity and het-
erogeneity as well as the biological variability of their internal structure, both
among and within species. Schmoldt et al.(1999) reviewed the progress that
has occurred in scanning technology, image analysis, and data utilization in
the sawing of for hardwood logs.
The final aim of the development of industrial X-ray computed scanning
technology is to develop the scientific and technological foundation that is
needed to make this nondestructive control method operational for sawmills.
2.4.2.1 Inspection of Logs for Lumber and Veneer
The main interest for the inspection of the internal structure of logs is related
to the improvement of the grade and volume yield in the sawmill. Prior to
studies for the development of commercial industrial tomographic log scan-
ners for sawmill use, feasibility studies, and laboratory investigations were
made. The stress was put on the detection and location of knots, decay zones,
decay areas, cracks, internal holes, abnormal variation of density, abnormal
growth rings, reaction wood, resin pockets, spiral grain, embedded metal, etc.
(Taylor et a1.l984; Funt and Bryant 1987; Swanson and Hailey 1987; Hodges
et al. 1990; Maroc et al. 1996; Niemz et al 1997; Bahyl 1998; Karsulovic et al.
1999; Oja and Temnerud 1999; Nystrom and Earl Kline 2000; Sepulveda and
Gronlund 2000; Varga et al. 2000). The high moisture content in green logs
makes the detection and location of defects especially difficult. Moreover, the
scanning must be performed at a rate compatible with log conveyance speed
in sawmills. During the transportation through the scanner, the logs must
exhibit little side motion, rotation or vibration. The automatic interpretation
of log tomograms started in the 1980s as feasibility studies with automated
defect identification and labeling using different algorithms, first for softwood
and later for hardwood logs. The main objective was to find a procedure
for detection of defects able to operate at real time speed so that scanning,
image reconstruction, interpretation, and display could be integrated into mill
processing.
Scanning applications in sawmilling and wood manufacturing processes, as
suggested by Johansson (1985), can be divided into three main domains:
- the control inspection, where production specimens are checked randomly;
- research and development for new technological procedures;
- on-line internal inspection before sawing at normal production speed.
The following is a review of the literature related to the internal inspection
of logs before sawing. In softwood logs, the knots are the most common grade-
reducing characteristic. As an example, the studies performed by Karsulovic
Applications 55
et al. (1999) are chosen to demonstrate the ability of the linear attenuation
coefficient of gamma radiation to detect the knots in Pinus radiata. The
scanning devise is shown in Fig. 2.26. The irradiation of the samples (40 x 125
x 300mm) was done with a gamma ray beam (24lAm) of 59.5keV, with a 3-mm-
diameter collimator for source and detector. The sample was 131 mm from the
detector, and the distance from the source to the detector was 145 mm.
The array for the irradiation was 5 x 5 mm. The variation of the linear attenu-
ation coefficient in radial and tangential directions of the specimen was mea-
sured. Three zones of gray were used to distinguish knots from the
surrounding zone of defect-free wood. A transition zone is clearly observed
between clear wood and a knot, caused by a strong deviation of fibers from the
natural anisotropic axes.
Using an industrial scanner with automatic image analysis, Taylor et al.
(1984) presented the characteristic image of knots shown in Fig. 2.27. This slice
was compared with the image analysis of the same section and with the pho-
tographic image of the cross section of a southern pine log to determine the
correct location of knots, for which a good agreement was obtained.
In current industrial processing in choosing the best sawing pattern, the
location of knots as well as the log geometry will be determined first. Skatter
(1998) established the cross-sectional shape of softwood logs from three X-ray
projections and modeled the cross section as an arbitrarily oriented ellipse. At
the same time, the exterior shape, the shape under the bark, and the heartwood
zone were determined. The comparison with experimental results on 40 Scots
pine logs gave 1.64% accuracy in the prediction of the model. Secondly, the
best log orientation will be selected to produce the lumber with the highest
value for each log. Funt and Bryant (1987) developed computer algorithms to
detect knots, decay, and cracks in hemlock logs with a medical scanner. The
decay zone was the hardest feature to distinguish and was identified on the
basis of a dark shade and a disruption in growth ring pattern. The program
used a low ring uniformity pattern as a criterion for decay identification. The
algorithm required 3 min/scan, which was very long compared with the speed
of log conveyance in the sawmill. Wagner et al. (1989, a,b) reported the uti-
lization of an ultra-fast medical scanning system for logs of 38cm diameter,
with 34 cross-sectional scans per second. They noted the necessity for the
development of a specific industrial scanner for sawmills. The second-genera-
tion scanners with translate-rotate geometry required 20s/scan. The third and
fourth generation with rotate geometry only required 1 s/scan which is the
maximum speed attainable today in rotating scanners. A short scanning time
can be achieved by eliminating all mechanical movement parts, for example,
by using an X-ray fan scanning system.
Schmoldt et al. (1993), and Schmoldt et al. (2000) emphasized the necessity
for the development of specific computer algorithms for hardwood logs, which
have a more complex internal structure than softwoods. An increase in the
number of log defects increases the difficulty in making the best sawing deci-
sion. For automated interpretation of tomograms, data collection, and image
a) 0
Fig. 2.26. Sound knots and surrounding area with different levels of gray. a Three levels of gray:
dark zone for the knot, white zone for the sound wood, and gray for the zone around the knot
(Krasulovic et al. 1999, courtesy of Forest Products J). b Tomographic image on a transverse
section of fir, where the knots are visible as well as annual ring pattern, bark, sapwood, com-
pression wood and cracks starting from the core. (Varga 2000, with permission)
Applications 57
........ - ... .
• , ,11'" 1 ....... , ",. • •• "'
I • • • Ut ••••••••• , •• ~ ••••
... ... .... ,..

• 1 •• 1 ~ It. "'" • II.~ "'.
..............
I ••• U • • • • • • • • • •• ••• • •
~.t
·I·f.~
.re
• It •••••• II • I ••• 1. •• ...
'l!:..' • • • •• ••• •• • Fig. 2.27. Automatic image analysis of the
..f" cross section of a bolt with four knots (Taylor

• •• ". ••• , , ... I •• , •• ."
··~········B
"" , , " " 1 " , • •
~ et al. 1984, courtesy of Forest Products J). a
• •• '""';1 . . . . . , • • • • • • "
r;--;- . . Tomographic image. b Photographic image
U •• , •
• t • ~ •• " I • I • I ... II
II •••••••••••••••••••••
(the bolt was split horizontally before sawing
· . . • . . . • . . • " !.. t.. . : . • ~
the disk). c Image analysis
analysis must be supported by robust methods of defect recognition developed

previously with extensive data sets for different hardwood species and log
configurations. The first decision in a sawmill is made for the processing of
logs sorted as veneer logs or as high-quality saw logs. The computer simula-
tion demonstrated an improvement of 21 % of the value yield from red oak logs
(Steele et al. 1990). Criticism of the simulated data can be made because of the
ideal shapes of logs (cylinders or truncated cones), equations used to generate
the defects, and low numbers of rotations.
Grundberg and Gronlund (1997) compared the accuracy of visual grading
of Scots pine logs with simulated grading of the same X-ray scanned logs using
an X-ray source of 160keV. In the cross direction, the resolution was of 256
pixels, with a gray resolution of original signal of 4096 levels. Figure 2.28 shows
the X-ray intensity variation across a log section. In this study, the logs were
separated into three classes (butt, intermediate, and top logs) and processed
with a normal cant sawing pattern All the boards were dried to 18% moisture
content. The central boards were scanned on all four sides and then graded.
The log classification was considered valid if at least one of the boards was
graded the same as the predicted log grade. The agreement between the
manual lumber grading and the simulated X-ray log scanner was between 71
and 89%. It was concluded that the simulation procedure and grading algo-
rithm was a good predictor for the grades of logs and lumber.
A large scale validation test of computed tomography software against
sawmill cutting was reported by Chiorescu and Gronlund (2000). This test was
based on the Swedish Stem Bank X-ray computed tomography database (Gron-
lund et al. 1994) that acquired information on 600 selected Scots pine saw logs.
The schematic path representation of the validation approach is presented in
Fig. 2.29. The validation approach has two branches, the first corresponding to
the simulated part of the study and the second to an actual medium-sized mill
located in northern Sweden. The conversion process from the sawmill and the
simulation is shown in Table 2.l3. The volume recovery (48.08%) of the sawmill
was very close to the simulation recovery (48.34%). A good prediction of the
quality and length of boards was noted.
Schmoldt (1996) and Schmoldt et al. (1996a) and Schmoldt et al. (1997)
developed specific X-ray computed tomography algorithms for hardwood logs.
Specific scanning parameters are required for X-ray imaging of logs of differ-
real
. ....
i
sim ulated
. . .-. ....
, I
Fig. 2.28. Real and simulated image of a log. (Grundberg and Gronlund 1997, with permission)
Table 2.13. The ability of the simulation software to predict board quality distribution.
(Chiorescu and Gronlund 2000, with permission)
Board quality Class

Quality 1 (%) Quality 2 (%) Quality 3 (%)
Sawmill 27.00 38.80. 34.20

Simulation 28.15 39.74 32.11
Prediction + 1.15 (overestimation) +0.094 (overestimation) -2.09 (underestimation)
Applications 59
raw material - A raw material - B
swedish stem bank sawmill supply
virtual sawmill sawmill
~ ~
[ common sawing process attributes 1
sawn timber - A sawn timber - B
comparison of the produced boards
Fig. 2.29. Validation approach of the computed tomography software against sawmill cutting.
(Chiorescu and Gronlund 2000, courtesy of Forest Products J)
ent hardwood species of different sizes, and with different image resolution
and contrast. Data visualization and condensation must form a clear view of
the geometry of the log and of its internal features. Only log defects that are
important for subsequent processing need to be identified. Statistical labeling
of accuracy was introduced, and segmentation and labeling into a single clas-
sification step were possible, largely by using local pixel neighborhoods. A
neural network based on statistical classification of accuracy by using nor-
malized pixel values as inputs to the classifier can formulate and apply aggre-
gate features (average and standard deviation), as well as texture-related
features. The recognition of different hardwood species was based on a
histogram-based preprocessing step that normalized the CT numbers for
density values prior to artificial neural network classification. Morphological
postprocessing is used to refine the shapes of the detected image. The 3D iden-
tification of defects proposed by Schmolt et al. (1997) is a major step in the
development of machine vision for wood species of very complex structure
such as oak and yellow poplar. Initial results using a 2D classifier produced a
lower classification accuracy than 3D data and it was suggested that the 2D
classifier could be used successfully for hardwoods. Previous investigations

published by McMillin (1982), Taylor et al. (1984), Funt and Bryant (1987), Zhu
et al. (1991), Hagman and Grundberg (1995) were limited to 2D image analy-
sis for softwoods only and no statistical approach was used.
2.4.2.2 Inspection of Poles
The large numbers of poles for telecommunications network in service in

Australia, Japan, Canada, the United States (Miller 1988), and Europe (Rinn
1991; Habermehl and Ridder 1992a,b, 1993, 1994, 1995, 1996, 1998), explain the
numerous projects that have been established to build and use field portable
X-ray computed tomography scanners for the detection of the internal condi-
tion of wooden power poles. The need for the development of a nondestruc-
tive inspection method was supported by the high replacement costs. Major
causes of internal deterioration of poles in the vicinity of the ground line are
decay fungi and termite infestation.
The prototype device built in Australia (Cavies et al. 1990) satisfied several
specifications: quantification of the internal structure of poles having the
maximum diameter of 450 mm, production of tomograms from 300 mm below
the ground line to 1000 mm above the ground line, relatively light (less than
50 kg), a spatial resolution of 5 mm or better ±50 kg/m 3 accuracy for the density
of negligible radiation hazard while in operation, and production of a cross
section scan within 5 min, an image of the cross section in 5min, and an image
that is easily interpreted by the field operator.
The instrument operates with a translate-rotate (first-generation) mode
for scanning poles of 450mm diameter. The detector is an X-ray tube that
is powered from a portable motor generator that can provide 160keV at
18mA. For each projection with 100 projections for 180 0 of rotation at a slice
thickness of 5 mm, 99 ray sums are collected. The image is reconstructed from
99 x 99 pixels. The complete data acquisition phase is completed in 3 minutes,
and the image is displayed as a 2D density map, with details on growth ring
pattern and density histograms for immediate assessment by the operator.
The spatial resolution is <5 mm. Data are stored on a floppy disk for further
analysis if necessary. The total weight of the entire unit is 53 kg including the
8-kg X-ray tube.
2.4.2.3 Inspection of Lumber
Pioneering work on X-ray computed tomography inspection of the internal

structure of lumber was done at the Forest Products Laboratory in Madison,
USA (Suryoatmono et al. 1994). The main objective was to determine the
density variation in a piece of lumber and to establish relationships between
its structural performance and its mesostructure characterized by the pres-
Applications 61
ence of macroscopic defects like knots. The final goal of tomograms is to

provide information for the optimization of the decisions in edging, trimming,
and grading during the production of lumber. The characteristics of the scan-
ning device were as follows:
- X-ray source of 50 keV, positioned 1810mm from the image plate coated with
photostimulable phosphor crystals, which act as energy traps when exposed
to ionizing radiation;
- time of exposure 0.5 s at 200 rnA;
- data readings with Philips Computed Radiography System to a grid of 0.9 X
0.9 mm. The readings are machine dependent numbers, ranging from 1 for
the highest radiation exposure to 1024 for the lowest radiation exposure.
- calibration of the system with two small clear specimens of southern pine
at 12% moisture content;
- constant mass attenuation factor assumed to be 0.2506m 3/kg;
- southern pine lumber samples size: 520 X 38 X 140 mm. with a single wide-
face knot ranging from 30 to 45 mm in diameter;
- high-density zones defined as having a density 20% greater than the speci-
men average.
The superposition of the real image on the graph of density variation along
the length of the specimen shows the transition zone of increasing density
within one knot diameter of the knot center point (Fig. 2.30). The knot density
is twice the average density of the board. It was concluded that the transition
zone density compensates for the mechanical weakening influence of the knot.
Successful lumber use for construction purposes is based on procedures to
obtain strength grades. For sawmills to choose the proper raw material and
predict the precise mechanical properties (modulus of rupture and modulus
of elasticity in bending) of each board is an important step forward in the pro-
duction of strength-graded lumber with special dimensions. Kliger et al. (1995)
demonstrated that the density and the knot area ratio can be used to explain
the variations in strength and stiffness of Norway spruce beams. Grundberg
and Gronlund (1997) showed that the knot volume of logs can be measured
with an X-ray log scanner. Wang et al. (1990) demonstrated a strong correla-
tion between the X-ray tomographic green density of heartwood and the
density of dry wood. Consequently, the hypothesis that relates density and knot
volume of logs to strength and stiffness of the corresponding lumber was
demonstrated by Oja et al (2000). For this purpose, eight Norway spruce (Picea
abies) logs were scanned. The tomograms were used for the simulation of knot
volume and green density of heartwood for a zone corresponding to four cen-
terboards. After scanning the logs, four centerboards were sawn from each log,
using a cant sawing pattern (45 X 145mm). The boards were proofloaded and
the bending strength and stiffness were determined using the Swedish norms,
SS-EN 408. Statistical treatment of experimental data (multivariate models
and partial least squares regressions) demonstrated reasonable correlations
(R2 ranging from 0.42 to 0.94) between knot volume and green density of
......-
longitudinal direction
sample 5
1200
1000
a::i 800
u
bb 600
C
.~
'"v~ 400
"0
200
0
0 50 100 150 200 250 300 350
a) longitudinal position (mm)
t
longitudinal direction
line 4
line 3
line 2
line I
sample 5
1200
1000
a::i 800
~ 600
C
.0
.;;; _._. line I
v~ 400
"0 -0 - line 2
200 ..... line 3

-line 4
0
0 20 40 60 80 100 120 140
b) transverse position (mm)
Applications 63
heartwood deduced from tomographic inspection of logs, and the stiffness and
strength measured on boards. It was concluded that the mechanical properties
of each board are partly related to the sawing position of the corresponding
board. From the information about the sawing position, it is therefore possible
to predict the strength and stiffness of each board.
In the near future, this important advancement in lumber inspection needs
to be confirmed with more studies with a larger number of specimens of the
same species as well as with other softwood and hardwood species.
2.4.3 Wood Technology and Wood Processing
Today, the application of X-ray computed tomography in the wood industry is

very limited because of the high costs of the device. Medical X-ray scanners
have been used for laboratory tests performed for drying control of lumber
and for the detection of voids in strand-based composites.
Specific scanners using gamma rays have also been designed for the mea-
surement of the densitometric profile of wood composites. A continuous data
profile of the board is provided by the device in a very short time, ranging from
2 to 5 min for a complete scanning of the board, depending of the sample thick-
ness. The data are displayed in 2D, with the density as a function of the spec-
imen thickness. No image of the internal structure of wood based composites
was required. The main advantage of this scanner is the relative compactness
of sources and detectors, without excessive shielding of the radiation source.
2.4.3.1 Control of Lumber Drying
Computed tomography during drying is an excellent nondestructive technique

to study the moisture flux in lumber as a function of drying time. However,
some difficulties related to the interpretation of tomograms have been
identified:
- accuracy of local moisture content determination because of shrinkage;
- determination of the edge position and of the local moisture content at the
wood surface with the conversion algorithm having an important density
step at the transition between air and wood.
To elucidate these aspects, theoretical modeling and experimental
approaches were proposed by Kanagawa and Hattori (1985), Kanagawa et al.
Fig.2.30. Density variation observed during length scanning of a southern pine board of 350 mm
length and 38 x 140mm section, containing a knot (Suryatmono et al. 1993). a Longitudinal
position; b transverse position. (Courtesy of Forest Products Society)
(1992), Pang et al. (1994), Pang (1996), Wiberg (1996), Arfvidsson et al. (1997),
Pang and Wiberg (1998), and Wiberg and Moren (1999).
Kanagawa and Hattori (1985) used a medical scanner for the study of the
distribution of moisture content during conventional kiln drying of 10.5 cm2
lumber of hemlock (Tsuga heterophylla) and of almaciga (Agatis sp.) from
green to 12% moisture content. The moisture distribution on sample sections
was estimated following the procedure specified by Hattori and Kanagawa
(1985) from the CT number of wet wood, oven-dry wood density, and volu-
metric shrinkage. A standard error of 1.1 % was measured for the reference
specimen with moisture content ranging between 2 and 27%. The errors were
caused by the lack of uniformity of the images and lack of perfect linearity
between the oven dry wood density, shrinkage and CT numbers.
The distribution of moisture on radial and tangential sections for both
species are given in Fig. 2.31. In green Agatis sp. (50-70% moisture content) a
high moisture content was observed at the limit of the annual ring and in ray
tissue in the center of the sample. The moisture gradient in the tangential direc-
tion was more important than in the radial direction. In green hemlock
>R
~ 100 100
1::
Q)
1::
0
<.)
a
Q)
en
.~
50 50
o
edge-grain center edge-grain flat-grain center flat-grain
surface position surface surface position surface
a) tangential direction b) radial direction

Fig.2.31. Moisture content distribution during drying of a sample of 10.5 x 10.5cm section of
Agatis spp. (Kanagawa and Hattori 1985). The average moisture content is noted as: 1 123%; 2
98%; 3 83%; 4 73%; 5 56%; 642%; 728%. (Courtesy of the Japan Wood Research Society)
Applications 65
(50-160% moisture content), moisture distribution in the tangential direction

was relatively constant around the central position of the sample, while in
radial direction a belt-shaped distribution was observed.
Pang (1996) and Pang and Wiberg (1998) studied the moisture distribution
in Pinus radiata boards of 100 x 40 x 1200mm during 30h of drying (dry-
bulb/wet-bulb temperatures were 90/62°C). The moisture distribution and
migration (Fig. 2.32) are shown in the wet board before drying (Fig. 2.32a), at
an early stage of drying (after 9.6h; Fig. 2.32b) and at a late stage of drying (i.e.
after 30.4h; Fig. 2.32c).A high moisture content is illustrated by the dark color.
As can be seen from the tomograms, a dry shell is formed between the dry
board surface and the wet front of wood. The thickness of the shell as predicted
by Pang's model is between 0.5 and 2 mm. Pang's model simulated the mass
transfer (moisture movement) and heat transfer following the width and the
thickness of the board and predicted the moisture distribution, temperature
profile and drying rate curves. The model proposed by Pang and Wiberg (1998)
predicted density distribution (Fig. 2. 32d-f) and moisture content distribution
(Fig. 2.32g-i). This model is more sophisticated than Pang's (1996) model,
including inputs of wood basic density, within-ring variations of wood density,
variation of moisture content, permeability of wood and permeability of late-
wood and earlywood, dimensions of lumber, drying conditions - temperature,
humidity and velocity of the air - and actual sawing pattern.
In the early stage of drying (9.6h), the moisture content near the surface
decreases to below the fiber saturation point, inducing shrinkage of the board
and corresponding drying stresses. However, the core of the board is still
wet and drying stress can produce checks on the board surface. In the late stage
of drying (30.4h; Fig. 2.32c), when the average moisture content was 7.4%,
the tomogram shows a high wet wood density in two latewood bands and
in the core region of the board. The model predicted (Fig. 2.32f) a relatively
high moisture content for the core zone, namely, 12 and 3.4% moisture content
for the outer area. The model was not able to predict the moisture content in
the two bands of latewood, probably due to the very high values of density
in this zone, which was not included in the model. This is only of minor impor-
tance, and the validity of the model was confirmed by the drying curve
expressed by the average moisture content of the board as a function of drying
time (Fig. 2.33) which shows a very good agreement between experimental
and predicted values. A small deviation of 6% was observed between 11 and
16h of drying, corresponding to the decrease in moisture content from 60 to
about 30%.
In view of these results, the modeling approach proposed by Pang and
Wiberg (1998) is a major contribution in the field of detection of moisture
distribution and validation of drying regimes using computed tomography.
The causes of drying defects such as kiln brown stain and surface checking
could be explained. At the same time, the model can be used to determine the
influence of sawing pattern, wood density and permeability in earlywood and
latewood, etc on the drying rate of lumber.
a b C
36-
36 E 36 E ·85C>950 E .540-590
75<>850
30 _E •.650-150 30 S .490-~
30S "~lIoo
24* :~~
~a_
24 ~ :.~::: 24~ ~= ~ a J40.390
18~ 18 18 5
s:
12; 12-2 12"E
6 6
.
III
6 i S
0
0 10 20 30 40 50 60 70 60 90 100 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
d Board wId1h (mm) e Board width (mm) r Board widlh (mm)
36-
E _1()O.l20
36 e :~r:~: 30 E • 110-'00 • 15-18
30.§. .'00.130 - . 60-00 .12-15
• 9-12
24 :ec 24 ~ ~ ~ :::: II 6 -9
~ a 0-20 a 3·6
18~ _18£
12"E = 12l?
6 S 6
.
III
0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70
it Board widlh (mm) h Board wId1h (mm) Board wid1h (mm)
Fig.2.32a-i. Dynamics of wood drying (Pang and Wiberg 1998). Experimental tomograms (with permission). a Scanning of the board before drying - corresponding to
wet saturated wood. b Scanning of the board before drying - corresponding to wet wood after 9.6 hours of drying, and to the early stage of the drying process. c Scan-
ning of the board before drying - corresponding to wet wood after 30.4 h of drying and to the late stage of the drying process. Predicted images for density distribution.
d Wet wood density before drying. e Wood density after 9.6h of drying. fWood density after 30.4h of drying. Predicted images for moisture content distribution. g Model
of predicted wet wood density before drying. h Model of predicted wet wood density after 9.6 h of drying. i Model of predicted wood density after 30.4 h of drying
Applications 67
,\.
140
:a...
~ 120 '\.
"-
~ 100 -- -
"
model predicted MC
~- -
-
•
~
.
80 "I> CT scanned MC
-
~
.~
60 •
" ...........
'" 40
~
.......
~
~ 20
--
1- ........ ~
oT
o 5 10 15 20 25 30 35
drying time ( h )
Fig. 2.33. Comparison between moisture content predicted by the model and the experimental
data deduced from the tomograms of Pinus radiata board, expressed by the drying curve. (Pang
and Wiberg 1998, with permission)
2.4.3.2 Control of Wood-Based Composites
The data concerning the utilization of X-ray computed tomography for the
characterization of wood-based composites are very sparse. Sugimori and Lam
(l999) reported the utilization of this technique for the study of macrovoid
distribution in strand-based-wood composites of 630kg/m3 density. The
macrovoids in strand-based composites, of various sizes and random distrib-
ution, are inherent to the forming process of boards. Their presence is due to
the imperfect bonding between the adhesive and the wood strands.
The X-ray computed tomograms were obtained with a medical scanner, set
at 120 k V, 200 rnA, 10-mm slice thickness, and 0.86-mm pixel size and 0.74-mm2
area, in a 512 x 512 pixel matrix. The image reconstruction was done every 3
s, and 130 slices were produced. The size of macrovoids was considered to be
::;;2 pixels, i.e., 1.48 mm2 • The images obtained were converted to black and white
with a conventional image-processing technique, and their size, number, posi-
tion, and major and minor diameters were measured with the medical soft-
ware. The variation along the length of a board of macrovoid area as well as
their distribution is shown in Fig. 2.34. The authors noted that the experimental
size measurements are in agreement with their mathematical prediction.
Gamma ray scanning densitometry (Feraz 1976; Ranta and May 1978) has
been more frequently used for the measurement of the densitometric profile
of wood-based composites. Winstorfer et al. (l986) measured the densitomet-
ric profile of particleboard (l2, 18 and 44mm thickness), waferboard (l3 mm
thickness), fiberboard (6 and 12mm thickness) and medium density fiber-
board {l0, 19mm thickness). A gamma source of 60keV was mounted in a col-
limating holder, and the sample was placed on a carriage that held it stationary
x = 376 - ,... X = 367

185
high densily~Ii;'.l
lowden i~
X"----
400 X direction ( pixel)
o
a)
'(
X=376 X=367
~JI ~
r-~~--------------------------------------~, 1~
.
-
"
t. ..
.. -;.. ..
-- .. .
., -
J
."-." ,.
.....
X ~_~~-L~--~·--LL'~·--~--~--------------"-·---~-·----~o
400 X direction ( pixel )
b)
Fig. 2.34. Macrovoid distribution in particleboard of 630kg/m3 (Sugimori and Lam 1999).
a X-ray image; b the same image after conversion with a 270-threshold value. (Courtesy of the
Japan Wood Research Society)
during counting and then automatically moved to the next step. At each step,
which can vary from 15 Jlm to 30 cm, the counter is automatically read, reset
and restarted. Counts at each step were taken every lOs. The average counting
time for a complete scanning of a board depends on the thickness, and is
between 2 and 5 min. The authors suggested using this as a standard method
for nondestructive evaluation with the accuracy and precision of density pro-
files for wood panel products.
Laufenberg (l986) also studied the use of gamma radiation to measure
the density gradient in reconstituted wood products. The mass attenuation
coefficients were determined for different resins: phenol-formaldehyde
(0.181 cm2/g), urea-formaldehyde (0.182 cm2/g) and isocyanate (0.178 cm2/g) as
well as water (0.196cm 2/g) and wood-red oak and Douglas fir (for both
Applications 69
0.183cm2/g). He noted that the large differences between the mass attenuation
coefficient of resins, wood, and water can induce errors in the prediction of
the wood density of the board. For this reason, the boards must be well con-
ditioned at a minimum 12% moisture content prior to density measurement.
In this case, the expected error of the density prediction is <1 %.
Elbez (1990) reported the utilization of computed tomography with a
medical scanner for the control and characterization of cured adhesives and
glue joints in balsa, spruce, poplar, oak and two tropical species, moabi and ipe.
The adhesives were urea-formaldehyde plasticized with furfurylic alcohol,
polyvinyl acetate with metallic salt and resorcinol phenolic formaldehyde. The
tomograms clearly showed structural details of the glue joints and of the adja-
cent zone composed of adhesive and wood. The density gradient was observed
with a maximum in the center of the joint and with a minimum in the prox-
imity of the wood. The minimum density is probably due to the lower degree
of condensation in urea formaldehyde glue or to the lower polymerization in
the case of the polyvinyl glue. Lindberg et al. (1996), using X-ray CT, measured
water diffusion through acrylate latex paint films with 0.1- and O.4-llm parti-
cle sizes. It was proven that the techniques employed were successful for the
measurement of water diffusion in films. As expected, a O.I-llm particle film
had a much lower water diffusion coefficient than an O.4-llm film.
2.4.4 Dendroarcheology, Dendrochronology, and Dendroclimatology
The physical and mechanical parameters of archeological buried wood, satu-

rated with water, were studied by Schniewind (1990). Compared with fresh
wood of the same species, the mechanical properties of archeological wood are
diminished because of abiotic deterioration.
Passialis (1997) noted that changes in chemical composition are due to a
slow hydrolysis of polysaccharides, with a notable increase in lignin content.
Morphological deterioration of cellular tissue and separation of cells in the
growth rings are found. Density is also reduced because of the losses of chem-
ical constituents and of deterioration of cell integrity. Kim and Singh (2000)
emphasized that all species can survive in a fully saturated environment for
thousands of years. The slow degradation of polysaccharides, lignin and
extractives can be caused by biological agents such as fungi and bacteria that
are tolerant to oxygen limiting conditions. Wood deterioration can be caused
by both microbial and chemical agents. Unger et al. (1988) reported the use of
X-ray computed tomography for the inspection of archeological oak wood
with different amounts of water, ranging between less from 185 to more than
400% moisture contents. The images clearly show the advanced degradation
zones.
Klein and Vogel (1993) noted the interest of using X-ray CT for the identi-
fication of wood species in dendrochronology. Madsen (1994) and Divos et al.
(1994) employed a gamma backscattering technique for inspection of ceilings

and roof structures of old buildings. Hoag and McKimmy (1988) proposed
direct scanning densitometry for dendroclimatological studies. A significant
advancement was observed in nondestructive inspection of fine art objects,
historic musical instruments, etc., using computed tomography. An example is
given by Fioravanti and Ricci (1991) for the study of the densitometric profiles
of a top plate of a violin. Onoe et al. (1984), and Schirone and Lo Monaco (1988)
also suggested the use of computed tomography in archeology and den-
drochronology. Unfortunately, computed tomography has not been extensively
used in dendrochronology, dendroarcheology, or dendroclimatology.
2.5 Summary
Ionizing radiation (X-ray or gamma ray) computed tomography in wood

science is a cornerstone in the field of the imaging of solid wood and wood-
based composites. High resolution X-ray imaging is a valuable tool for nonde-
structive evaluation and inspection of trees, logs, poles, boards, etc. and for
nondestructive control in wood processing. The ability of this method to iden-
tify and quantify the internal inhomogeneities and defects in wood products,
the distribution of moisture inside the trunk, etc. explains its importance and
increasing implementation in wood science and technology.
The aim of computed tomography is to create a precise map of the internal
inhomogeneities of a sample that can be a trunk, piece of lumber, laboratory
testing specimen, etc. This map is obtained through the determination of the
attenuation coefficient of ionizing radiation in a single thin layer of the sample.
The images produced are reconstructed from projections using advanced com-
putational and mathematical procedures underlying the data collection, image
reconstruction and image display. The spatial information in 2D or in 3D is
able to discern zones of low attenuation contrast. First generation scanners
produced tomograms with one source and one detector acting in parallel by
translation. The sample was rotated by a 10 -step angle and the whole image
obtained for 180 0 of collected data. Second-generation images were also
obtained by translation, using an array of detectors that made simultaneous
measurements through different angles in a single transverse inspection. The
sample was rotated by the array beam angle. The third generation of tomo-
grams were produced by a fanning movement, with a scanner having many
detectors located on an arc focused on the radiation source. The fourth scanner
generation has also a fan system of detection, and the detection array is located
on a circle that surrounds the source and the sample.
In inhomogeneous materials like wood, the attenuation coefficient depends
on both the quantum energy of the ionizing radiation and on the chemical
composition of the sample. Because of the high quantum energy of radiation
and the low atomic number of the chemical constituents of wood, the attenu-
Summary 71
ation phenomenon is caused mainly by the Compton effect and, consequently,

is proportional to the mass density of wood. Each zone of the specimen having
the same density is evaluated statistically and is represented by one color, or
by a gray scale.
Worldwide, during the last 20 years, stress has been put on understanding
of the basic aspects of ionizing computed tomography in view of the develop-
ment of specific industrial scanners for routine quality control of wood prod-
ucts. The main advantages of the ionizing radiation computed tomography
compared with conventional radiography are:
- elimination of the intermediate steps involving photographic film and
optical densitometry;
- data available in real time;
- improvement of the calibration procedure, since the mass attenuation coef-
ficient can be determined directly, using a scintillation detector with stan-
dard radiation pulse shaping and counting equipment that allows the user
to select the energy range counter;
- large-volume material can be inspected quickly and implementation of
scanning technology in sawmills and other mills for wood will have an
important payback for wood processors.
The scanners for ionizing radiation computed tomography can be fixed or
portable. Medical scanners were used in the 1980s for imaging wood speci-
mens, because the physical properties of wood do not differ greatly from that
of the human body. The development of non-medical scanners, initially exper-
imental and later commercial, occurred in the mid-1970s. The industrial
systems in wood industry require: high energy X-ray sources, accommodation
for a wide range of specimen sizes and densities and operations in industrial
environments with a speed consistent with efficient wood processing.
Fixed and portable equipment are used in wood science and technology for
different purposes. If the object can be brought to the scanner as in the case
of logs and boards, fixed equipment is recommended. In this case, an X-ray
source is commonly used. Portable equipment was designed to be used for in
situ inspection of trees, poles, and building elements. Gamma rays were used
in this case because of many advantages with respect to the wide range of ener-
gies available, portability and cost. In spite of these superior characteristics,
gamma ray sources are of low intensity compared with X-rays.
The factors affecting the quality of the image are: beam path, spatial and
contrast resolution, anisotropic direction of wood and beam hardening. X-rays
travel in a straight line through materials. Two main parameters are related to
the beam path: range of transmission probability and combined effects. The
range of transmission is defined as the radiation resolution of a given wood
X-ray experiment, in reference to the separability of transmission probabilities
resulting from density variations. The combined effects are a global factor
taking into consideration the combination of X-ray energy, wood specimen
thickness, and heterogeneities induced by the anatomical structure.
The maximum radiation resolution is determined by density variations in

wood specimens, energy and intensity of X-rays, spatial and contrast resolu-
tion and beam hardening. Because X-ray attenuation is highly dependent on
the energy of the X-rays, it is always necessary to provide this information.
The main applications of ionizing radiation computed tomography dis-
cussed in this chapter are:
- examination and inspection of trees, related to the growth rate assessment,
water content in trees and pollution effects;
- wood quality assessment, in the inspection of logs and poles;
- processing of wood for control of lumber drying and the quality of wood-
based composites;
- in dendrochronology and related fields, such as dendroarcheology and den-
droclimatology.
2.6 Annexes
Annex A. Measured and calculated CT numbers for wood of different

densities and 0% moisture content. (Lindgren 1991a)
Density (kg/m3) Measured CT number Calculated CT number

415 -603 -606
419 -601 -602
496 -534 -529
499 -529 -526
619 -414 -412
612 -419 -418
352 -670 -666
468 -554 -555
400 -621 -620
456 -570 -567
506 -518 -519
Annexe B. Density and CT number for different species.

(Klein and Vogel 1993)
Species Density (kg/m3) CT number (H)

Lophira alata 1030 62.6
Millettia laurentii 750 210.1
AJzelia spp. 700 178.9
Quercus robur 650 483.9
Annexes 73
Species Density (kg/m3) CT number (H)

Fraxinus excelsior 650 347.8
Pinus palustris 630 441.6
Salix spp. 520 666.6
Tilia grandifolia 490 510.4
Tilia spp. 490 640.3
Populus nigra 450 649.0
Pinus cembra 450 641.7
Picea sitchensis 430 616.4
Picea abies 430 607.6
Bombax spp. 400 874.2
3 Thermal Imaging
3.1 Introduction
Thermography is a generic term for a variety of techniques used to visualize

the temperature at the surface of objects and refers to imaging of a full-field
temperature distribution. Active heating or passive heating procedures can
be used for thermographic imaging of solids (Liezers et al. 1985; Puttick
1987; Vavilov 1992). Thermography provides a map of surface temperature
distribution. It can be said that thermography is a non contact and nonde-
structive method that gives information about the thermal properties of
a sample. The main thermographic methods for solids fall into three major
classes: pulse thermography (Lau et al. 1991; Thomas et al. 1995), heating
up thermography (Reynolds and Wells 1984), and lock-in thermography
(Busse et al. 1992b; Busse 2001). The main differences between the thermo-
graphic methods are determined by the type of heating and by evaluation of
the thermal properties of the samples. The pulse thermography system records
the transient thermal response of an object to a pulse of thermal excitation,
provided by a flash tube.
Heating-up thermography utilizes heat sources such as: commercial photo-
graphic lighting devices, which deliver energy up to 5 J/cm2 in IllS at high rep-
etition rate, hot liquids, and air dryers of the high power "heat gun" type. As
an alternative to heating, cooling can be used with an aerosol freezer spray.
Video cameras in visible light wavelengths record the images in a rapid corre-
sponding time. The transient character of the temperature field demands a
dynamic tuning of the thermo-vision system. Sometimes, the large local
nonuniformity of heating can induce noisy signals that influence the image
quality obtained throughout the whole scanned area. The development of
infrared video cameras has extended the wavelength range of visible light
video cameras to the thermal infrared range, between 3 and 121lm. Images
obtained with infrared cameras correspond to a mixture of emitted and
reflected radiation of the object and of its surroundings and this test is par-
ticularly severe in the range of 8-121lm wavelength, which corresponds to the
peak of blackbody radiation at room temperature. For this reason, a calibra-
tion test must always be performed. In conventional thermal wave-imaging
techniques, this problem is overcome by the use of a lock-in analyzer, syn-
chronized to the source of the thermal wave.
76 Thermal Imaging
Lock-in thermography can provide three types of images in a very short

time: thermographic, phase and magnitude images. The development of
lock-in vibro-thermography was an important step in thermal wave imaging
which allows the utilization of thermographic equipment to image the average
temperature distribution of oscillating components. The thermal waves are
generated by a periodical stress at very low frequency (0.03 Hz), which do
not induce overheating of the surface of the sample. Vibro-thermography
is a very useful method for inspection of solid wood and wood-based com-
posites because of the high emissivity and low thermal conductivity of these
materials.
In wood science, thermography is a relatively new field of study. Thermal
imaging of subsurface temperature distribution on wood-based composites to
ascertain the integrity of the subsurface structure was introduced less than 10
years ago. We note here the pioneering work by Busse (1988, 1994,2001) and
coworkers at Stuttgart University in Germany who developed these techniques
for polymers and wood-based composites.
Heat propagation phenomena in solids are influenced by the thermal
impedance of the materials, which varies as a function of the interaction of
thermal waves with the structure. Since these interactions are associated with
a modification of the thermal properties, it is possible to obtain a map of the
sample characteristics and consequently to obtain an image.
The active heating procedure, or stress-generated thermal field, was used in
early work on wood thermographic imaging (Masuda et al. 1995; Masuda and
Takahashi 1998). This makes use of the fact that under fatigue cycling loading
in bending or compression, the temperature rise is determined by the increase
in energy in the hysteresis loop. The material is cyclically stressed while its
surface temperature is monitored thermographically. Heat dissipation results
in a temperature distribution on the material surface that is related to the stress
distribution in the material. Loading of the specimen can be done either by
direct fatigue or by vibrational means.
The advantage of this active heat generation technique is that the influence
of defects on the mechanical properties of solids can be easily and rapidly
examined. The quantitative evaluation of a stress field can be obtained with a
highly sensitive infrared camera capable of detecting changes of 10-3 K in
regions of less than 1 mm. The main disadvantage of the cycling loading of the
material is related to the possibility of inducing new damage in the sample
when trying to locate existing defects. The use of a low-amplitude vibrational
excitation can be an alternative and can alleviate the damage propagation. With
pulse thermography the response of the sample is recorded after its warm-up
period with a short energy pulse (light, ultrasonic pulse) and an infrared image
is obtained. This image indicates the inhomogeneities and defects at the
surface of the sample or at different depths.
With lock-in vibrational thermography, which is a combination of thermal
wave technique and thermography, a large sample depth range can be
inspected, and no overheating occurs at the surface. The sinusoidal thermal
Introduction 77
wave is excited on the whole surface of the sample and then monitored, using
both the thermographic technique and lock-in data analysis which very rapidly
provide a phase angle image. This remote method was used for detection of
flaws such as delaminations in wood composites (Wu and Busse 1995, 1996;
Wu et al. 1996, 1997b) because of its sensitivity to subsurface defects and its
imaging capability in short time.
In the case of the so called passive method, the material is subjected to
heating from an external source. The thermal image obtained is the result of
a very complex interaction between the heating source, the material and the
presence or absence of defects. The rate of heat application and the mode
of heating by contact or by radiation are factors of major importance for
obtaining good images with correct resolution. The thermal gradient and
not the temperature itself makes it possible to obtain information about the
integrity of the structure. The thermal gradient is calculated from the temper-
ature distribution. The heat application may be at a single point or over a wide
area. This passive mode produces images that are transient and therefore
difficult to capture for a material that is not subjected to any mechanical stress.
The elimination of the loading fixture makes the rapid testing of objects
of various geometries in situ possible. The time interval of interest for passive
thermal testing is a few seconds. During this time the surface temperature
gradients reach the maximum and yield the greatest resolution of the thermal
image.
An advantage of the passive heating method over the active heating proce-
dure is its ability to produce a temperature distribution without resorting to
mechanical loading of the material. The thermal stress or the thermal shock
is too low to damage the material. The disadvantage of the passive method is
that the thermal images are transient and require a fast recording system to
capture the most interesting images during the test.
Electronic imaging systems or chemical imaging systems can be used for
the visualization of the temperature distribution on surfaces. The most widely
used system is an electronic system with a scanning infrared camera, which
produces an image of an object through electronic detection of infrared radi-
ation emitted from the object.
Chemical imaging is obtained with liquid crystals applied to the surface of
the object. They selectively scatter the light according to their temperature.
The thermal resolution is of 0.007 DC and the spatial resolution of 111m (Wilson
and Charles 1981). A photographic recording system is attached to capture
liquid-crystal images during the test.
Catena (1992) reported the use of an infrared passive technique for the
detection of cavities and for phytopathological diagnosis of ornamental trees.
The infrared camera scanning technique was also used for the detection
of different defects, such as delamination in wood composites (Berglind and
Dillenz 2000) or locating knots and voids in lumber (Tanaka 1994; Quin et al.
1998; Steele et al. 2000a,b; Tanaka and Divos 2000). In wood technology,
the scanning infrared imaging technique seems to be one of the most con-
78 Thermal Imaging
venient remote sensing methods available today for thermographic inspection

of trees, wood and wood-based composites in situ.
The aim of this chapter is to provide a general description of thermal
imaging techniques for solid wood and wood-based composites The methods
described here are divided into two main groups: the active and the passive
methods, for imaging defects in trees and in lumber, delaminations in wood-
based composites and in different types of joints for lumber connection.
3.2 Basic Aspects
Significant progress in the understanding of thermal properties of solids was

made with the understanding of the physics of diffusion waves. Diffusion
waves are neither electromagnetic nor elastic but thermal and their mathe-
matical formalism describes how a temperature modulation propagates in a
medium. In the 19th century, Fourier, Angstrom and Bell were the pioneers in
investigating thermal waves. Only 100 years later, the advance of laser tech-
nology made possible the coherent explanation of the optoacoustic or pho-
to acoustic effect (Rosencwaig 1983) induced by the propagation of thermal
waves.
Mandelis (2000), in an article devoted to diffusion waves and their uses,
noted that "diffusion waves lack wave front, can't be beamed and don't travel
very far, yet they form the basis of several new and revolutionary measurement
technologies." Diffusion waves have a complex wave vector and their transport
takes the form of a spatial diffusion gradient, which obeys a linear law. In the
imaging application of thermal waves, the map of the characteristics (magni-
tude or phase angle) of a locally produced signal shows the thermal parame-
ters of the structure under test. Busse (1988) noted that the image generated
with "locally observed phase shift is usually more significant" than that
obtained with the magnitude of the signal.
Thermal waves can be generated by any modulated energy. Imaging with
thermal waves is based on the dynamic effect of the modulation of the tem-
perature by the structure.
When a static effect is used, the image is obtained with a thermal infrared
emission that shows the static temperature distribution weighted by a local
infrared emission coefficient. In both cases, physical contact with the samples
is not necessary.
Heat flow in a solid is governed by Fourier Eq. (3.1)
k d 2 T/dx 2 = pC dT/dt (3.1)
where T is the temperature at depth x beneath the surface at time t, k is the
thermal conductivity, p is the density, and C is the specific heat capacity of
the solid.
Basic Aspects 79
If the angular frequency of the intensity-modulated laser heat source is rn,

the solution of Eq. (3.1) has the form
T(x,t) = Toe-x/llcos(21t x/A-rot) (3.2)
with thermal waves, the sample undergoes periodic heating when it absorbs
intensity-modulated laser radiation. If diffusion is periodical, produced by
absorption of radiation, the thermal diffusion length is defined as the distance
where the temperature decrease due to conduction is e- 1 times that at the origin
of the heating at the surface of the sample. The signal waveform phase lag
relative to the excitation waveform increases by 57 degrees.
The thermal diffusion length J..l is given by Eq. (3.3)
J..l = [2 k/ ro p C]1/2 (3.3)
where ro is the radial frequency at which the light is intensity modulated. In
metals J..l is about 1 mm. For other materials, the thermal diffusion length is
smaller. At the same frequency, the thermal wavelength 21tJ..l is much smaller
than the acoustic wavelength.
The phase velocity is the speed with which the temperature modulation
moves along the z direction of the specimen and is given by:
v = [2 rok/p Ct'2 (3.4)
It depends on frequency, thermal conductivity, density and the specific heat
of the sample. The dependence of the phase velocity on frequency undergoes
a strong dispersion. On the other hand, the wave velocity is not dependent on
the geometry of the sample in which the wave propagates.
The thermal wave length is given by:
A = 21tJ..l (3.5)
Thus, high frequency thermal waves propagate at high velocities but because
of their short diffusion lengths, they decay rapidly. In contrast, low frequency
thermal waves penetrate deeply into the subsurface with a low velocity.
It has been demonstrated that thermal wave imaging phenomena can be
explained with an extreme near-field diffraction theory, and standard scat-
tering techniques can be employed for the calculation of thermal wave imag-
ing (Rosencwaig 1983). Measurement of the thermal diffusion length and of
thermal conductivity can give information about the thickness of the sample
and can be used as a routine technique for depth profiling of a subsurface
structure. Most applications of relevance to quality control can be imagined
for depth profiling of layered wood-based composites or for measuring the
thickness of paint or varnish layers as demonstrated by Wu and Busse (1996).
Moreover, for thermal waves, the real and imaginary parts of wave numbers
are equal. This characteristic allows thermally driven waves to be observed at
any frequency with a very small wave vector magnitude, which permits iden-
tification of subsurface features in the submillimeter range.
80 Thermal Imaging
The spatial damped nature of thermal waves has two major advantages over
other tomographic techniques: the high near-surface feature resolution and at
the same time less susceptibility than other waves to the appearance of spuri-
ous signals from a remote interface. For these reasons, tomographic techniques
were developed for clinical applications and for nondestructive evaluation of
critical and valuable manufacturing components.
A laser beam that has a high spectral selectivity, spatial coherence and res-
olution is usually the light source used for generation of thermal waves. The
choice of laser characteristics (wavelength and power) depends on the thermal
properties of the sample. He-Ne laser and Ar+ laser were used for a power level
of about 20 W. They have an acceptable ratio of signal to noise and cause no
optical damage to the sample (Busse 1988). Tsuchikawa et al. (2000) proposed
the utilization of a semiconductor near-infrared laser of 830 nm wavelength
and 48 m W output power for the study of the effect of physical characteristics
of a specimen (sample thickness, fiber direction, surface roughness, moisture
content) and the geometric conditions of illumination on the transmitted
output power.
The relationship between the sample thickness and transmitted output
power for Sitka spruce and beech specimens (Fig. 3.1) shows an exponential
reduction of power for both species. This graph argued in favor of the devel-
opment of a nondestructive technique for measurement of veneer thickness.
The thickness of the sample can be controlled by varying the modulation fre-
quency of the incident radiation and hence the thermal wavelength.
The combined effect of light irradiation and of the anisotropic directions of
wood is shown in Fig. 3.2. It can be noted that the variation of output power
is strictly related to the geometric conditions of the experiment. The effect of
800
incident angle : () =0 0
:i' 700 diving angle : a =90 0
3
~ 600 distance between sample
and detector: dsd =2mm
""
<l)
~ 500
0.
"'[400
g 300
"0
<l)
.§ 200
g'" 100
oI 2
a) b)
Fig.3.1a,b. Transmitted output power under air-dried conditions as a function of sample thick-
ness (Tsuchikawa et al. 2000). a Relationships for beech and Sitka spruce. b Specimen geometry
and parameters. (Courtesy of Forest Products J)
incident angle 0 (deg.) incident angle e (deg.)
incident angle e (deg.) incident angle e (deg.)

• :n=O
O:n = 1 species: Sitka spruce sample thickness: d=2.3mm
... :n =2 distance between sample and detector: dsd=9mm
o :n=3 number of grinding by the abrasive paper #400
• :n=4
Fig.3.2a-d. The combined effects of light irradiation, anisotropic directions and surface rough-
ness of Sitka spruce specimens on transmitted output power (Tsuchikawa et al. 2000). a Diving
angle a = 0; light irradiation in LR plane and propagation in T direction. b Diving angle a = 45°;
light irradiation in LR plane and propagation at 45° in T direction. c Diving angle a = 0; light
irradiation in LR plane and propagation in R direction. d Diving angle a = 45°; light irradiation
in LR plane and propagation at 45° in R direction. (Courtesy of Forest Products J)
moisture content was related to the reduced transmitted output power, which
corresponds to the distance between the sample and the detector of 1 mm for
Sitka spruce and beech specimens of different thicknesses. For both species
and for all thicknesses, the reduced transmitted output power increases with
82 Thermal Imaging
increasing moisture content. This means that the output power increases with
the presence of free water in cell lumina and influences the conditions of light
reflection and scattering.
The second part of this section is devoted to an analysis of some particular
aspects related to thermal conductivity of wood. Because of its anisotropic and
porous nature, the thermal behavior of wood must be related to the parallel
and perpendicular direction of the fibers and to the hygroscopic level, below
and below fiber saturation point. The relationships between the thermal con-
ductivity of wood and some physical parameters such as the density or spe-
cific gravity and the porosity were studied by Siau (1995). The model proposed
by Siau (1995) shown in Fig. 3.3 was used for the derivation of the thermal con-
ductivity equations, for which three elements were considered: the thermal
conductivity of the cell wall substance, of bound water, and of air.
Below the fiber saturation point, when the heat flux is applied in the direc-
tion of the fiber axis, the thermal conductivity is given in Eq. (3.6):
(3.6)
where kL is the thermal conductivity in the longitudinal direction, k Lw is the
thermal conductivity of the cell wall in the longitudinal direction, Va is the
porosity of the wood, ka is the thermal conductivity of the dead air.
When the heat flux is across the cell walls, the transverse thermal con-
ductivity k T can be deduced in a very simplified form by Eq. (3.6) for which
the thermal conductivity of cell-wall substance was 0.318 and 0.401 W/mK for
bound water and 0.024 W/mK for dead air in the cell lumina.
ky = 0.50 - 0.46a (3.7)
where a x a is the lumen size that corresponds to the porosity Va = a x a
a) geometric model of a wood cell b) relative proportion of cell-wall substance,

bound water and air in wood
-
cell-wall
flux substance
oun water
air
Fig.3.3a,b. Simplified model of a cell wall and relative proportions of cell-wall substances, air
and bound water in wood (Siau 1995). a Geometric model of wood cell. b Relative proportion of
cell-wall substance, bound water and air in wood. (Courtesy of Wood Fiber Science)
Equipment for Thermal Imaging 83
crosswals Fig.3.4. The flux of thermal energy in

sidewall lumen and in cross walls and in side
walls (Siau 1995; courtesy of Wood
Fiber Science)
flux ~ lumen
sidewall
The relative flux concentration through the cross walls and side walls is
given in Fig. 3.4. It should be noted that because of the non uniformity of the
flux, the entire width of the wall is not available for conduction.
Above the fiber saturation point, the empirical equation for the determina-
tion of kT as a function of the specific gravity of the wood (G) and of the
moisture content (MC) is given in Eq. (3.8)
kT = G(0.27 + 0.0055 MC) + 0.024 Va (3.8)
The relationship between the longitudinal and transverse thermal conduc-
tivity is given by Eq. (3.9):
kL =(2.25 ... 2.75)kT (3.9)
Thermal properties of wood and wood-based composites have been
reported in several reference books and articles (Weatherwax and Stamm 1947;
Kollmann and Cote 1968; Giordano 1971; USDA Forest Prod. Laboratory 1972;
Siau 1995; Olek et al. 2000) Some values of thermal parameters are given in
Tables 3.1 and 3.2.
3.3 Equipment for Thermal Imaging
In this section, we describe the imaging equipment used for thermal imaging
of solid wood and of wood-based composites. The most popular device for
thermal imaging is the infrared scanning camera, which is a thermal wave
detector in the infrared domain and which can be used for both field and
laboratory measurements. The popular infrared camera can be used for all
thermal wave-type imaging. The thermal wave describes how a temperature
84 Thermal Imaging
Table3.1. Some values of thermal conductivity of wood and other materials in the SI interna-
tional system (Siau 1995).(courtesy of Wood Fiber Science)
No. Material Thermal conductivity

(W/mK)
Wood, 450kg/m 3; 12% MC, perpendicular to fibers 0.l3

2 Wood, 450kg/m 3; 12% MC, parallel to fibers 0.31
3 Wood, 700kg/m 3; 12% MC, perpendicular to fibers 0.18
4 Wood, 700kg/m3; 12% MC, parallel to fibers 0.44
5 Cell wall substance perpendicular 0.44
6 Cell wall substance parallel 0.88
7 Water, free 0.59
8 Air, dead 0.024
9 Douglas fir plywood 0.12
10 Hardboard, medium density 0.10
11 Particleboard, medium density 0.l3
12 Cement mortar, common brick 0.72
l3 Concrete 0.93
14 Stone, lime, sand 1.80
15 Glass 1.05 1.05
16 Fiberglass 0.039
17 Silver 419
18 Cooper 396
19 Aluminum 202
20 Stainless steel 16.3
Table 3.2. Thermal diffusivity of wood as a function of moisture content and density. (Data from
Kollmann and Cote 1968)
Moisture Diffusivity (10- 7 m 2 /s) for different wood density

content (%)
Density Density Density Density
(200kg/m 3) (400kg/m3) (600kg/m3) (800kg/m3)
10 1.89 1.56 1.42 1.36

20 1.75 1.47 1.36 1.33
30 1.64 1.39 1.31 1.28
50 1.50 1.31 1.22 1.19
100 1.44 1.14 1.06
modulation propagates if a sequence of thermographic images is recorded

during one or more modulation cycles of excitation. If a Fourier transform of
the time-dependent signal is performed at each pixel, the whole sequence is
compressed into an amplitude image and a phase angle image. The latter is
more relevant since it has a larger depth range and is less dependent on surface
topography and optical surface features (Busse 1979, 2001; Rosencwaig and
Busse 1980; Lehto et al. 1981; Bennet and Patty 1982). This kind of combining
thermography with imaging based on periodical excitation and thermal wave
analysis is named "lock-in thermography" (Carlomagno and Berardi 1976;

Beaudoin et al. 1985; Kuo et al. 1987; Busse et al. 1992b; Busse 2001). In previ-
ous sections, we have seen that the discontinuities in objects scanned with
infrared equipment can be observed on thermographic images since a tissue
discontinuity produces a difference in thermal conductivity that yields a non-
homogeneous distribution of surface temperatures.
The video infrared camera converts the electromagnetic thermal energy
radiated from the object under test into electronic video signals. Furthermore,
the electronic signals are amplified and transmitted to a monitor that displays
the image on a screen. The presence of internal cavities or damaged zones
induces a surface temperature different from that of sound wood.
As noted by Catena and Catena (2000), the infrared scanning camera used
for the detection of cavities in trees (Fig. 3.5) is a light, mobile piece of equip-
ment (3 kg). Only objects larger than the instantaneous field of view can be
Fig. 3.5. Infrared device for the detection of cavities in trees. (Catena and Catena 2000, with
permission)
86 Thermal Imaging
seen by the infrared camera (Fig. 3.6). The equipment used by Catena and
Catena (2000) for imaging of pathological attacks and internal cavities in trees
is composed of the following subsystems:
- a set of infrared lenses that focus the electromagnetic energy radiated from
the object.
- matrix of 320 x 240 microbolometers pick up the temperature at the tree
surface
- an electronic device that displays the image on the screen
For the inspection of trees, it is possible to use any infrared camera with
high thermal sensitivity and geometric resolution. The images obtained by
Catena and Catena (2000) had a thermal sensitivity of 0.1 °C at ambient tem-
perature and a geometric resolution of 1.3mm at 1-m distance from the tree.
The camera is mounted on a tripod and can be stationed at a variable distance
of up to 20 m from the tree. The average time for the inspection of one tree is
about 2 min. The equipment has the advantage of being non contact and very
easy to use without the need for additional personnel. Table 3.3 gives the char-
acteristics of the lens for the instantaneous field of view of the infrared camera
used by Catena et al. (1990.) for the detection of large internal cavities of trees.
p ---- -
camera-objects
=
I. -
- -==-::.=...-=...-.:.
- ---
distance tEe
F.O.V. (m) 0.13 0.5 0.8 1.4 1.7 2.1 2.5 3.2 3.5
I.F.O.V. 0.13 0.39 0.65 1.04 1.30 1.56 1.95 2.34 2.6
(mxl0- 2)
Fig. 3.6. Geometric description of camera field of view and of the instantaneous field of view of
the sensing element equipped with a 10° X 10° lens. (Catena 1993a, with permission)
Table 3.3. Thermal scanner lens for the instantaneous field of view of sensing elements and the
corresponding field of view of the camera. (Catena et al. 1990, with permission)
Lens Instantaneous field of view The square width side of the field
of scanning element (em) of view of camera (m)
10° x 10° 1.3 1.75

25° x 25° 2.5 4.45
45° x 45° 3.5 8.16
The infrared camera used by Catena (1993a,b) was used in the ambient tem-
perature range from -15 to +55°C. The sensitivity was 0.1 °C for +30 °C object
temperature.
For the nondestructive evaluation of the mechanical behavior of wood
under loading, Luong (1996) used mobile equipment (AGA 782 SW) that com-
prises a set of infrared lenses that focus the electromagnetic energy radiated
by the object, an electro-optical mechanism that determines the field of view
with a scanning rate of 25 fields/s, a set of relay optics containing a selectable
aperture unit and a filter cassette unit, a photo-voltaic shortwave infrared
detector that delivers a signal proportional to the radiation from the object
with a spectral response between 3.5 and 5.6 !lm, a liquid nitrogen Dewar
chamber and a control electronics unit that produces a video signal on the
screen. Figure 3.7 shows the utilization of an infrared camera for laboratory
measurements for thermographic detection of artificial defects and delamina-
tions in plywood. The specimens were illuminated by halogen lamps, and the
temperature was measured in defective and clear zones.
Transient thermography, for which the heat flow is generated by a short light
pulse, is able to give more detailed information about the inspected object, by
a most sophisticated analysis of the temperature differential. Lock-in ther-
mography combines conventional thermography with' modulated thermal
waves. The block diagram of the experimental device for thermal wave imaging
with lock-in thermography is shown in Fig. 3.8.
The sample is exposed to a sinusoidal illumination of a heat beam. A com-
puter controls the sinusoidal illumination power produced by a conventional
halogen lamp. An infrared filter is used to avoid the emission of infrared radi-
ation. During each modulation period, successive scans are performed to
obtain the image from temperature differentials from the magnitude and phase
of the signals. The lock-in thermographic equipment used for the detection of
delaminations in wood-based composites is shown in Fig. 3.9. The phase angle
incandescent lamp
....-------",(1
sample ~
§~E>m'=~'
(mm)
~
420 ~
650
o
color monitor
color
video copy
processor
Fig. 3.7. Infrared camera for laboratory measurements, for thermographic detection of artificial
defects and delaminations in plywood (Xu et a1. 1994; courtesy of Forest Products Society)
88 Thermal Imaging
Fig.3.8. Block diagram of the experimental device

for thermal wave imaging with lock-in thermogra-
phy. (Wu et al. 1992, with permission)
computer
and
amplifier
sy tern
AGEMA lR camera
sample
(Pc!
~
power supply
Fig. 3.9. Equipment used for the detection of delamination in wood composites samples. (Wu
images were taken with 0.03,0.06 and 0.12Hz modulation frequencies. Three
minutes are necessary to obtain an image.
Lastly, we wish to cite one of the most interesting techniques, which is the
lock-in vibro-thermal method (Busse et al. 1992a), for which the thermal waves
are generated by a periodical stress inside the sample and not only at the
surface. The phase of thermal waves with respect to the mechanical excitation
provides information about the integrity of the samples. Figure 3.10 shows
the experimental set-up used by Rantala et al. (1996) for polymer materials.
Applications 89
uitcasonic transducer
and water contamer
JR camera and the control unit
r-. r-I
I ! • ~
o
Fig.3.10a,b. Lock-in vibro-thermography setup. a Amplitude modulated frequency in tension.
b Modulated ultrasonic excitation. (Rantala et al. 1996, with permission)
The amplitude modulation can be obtained with a mechanical shaker or with

ultrasonic excitation. To our knowledge, this method has not yet been used in
wood science, but its high efficiency for imaging of polymers encouraged us
to present some details at the end of this chapter with the hope that one day
it can be a source of inspiration for further research. Ultrasound lock-in ther-
mography is a very interesting research field as was noted by Hennecke et al.
(1979) and Salerno et a1. (1997, 1998). Ultrasound burst phase thermography
combines the advantages of lock-in and pulse thermography as pointed out by
Dillenz et a1. (2000b).
3.4 Applications
The field of application of thermal wave nondestructive testing in forestry

science is very large, ranging from forest fire surveillance with live infrared
90 Thermal Imaging
imaging superimposed on visible terrain view (Tice and Euskirchen 1976) to

mapping forests (Catena 1991; Castagnoli et al.1997), recording thermography
on forest environments (Wickmann 1985), locating potential frost pockets in
the forest (Laurence and Banner 1980), establishing the effect of acid rain on
trees (Catalano et al1986), inspection of the sanitary state of urban green zones
(Catena et al. 1995) or making decisions regarding the possibility of saving
trees of historical interest (Catena 1989,1997).
Thermal infrared imaging has also been used to study the consequences of
induced embolism in the path of sap flow in living trees (Anfodillo et al.1993).
The field of infrared thermography is endless (Nicolotti and Miglietta 1998).
An interesting application was reported by Wisniewski et al. (1997) who
studied the thermal response of plants during freezing. This technique could
be extended to a study of frost crack propagation in standing trees. Knowledge
of ice nucleation in wood is important in devising strategies for frost control.
In the following pages, we will consider some applications of thermal
imaging related to specific problems of wood in living trees, in timber and in
wood-based composites such as imaging of cavities in standing ornamental
trees, imaging of the internal structure of lumber at the macroscopic level,
imaging of moisture distribution in lumber, imaging of different stages of solid
wood rupture phenomena, and imaging of defects in wood-based composites
and in lumber joints.
3.4.1 Imaging of Defects in Trees
The assessment of the stability of weakened ornamental trees in parks and

gardens is today a public safety hazard problem. We have to emphasize here
Catena's pioneering activity in this field in his laboratory in Rome (Istituto
Superiore di Sanita). Thermal detection of cavities and internal decayed tissue
in trees has been reported in several publications since 1986 by Catena and
coworkers. (Catena from 1989,1991,1992, 1993a,b, 1997,2000). The validity of
the thermographic method to observe temperature differentials with an
infrared camera was demonstrated for softwood and hardwood trees.
Figure 3.11 shows the photographic and infrared image of an old Acer
plata no ides tree. On the photographic image, a big cavity is observed, which
corresponds to an old pruned branch. The question to be answered is whether
this cavity has ramifications within branches. The presence of internal cavities
induces a variation of the surface temperature, which will appear on a moni-
toring screen as a darker or lighter area, as can be seen from Fig. 3.11. The
size and the precise location of the cavities can be easily observed on the
infrared image. The sound wood is represented in green. The blue color defines
well the presence of cavities in trunk and branches. The cavity observed in the
stem is extended into the bifurcation of the branches. The cavities are smaller
in the branches than in the stem. Catena and Catena (2000) also demonstrated
Applications 91
Fig. 3.lla,b. Cavities in standing trees. a photographic image. b thermographic image of cavity
(in blue) in trunk and in branches. (Catena and Catena 2000, with permission)
that pathological attack on the roots of trees can be observed with thermal
imaging. For the nondestructive estimation of the sanitary state of trees in
parks and public gardens, the authors suggested different periods for the
infrared inspections, ranging from 1 year for trees having advanced deteriora-
tion to 3 years for trees with 'medium deterioration' and up to 8 years for
healthy trees.
3.4.2 Timber Imaging
Thermographic images of timber have been obtained with active and passive
methods. The active methods generate the required temperature field in the
volume of the component under test, which acts as a distributor of the heat
source caused by hysteresis or by thermoelastic effects. One of the major inter-
ests in the development of a thermal imaging technique for the lumber indus-
try is related to the improvement of the accuracy of stress grading. Masuda
and Takahashi (1999,2000) studied the feasibility of this methodology using
western hemlock. Steele et al. (1998, 2000a,b) used passive methods for detec-
92 Thermal Imaging
tion of knots in softwoods and hardwoods. Sadoh and Murata (1993) and
Murata and Sadoh (1994), Tanaka (1994) as well as Naito et al. (2000) reported
the thermographic measurements of the slope of grain in lumber. In this
section, both approaches for timber imaging will be discussed.
3.4.2.1 Imaging of Knots
For the detection of knots, Masuda and Takahashi (1999) reported measure-
ments on western hemlock lumber under cycling loading in bending and com-
pression. For bending loading, the measurements were performed on lumber
with knots. The specimens had a 45 x 105mm cross section and a length
of 600 mm. The lumber was tested under cycling bending at four points,
starting with 5 Hz. This frequency decreased with increasing bending load
because of experimental equipment limitations. The tension and compression
sides of the specimens were inspected. Two experimental configurations are
discussed, the first one with a specimen including an edge knot at the tension
side and the second one with a specimen including a central knot, for which
the variation of the temperature distribution as a function of loading are
shown. For the bending loading, a temperature rise was observed from 1000 N
loading, corresponding to 120N/cm2• The ultimate failure load was observed
at 2973N, corresponding to 360N/cm 2• Thermographic detection of defects
was started with one third of ultimate strength. Improvement of defect
detection can be obtained with an increase in repetition loading to more than
50 times. The hysteresis effect favored the low thermal conductivity of the
defective zone. Imaging of wood structure and knots was possible under high
speed cycling loading. Clearly defined annual rings were observed at 6000 N
(Fig. 3.12).
I.
18.i
E
~ "
~
~
17,,.
11
16.-5
2000 <000 6000 iIOOIl 10000
a) b) ~1Cdloodt~Gf)
Fig.3.12a,b. Imaging of wood structure and knots at 6000N cyclic loading oflumber. a Image of
annual rings around different knots at 6000N. b L left edge knot; R right edge knot; N central
line of the sample; A, B, C knots in the central part of the sample. (Masuda and Takahashi 1999,
with permission)
Applications 93
The relationships between the temperature and the hysteresis curves

of deformation and of energy loss by a hysteresis loop are shown in Fig. 3.13.
Temperature rises in compression were much higher than that in bending. In
compression loading, high speed cycling loading induced larger hysteresis heat
than did bending, but only small differences were observed between static and
dynamic compression.
For improvement of the accuracy of stress grading of lumber it was
suggested that measurements of the static modulus of elasticity should be
nol9
1600
2.5
1400
1200
1000
c
~ 800
-:iI
.Q
600
400
0.5
200
10 15
a) Energy in a hysteresis loop (1) b) Energy in a hysteresis loop (J)
no 13.
2.5
E
0 1.5
·c
2
"
~
g
0.5 0.5
-0.5 L-_ _----1_ _ _--L.._ _ _.L-_ _-----l

-0.5 ' - - - - - - - - ' - - - - - ' - - - - " - - - - - - - - '
~5 15 05 15
c) Energy in a hysteresis loop (J) d) Energy in a hysteresis loop (1)
Fig.3.13a-d. Relationship between temperature rise and energy loss with hysteresis. a Hystere-
sis loop on load-deflection curve. b Static bending. c Static compression. d Dynamic compres-
sion. The number of the specimen is referred to on each curve. (Masuda and Takahashi 1999, with
permission)
94 Thermal Imaging
associated with thermal imaging of the temperature rise induced by the energy
loss in the hysteresis loop during cycling compression loading. Passive
methods for detecting the differential thermal response of knots and of clear
wood were reported by Steele et al. (2000a). The thermographic images were
produced by two methods: by heating with a radio frequency heating system
during lOs at 35kHz, and by heating with infrared quartz lamps of 6000W at
in-line lumber production speed. Temperature differences and differences
between several physical parameters of wood are shown in Tables 3.4 and 3.5.
For softwoods, the difference between the physical parameters of knots and
that of clear wood was more important than in hardwoods.
It was concluded that for softwoods the temperature difference of 2.07°C
between knots and clear wood was significantly large enough to allow the
detection of knots with thermographic imaging methods. For hardwoods, the
Table 3.4. Temperature differences between clear wood and knots in lumber as a function of
species and of the heating system. (Data from Steele et al. 2000a, with permission)
Species Heating Radio frequency System Infrared system

system Temperature Temperature difference (0C)
difference (0C)
Softwoods
Loblolly pine 3.29 0.63
Eastern white pine 2.10 2.33
White spruce 0.80 1.19
Ponderosa pine 2.66 1.22
Douglas fir 2.07 2.35
Eastern red cedar 1.49 0.65
Hardwoods
Red oak 0.40 0.75
Yellow poplar 0.10 0040
Black walnut 0.19 0.12
Basswood 0.12 0.77
Black cherry 0.56 0.66
White ash 0.49 0.95
Table 3.5. Difference between the mean values of physical parameters of clear wood and of knots
in different softwoods and hardwoods. (Data from Steele et al. 1998, 2000a, with permission)
Parameters Softwoods Hardwoods Ratio
Specific gravity (kgm3 ) 400 180 2.22

Moisture content (%) 1.12 0.51 2.24
Extractive conrtent (%) 12.11 0.58 20.87
Fibril angle (0) 14.11 6.22 2.2rJ
Temperature with radio-frequency system 2.04 0.31 6.58
Temperature with infrared system lAO 0.61 2.29
Applications 95
difference is very small and only 0.31°C, which makes the differentiation
between clear wood and knots quite difficult. Rapid heating with infrared
lamps is less efficient than radio frequency heating for lOs. The on-line valid-
ity of the infrared heating system is strongly dependent on the technological
development of new and more sensitive infrared cameras.
The experimental data reported by Steele et al. (2000a,b) were in agreement
with data obtained by Murata and Sadoh (1994) on heat absorption by knots
of hinoki (Chamaecyparis obtusa) and sugi (Cryptomeria japopnica) lumber.
The rise in temperature of the lumber surface irradiated with incandescent
lamps indicated an increasing heat absorbency by the knots of about 1.6 times
that of the clear surrounding wood.
Quin et al. (1998) employed an infrared camera to determine the influence
of the radiant heating periods and the subsequent cooling times on tempera-
ture differences between knots and clear wood in southern yellow pines. The
study demonstrated that knots can be detected regardless of their diameter in
heated lumber, depending upon the length of heating period and of elapsed
cooling time.
For in situ detection of knots, Tanaka and Divos (2000) proposed measure-
ments of temperature differences between knots, clear wood and decayed
wood in sugi samples of 10 x 10 x 80mm, using the daily natural temperature
variations ranging between 26.3 and 3S.7°C. Figure 3.14 shows the tempera-
ture differences between knots and decayed areas and the temperature of the
air during daytime. The air temperature increases from 5 a.m. to 4 p.m. The
evolution of temperature differences for knots is opposite to that for decayed
areas. A good image resolution was achieved with a temperature difference
of 0.7°C, which implies that the method can be successfully used for the
detection of defects in situ. The detection of knots on wood boards with ultra-
sound thermography was also studied. The amplitude image was obtained with
0.8 r----r=--,
0.6
G '----===::..:.J
t.- 0.4
8
e0.2
~ .,""
~ 0 5 .·. -"'6'''''-.6~:--IT+---rj(-1"""
. 1 ---'1""
2 ---' 14- 1"-5- : Ir hh----.-j
13--'-
a-0.2 clock time (hours) .,i
~ }... 1 j
t
v . ~I~ ~ I~ ~ ~ 1'..
., ' .\.
III! , - \ .. ••
~-0.4 " ' w~,
£ \ l ~ ( .~ .'r{' '.l •
06
-. ~
V ~
-0.8
Fig. 3.14. Detection of knots and surrounding area. Temperature difference of knot and decay
areas compared with the temperature of the air during daytime. (Tanaka and Divos 2000, with
permission)
96 Thermal Imaging
ultrasound lock-in thermography. This method works, but unfortunately the

input coupling of high power ultrasound is very critical. The damage thresh-
old can also be exceeded in wood.
3.4.2.2 Imaging of the Slope of Grain
Modulation of the temperature is affected by the anisotropic nature of wood,

which creates an elliptical pattern of temperature rise around the heat source.
The spatial dependence of the temperature modulation is affected by the
thermal structure of the sample. Consequently, the determination of the slope
of grain with a thermal method is an inverse problem approach.
Thermal conductivity measured by the laser flash method as a function of
the grain angle is shown in Fig. 3.15. The relationship between the thermal con-
ductivity and the grain angle deduced by Murata and Sadoh (1994) is given for
hinoki and sugi in Eqs. (3.10) and (3.11).
A. = 0.154 P sin 2 8+ 0.476 p cos 2 8+ 0.025 (3.10)
A. = 0.146 P sin 28+ 0.471 p cos 2 8+ 0.025 (3.11)
where A. is the thermal conductivity (kcal/mh°C), p is the density (g/cm 3) and
8 is the grain angle (degree).
The correlation between the slope of grain estimated by thermography
and by visual inspection for three different species, sugi, western hemlock
and buna, is significant and is between 0.71 and 0.87. Bernard et al. (1996)
measured the infrared birefringence and determined the slope of grain
GO.6
o A
.t::
~
6 0.4
o
';>
'is
.g0.2
c:
8
<0
.sE 0 0.4 0.6 0.8 1.0

density (glcm 3 )
1.2 o 30 60
grain angle (deg.)
90
Fig.3.15A,B. Relationship between thermal conductivity, density and grain angle. A Thermal
conductivity and density at constant grain angle. Filled circles The angle of heat flow is 0°; open
circles 15°;filled squares 30°; open squares 45°; filled diamonds 60°; open diamonds 75°;filled tri-
angles 90°. B Thermal conductivity and grain angle at constant density, symbolized as wood with
density ranging between 360 and 680kg/m3. Open circles The density for 360g/m3; open squares
for 500kg/m3; open diamonds 680kg/m 3: open triangeles also for 680kg/m3. (Murata and Sadoh
1994; courtesy of the Japan Wood Research Society)
Applications 9?
and noted that this technique is insensitive to wood moisture content and
temperature.
Naito et al. (2000) verified the feasibility of thermographic imaging, using
the finite element method to demonstrate the transient temperature on the
wood surface. The experimental device is shown in Fig. 3.16. The finite element
is seen in Fig 3.17. The variation of the temperature distribution with heating
time, rising from 10 to 60s is shown in Fig. 3.18. These results were compared
with thermographic images. Good agreement was observed between finite
(deflexion) (load)
····..................................
···· ....
··· ..
L-----I'--,-,
CCD camera;
·............... ..... ....... -.....

computer
Fig. 3.16. Experimental device. (Naito et al. 1998, with permission)
(uni t: mill)
Fig.3.1? Thermographic measurements of the slope of grain. Finite element model. The heat
flux is represented by the vertical arrows in the central region. (Naito et al. 2000; courtesy of the
Japan Wood Research Society).
98 Thermal Imaging
Fig. 3.18. Temperature distribution as a function of heating and slope of grain deduced with the
finite element method time obtained with the finite element method. (Naito et al. 2000; courtesy
of the Japan Wood Research Society)
elements simulation and the real experimental images. This allows the con-
clusion that the slope of grain can be measured with an infrared camera.
3.4.2.3 Imaging of Wood Density
Density variation mapping of the annual ring with high resolution (100 Jlm)
can be achieved with the far-infrared spectroscopy method in the range of T
Hz. The experimental device as described by Koch et al. (1998) is presented in
Fig. 3.19. Density imaging is obtained with coherent far-infrared pulses which
are generated and detected by small photoconductive dipole antennae gated
by ultra-short laser pulses in the near-infrared. The pulses are "guided by off-
axis parabolic mirrors to form an intermediate focus of sub-millimeter diam-
eter through which the wood samples are scanned in a raster pattern. At each
sample point a complete transmitted waveform is acquired and subsequently
Fourier transformed. The transmitted intensity integrated over a given window
is translated into a gray value to finally obtain a pixel. The data acquisition rate
is 20-50 Hz and the time required to generate a two-dimensional image is
typically 10 min, depending on sample size and resolution".
Figure 3.20 shows the density mapping of beech and balsa obtained by the
transmission method through a 1.7-mm-thick microscopic section. In beech,
the dark blue and red areas are zones of low transmission in latewood and in
medullary rays, ranging between 700 and 1400 kg/m3. It is evident that the large
Applications 99
eyeiece
Fig.3.19. Experimental device for T Hz imaging of density variations in the annual ring. (Koch
et al. 1998, courtesy of Wood Science and Technology)
rays have a higher density than the latewood. This is in agreement with data
obtained by Keller and Thiercelin (1975) with an X-ray technique. In early-
wood containing vessels of about 250-l1m diameter, the density is much lower
than in latewood, ranging between 500 and 600kg/m3.
It is well known that in contrast to beech, in balsa only the vessels and fibers
can be identified. In Fig. 3.20, the vessels are very well identified as large yellow
zones. The large dominant red zones are produced by the presence of fibers
with a density of about 470 kg/m\ while the radially oriented blue zones cor-
respond to the parenchyma ray cells with a relatively low density of about
300 kg/m3 and where starch is a major component.
Concluding this study, it can be said that the false color image of density
variation in transverse sections of wood gives us a better understanding of the
anatomical structure of wood.
Since the aim of our analysis is to show the way in which thermal techniques
can illuminate wood structure, it is important to note the pioneering
work of by Moksin (1993), who measured the decay time of opto-thermal
signals observed across a pine sample scanned with a pulsed laser source
with 100-l1m step size. The opto-thermal decay time 't was calculated from
Eq. (3.12)
(3.12)
where a is the absorption coefficient and D is the thermal diffusivity of the
sample.
Figure 3.21 shows the decay curve profile of growth rings pine scanned with
a step size of 100 11m, which is consistent with the earlywood and latewood
zones of the annual ring. The minima represent the thinner and dark zones of
latewood, which are more efficient in heat dissipation than the clear earlywood
100 Thermal Imaging
--
0
Density (g/cm 3 )
2 1.3000 - 104000
1.2000 - 1 .3000
4 1.1000 - 1.2000
-
1.0000 - 1.1000
6 0.9000 - 1.0000
0.8000 - 0.9000
8 0.7000 - 0.8000
0.6000 - 0.7000
10 0.5000 - 0.6000
004000 - 0.5000
12 0.3000 - 0.4000
0.2000 - 0.3000
0.1000 - 0.2000
14
0-0.1000
a 2 4 6 8 10 12 14
a) x (mm)
3.0
Xylem vessels
2.5
Transmission
2.0 _ 567·600
_ 535-567
1.5 502 - 535
_ 470 - 502
1.0 _ 437 - 470
405 - 437
0.5 _ 372-405
_ 340 - 372
0
_ 308-340
0 2 3 4 5
275 - 308
b) x (mm)
Fig. 3.20. Density mapping of a transverse section of two species: a beech and b balsa. The size
of the specimen was 14 x 14 x 1.7mm. (Koch et al. 1998; courtesy of Wood Science Technology)
areas. The distance between two minima corresponds to an annual ring width
of 4.4mm.
For the future, we suggest that this technique be utilized for the mea-
surement of the density profile of wood. This opto-thermal method can be
validated with the X-ray densitometric method.
Applications 101
Fig.3.21. Decay curves of an opto-thermal signal

in pine. (Moksin 1993, courtesy of Wood Science
Technology)
10
o~!:=====:!Y
o 10 20 30 40 50 60
time / IO-\
3.4.2.4 Imaging of Moisture Content Distribution
Indirect evaluation of moisture content in wood and its distribution can be

performed with photothermal techniques as demonstrated by various authors
(Cielo et al. 1988; Bernatowicz 1994). The photothermal parameters related to
wood moisture content are: thermal conductivity, K, thermal diffusivity, u,
thermal effusivity e, wood specific heat, C, wood density, p, and surface losses.
The specific heat for wood is between 0.25 and 0.35 ca1!gK for wood densities
ranging from 300 to 900 kg/m3. The specific heat of wood is lower than that of
water which is 1 ca1!gK.
Measurements of the photothermal parameters of a wood sample can be
performed on the front or back surfaces. The front surface measurements can
be carried on a solid wood specimen of normal thickness, on which the surface
temperature measurements need shorter observation times than for that on
the back surface. The back surface measurements are convenient because they
are not affected by reflected light noise. The value of the thermal diffusivity
can be evaluated from the shape of the thermograms without any absolute cal-
ibration of the detector. Cielo et al. (1988) noted that surface temperature mea-
surements are more sensitive to lumber moisture content than thermal
diffusivity measurements.
In one-dimensional approximation, the surface temperature rise ~ T
induced by a constant power flux Q absorbed at the front surface of a semi-
infinite irradiated material during a time period 't is expressed by Eq. (3.13), if
the time t < 't
~T = 2 Q(d'2(1tK P CfI/2 - P(T) (3.13)
102 Thermal Imaging
and if the time t > t:

LlT = 2 Q(TI K P CfI/2[cd/2 - Ct - t)1/2] - PCT) (3.14)
where peT) is the temperature-dependent heat losses due to air convection,
radiation and phase changes.
The thermal diffusivity of the material, in this case of spruce submitted to
laser irradiation for 5 s, can be calculated from:
(3.15)
where d is the sample thickness and t is the half-rise time.

The thermal effusivity e can be calculated from Eq. (3.16) if the absorbed
power Q and the heat losses are known:
e= CK pC)I/2 (3.16)
The anisotropic character of solid wood was demonstrated also for its
thermal properties as can be seen from Table 3.6 and Fig. 3.22. The data in
Table 3.6 were obtained with a flash thermometry analysis, the front of the 3.4-
mm veneer sample was heated for 1 s, and the back surface temperature was
recorded with an infrared detector. The thermal diffusivity in the longitudinal
direction is twice that in the radial or tangential directions. The isotherms are
quite circular in the radial tangential plane. In the longitudinal-tangential
plane the fast axis is in the fiber direction.
The effect of the moisture content in the range 0-160%, on LlT and a
deduced with Eq. (3.13) are shown in Fig. 3.23 for softwood veneer, red cedar
and spruce. Both of the thermal parameters studied decreased with increasing
moisture content. The variation as a function of moisture content of the ratio
between the thermal characteristics (conductivity, effusivity and temperature
rise) at a given moisture content and under oven-dry conditions is given in
Fig. 3.24. An increase in all characteristics was observed with the increase in
moisture content. The contribution of these three variables to thermal phe-
nomena is of the same order of magnitude. Above 30-40% moisture content,
the increase in the temperature rise ratio is much more important than the
conductivity and effusivity ratios. Referring to Eq. (3.13), it can be noted that
at low moisture contents surface losses have a relatively small effect on the tem-
perature rise. At high moisture contents, corresponding to the presence of free
Table 3.6. Thermal diffusivities in red cedar at 30 percent

moisture content as a function of wood anisotropic direction.
(Cielo et al. 1988, with permission)
Thermal propagation direction Thermal diffusivity (mm2/s)
Longitudinal 0.41
Radial 0.15
Tangential 0.16
Applications 103
Fig.3.22a,b. Thermal anisotropy of a 3.4-mm softwood veneer sheet at 40% moisture content in
a radial tangential plane; b longitudinal tangential plane. The temperature distribution is dis-
played over a 2 x 3cm area. (Cielo et al. 1988, with permission)
104 Thermal Imaging
LiT (K) (mm2/s) Fig.3.23. Thermal diffusivity (a) and tem-

perature differential (~T) as a function of
80l:l'~ (a) S0 ftw00 d Veneer J o . 8 moisture content on the front surface of the
40, ~ 0.4 specimen, on which the temperature increases
20r: ~b)~o___ 0.2

after 5 WI cm2 laser irradiation for 4 s. (Cielo et
aL 1988, with permission)
10 L-..I.-_'--_~J._._J_ 0.1
:~~'. (b) ,
RodCeruu J'~:: 0.2
10 0.1
80!"..., Spruce Sapwood . 0.8
40 ~ ·0.4
20 (b)~ .0.2
10 ~ '-----.J_-=:::.:,-.. . :.0.1
o 40 80 120 160
percent moisture content
6 -
o LiTdlLiT
5 .. e/ed
-KlKd
4
_A-
A--
(b) ..
.g
e 3 - o=----=======. ____ .__ e-
~:-:::::::--- (a)
2- ~.
I ~~ .............-.- ............... _........ -.- .... _.......
01_--1--
o 20
1- - - ' - -
40 60 80 100 120
percent moisture content
Fig.3.24. Relationships between moisture content and the ratios of thermal conductivity; a Filled
circles Thermal effusivity: b filled triangles front surface temperature variation; and c open circles
under laser irradiation for a softwood veneer sample. (Cielo et aI. 1988, with permission)
water, the effect of the rise in temperature is predominant, probably because

of the losses due to surface evaporation.
Figure 3.25 shows a thermal image of a veneer sheet with zones of differ-
ent moisture contents, which was artificially moistened on the left side
and was heated at 10 K above ambient and finally allowed to cool down by
vaporization. The knots dry faster than the surrounding clear wood because
Applications 105
Fig. 3.25. Thermal image of a veneer sheet of 30 x 30 x OAcm with zones of different tempera-
tures and moisture content. (Cielo et al. 1988, with permission)
lO6 Thermal Imaging
142m
-
I I
I I IW WI I J
I I r-
p
<nfrared heaters C
S
J~ h I ....
I I IW WI I I
I L f-1 L ...J
I
A=3/16 Allen Screw
C= moveable clamp 2.84 ill
p= photocell r-
S= infrared heat sensor p
W=wignut C
S
....
~ h
W W I
r-
p
v-,
S
~
Fig. 3.26. Infrared measuring device for green veneer and lumber S Infrared heat sensor;
P photocell; W wingnut. (Troughton and Clarke 1987; courtesy of Forest Products J)
of their particular structure. The moisture content of such a specimen con-

taining various defects exhibits important local variations. The moisture
content of the whole surface can be expressed by an average value with a wide
distribution.
For in-line measurements of the moisture content in unseasoned veneer and
lumber Troughton and Clarke (1987) used the system shown in Fig. 3.26.
The device designed for a mill trial is equipped with two 5000-W heaters and
three infrared sensors connected with a computer and used for unseasoned
60-mm lumber with a speed of l3 m/min. The relationships between the mois-
ture content and the surface temperature for veneer and lumber are shown in
Fig. 3.27.
3.4.2.5 Imaging of Wood Rupture Phenomena
Thermal wave generation in wood under stress has opened new directions
for studies of wood rupture phenomena. The propagation of thermal waves in
specimens under stress can be monitored by a non contact infrared technique,
Applications 107
OJ
..§ 160i,
20 40 60 80 100 120 140 160 180

a) moisture content %
30 40 50 60 70 80 90 100 110 120 130

b) moisture content %
G:
"--'l30
~
1;j
[ 120 1
8 I
B I
~II 0 ~
] I
g100 II
~
90 1- - - -
20 40 60 80 100 120 140 160 180 200
c) moisture content %
Fig.3.27a-c. Relationships between moisture content and surface temperature. a Relationship

between veneer surface temperature and moisture content. b Relationship between final surface
temperature and moisture content (%) for unseasoned 2 x 6-in.lumber. c Relationship between
final surface temperature and moisture content (%) for 1/8-in. spruce veneer. (Troughton and
Clarke 1987; courtesy of Forest Products J)
108 Thermal Imaging
as demonstrated by Luong (1995, 1996), Okumura et al (1996) and Naito et al.

(1998).
Infrared imaging on pine specimens under an unconfined compression test
is shown in Fig. 3.28. The effect of anisotropy was demonstrated using three
different specimens having the major axes in longitudinal (L), radial (R) and
tangential (T) directions. Analysis of the thermal images was performed "by
discriminating the intrinsic dissipation from thermal noise by subtracting the
thermal image at a reference time from the thermal image taken at a given
applied stress level." It was observed that for each anisotropic direction the
rupture mechanism is different, and typical modes of intrinsic dissipation
localization describe in colors the damage process.
longitudinal radial tangential

Fig.3.28a,b. Infrared imaging of pine specimens under compression. a Specimens with major
axis in longitudinal, radial and tangential directions. b Localization of intrinsic dissipation
in wood specimens as a function of anisotropy (each color hue corresponds to 0.2 DC) in
longitudinal, radial and tangential directions. (Luong 1996, with permission)
Applications 109
Okumura et al. (1996) studied the behavior of three tree species under static
unconfined compression. The profile information of the temperature distrib-
ution was related to strain measurements obtained on an array of 66 points
(Fig. 3.29). With specimens under longitudinal compression, a rapid increase
in temperature from 19.6 to 22°C along the fracture plane was observed (Fig.
3.30 at point d). The temperature increase in specimens under radial and tan-
gential compression (Fig. 3.31) was between 20 and 21.2 DC. In both cases, the
highest temperature corresponds to the fracture line. As can be seen, the most
important increase in the temperature was observed in specimens under com-
pression in the longitudinal direction.
Narito et al. (1998) studied the temperature distribution in yellow cedar
specimens under static bending test in four configurations. In this section, we
present the thermographic images obtained for specimens under three-point
bending (Fig. 3.32), with a 5-mm-diameter hole drilled in the compressed zone
of the specimen. At a very low loading level, the temperature increased 0.6°C
under the loading point and the presence on the hole is not observable.
Increase in the loading resulted in a better image of the hole. The rise in
local temperature around the hole is similar to that in the fractured zone of
the specimen.
a) b)
~1,
~I 38
A==;?[12 I 9
c)
: . ·. ~18'30
..................... ~-30
I 87 -100 I
Fig.3.29a-c. Specimens for testing under compression loading. a Longitudinal loading on hinoki
specimen; b radial loading on hinoki specimen; c loading on hinoki, sugi and white seraya
composite specimens. Note that for strain measurements strain gauges were used as a lattice
represented by the lattice dots. (Okumura et al.1996; courtesy of Wood Research Institute, Kyoto
University)
110 Thermal Imaging
..... .
SpecImen hlnold
2 4
a) b) a AmQCJnt of compression (mm)
c)
Fig.3.30a-c. Infrared imaging of specimens under compression loading related to stress distri-
bution in the longitudinal direction. a specimen; b load strain relationship; c thermographic
images corresponding to the points a, b, c, d, e, f from b. (Okumura et al. 1996; courtesy of Wood
Research Institute, Kyoto University)
3.4.3 Imaging of Delaminations in Wood-Based Composites
Detection of delaminations in wood-based composites with non contact and

nondestructive methods is one of the most important aims of industrial
quality inspection. Thermographic methods detect the delamination areas in
composites, using a heat flow that is disturbed by the presence of the defect. It
also causes a variation in temperature at the surface of the inspected object.
Heat can be applied in the front or to the back surface of the laminate. In the
second case, the front surface temperature is observed. In this way, the heat
flow is conducted from the heated back surface toward the cooler front surface.
This heat conduction phenomenon is very complex and is determined by the
anisotropic and heterogeneous nature of the laminate. The defect present
between two laminae is a barrier to the heat flow. The heat flow, which pro-
duces an image on the front of the sample, "incorporates" the presence of the
J
Compression
load
t
~... .-. . . .
It • • • • • .- ~ .
CH1 .. CH2
. . . .. ..
- - . ~~
~
.. :,... ~-~~.
"E-
oo- ..
CH3 ~
".o·
.......
.. . . ::>
'"
Fig.3.31. Infrared imaging of specimens under compression loading related to the stress distribution in the transverse plane.
Images of specimens at rupture under different compression loads applied in radial (CH4) tangential (CH1) and inclined --
directions vs. the annual rings (CH3 inclined vs. T and CH2 inclined vs. R). (Okumura et al. 1996; courtesy of Wood Research
Institute, Kyoto University)
Hole of 5 mm N
-
diameter
>-3
::r
(l)
(- ,)/: 61J- 20 a
eo.
280
s
~
&.
a) 300
150
'5;100
~
....
~ 50
1/' ~
0
0 5 10 15 20
b) Deflectfon(mm) c)
Fig. 3.32a-c. Infrared imaging of specimen under static bending loading. a Specimen with a central hole under static bending. b Stress strain
curve. c Temperature distribution as a function of loading and strain, corresponding to points 1, 2, 3, and 4. (Naito et a1.1998; courtesy of Wood
Research Institute Kyoto University)
Applications 113
defect. The thermal gradient on the surface is one of the most important para-
meters to by studied, when the temperature distribution is known.
During the last 20 years, a large number of important publications have been
devoted to the detection of delaminations in different materials, such as those
used in the aerospace industry, the composites for a large variety of applica-
tions (Henneke and Jones 1979; Charles and Wilson 1981; Kou et al.1987; Busse
et al. 1992a; Wu 1992; Wu et al. 1992, 1997a; Bauer et al. 1992; Lebowitz et al.
1996; Dillenz et al. 2000) for ceramics (Lyamshev et al.1995), for materials with
micro inclusions (Willis et al. 1997,2001), for electromagnetic materials (Bal-
ageas et al. 1996), etc. The techniques developed for the wood industries are an
extension of technological progress achieved in this field for different wood-
based composite materials (Schulte et al. 1995). The purpose of this section is
to describe the thermographic methods used for imaging of delaminations in
a large variety of wood-based composites such as: glued laminated lumber,
plywood, medium-density fiber boards and chipboards.
3.4.3.1 Detection of Adhesion Defects in Laminated Wood Composites
The methods for the detection of delaminations in glued laminated wood

discussed in this section were reported by Xu et al. (l993, 1994), Masuda and
Takahashi (2000) and Berglind and Dillenz (2000).
Xu et al. (l994) used an infrared camera to detect the delamination defects
between two veneer laminae of different thicknesses (1.3, 2 and 3.1 mm). The
surface of the artificially produced defects varied from lOx 10 to 50 x 50 mm2 •
The specimens were heated with incandescent lamps. Thermal images clearly
show the area of the implanted defect surrounded by a sound zone, in which
the adhesion between the two laminae is correct.
For the detection of delamination under cyclic bending, Masuda and
Takahashi (2000) used an active laboratory infrared thermographic technique.
The specimens were composed of two laminae to which were glued several
zones of glue discontinuities ranging from 4 to 16cm in size as can be seen in
Fig. 3.33. They were tested between 63 and 72% cyclic loading of the ultimate
failure load. The thermographic image for the specimen with two 16 delami-
nation zones is given in Fig. 3.34 in which the temperature rise was observed
at the glued end of each zone and at the central loading point, corresponding
to tension and compression zones, respectively.
The mechanical performance of the specimens and the corresponding para-
meters for thermally detected defects are shown in Table 3.7. The higher failure
load was observed on the specimen with two zones of 8-cm delamination,
which also corresponds to the higher Young's modulus and failure stress. It was
concluded that the proposed technique can be used for the detection of delam-
ination of specimens.
For industrial applications, Berglind and Dillenz (2000) studied the feasi-
bility of a heating-up thermography method for the detection of delamina-
114 Thermal Imaging
f \ I
glued line
t
/ '''\ -*
~------------~~~/-- ~£~\-----
f f
f f
Fig. 3.33. Glued laminated specimen with different zones of discontinuities in glue layer. (Masuda
and Takahashi 2000, with permission)
Fig. 3.34. The thermographic image for a specimen with two 16-cm delamination zones. (Masuda
and Takahashi 2000, with permission)
Table3.7. Mechanical performance of the specimens and the corresponding parameters for
thermally detected defects. (Masuda and Takahashi 2000, with permission)
Mechanical parameters Delamination Delamination Delamination

length in four zones length in two zones length in two zones
of 4cm each of 8cm each of 16cm each
Detected load, P (kN) 10.3 12.1 5.9

Failure load, P, (kN) 15.3 16.8 9.03
Ratio PIP, (0/0) 67 72 63
Density (kg/m3) 509 551 458
Bending Young's modulus, 9.16 11.17 7.49
three points (GPa)
Failure stress, Om.x (MPa) 83.9 91.1 48.4
Applications 115
~_ _."._~Dm
),6
.. 1" h •
•
Fig. 3.35. The geometry of specimens used for the detection of the delaminations between wood
and urea-formaldehyde glue. (Berglind and Dillenz 2000, with permission)
tions, the defect depth and in plane size of a specimen, as shown in Fig. 3.35.
The heat camera operated with infrared radiation between 8 and 12 ~m and
had a temperature resolution of 0.08 K. The heat source was produced by two
halogen lamps of 1000 W each, which were placed l. 7 m in front of the test
piece.
The image contrast was calculated as the difference between the signal para-
meter corresponding to the defect and to the sound area, as can be seen in
Fig. 3.36. The multiple correlation equation (Eq. 3.17) between the contrast,
C, the defect width, D, and the defect depth, L, below the surface layer of the
exotic wood merbau had a coefficient of determination of R2 = 0.90, which
means that the variability of the population not explained by this regression
is only 10%.
C = 1.3+0.28 D-0.91 L (3.17)
The relationship between the contrast and the thermally determined defect
size is shown in Fig. 3.37, in which large defects result in a higher contrast com-
pared to smaller defects, and defects below 3 mm cannot be detected.
Several factors such as inhomogeneous lighting of the specimen surface and
natural variability of wood species, can influence the temperature distribution
and the corresponding thermal image contrast. The effect of wood species on
the detection of subsurface defects of 24-mm width and 2-mm depth is shown
in Fig. 3. 37 in which four species were analyzed namely: oak, merbau, pine and
alder. A higher contrast was observed for merbau, a lower for alder, depending
on the degree of "darkness" of each species. "Dark" species have a high absorp-
tion that gives rise to a greater heat flow that propagates into the specimen,
reflects again the defect, returns to the surface and is detected with a corre-
sponding contrast by the infrared camera. The sources of errors that can influ-
ence the surface temperature of the specimen are: inhomogeneous lighting
of the surface and the natural variability of wood, differences between early-
wood and latewood, zones of vessels and fibers, the presence of knots and of
reaction wood. The penetration of thermal waves was between 2 and 4 mm,
depending of wood species.
116 Thermal Imaging
.. ." "'
specimen length (mm)

a) b)
Fig.3.36a,b. Infrared image of specimens with defective glue zones and the corresponding signal
as a function of specimen length. a Infrared image where the bright zones corresponds to the
defective areas. The maximum contrast also corresponds to defective zones. b The signal profile
vs. the length of the specimen. (Berglind and Dillenz 2000, with permission)
10
-----
A
9
.,
----
8
::I
c;; ~
> /Z.
..,.y -......
..'""
~
6 ...... Qak
5 -~/ ~ ..... Merbau

'I-..
~
-- Pine
$ ~
-----
4 -- Alder
/ "'-......
lj\
!!
3
~
U
2
o
10 100
log (Thermal defect size (mm/mm))
Fig. 3.37. Contrast in gray scale between a defective glued area with delaminations and sound
area vs. the logarithm of the defect size for different species. (Berglind and Dillenz 2000, with
permission)
Applications 117
3.4.3.2 Detection of Subsurface Defects Under a Veneer Lamina
The plywood composed of veneer laminae is probably the most ancient wood
composite known from antiquity. The decorative value of fine veneer is highly
prized. The veneer is produced from solid wood as thin sheets, usually cut on
a rotary lathe. The thickness of veneer sheets that compose the plywood can
be below 1 mm and the plywood thickness ranges between several millimeters
to several centimeters. The laminae are alternated with grain at right angles.
Today, plywood has a wide field of utilization, from furniture to material
for building framing and other structural applications. The development of
synthetic adhesives and the improvement of quality control contributed to the
rapid expansion of the plywood industry, despite the declining size and quality
of the timber resource.
For the inspection of defects in wood composites under a veneer lamina,
Wu and Busse (1995) suggested the utilization of the lock-in thermography
method. The feasibility of this technique was demonstrated, using a sample
composed of several veneer laminae of different thickness, glued on chipboard
or on a solid wood block of a different tree species. The phase images were
taken at 0.03 Hz. The following items were studied:
- the detection of artificially induced defects in the chipboard. through the
surface veneer sheets of various thicknesses. For this purpose the veneer
thickness was varied in five steps from 0.5 to 2.5 mm. At the same time, under
the surface, holes of different diameters were made in the chipboard, as
can be seen in Fig. 3.38. Holes with a diameter greater than 4 mm were
detected.
- the detection of natural defects such as knots present on solid wood blocks
covered with veneer sheets (Fig. 3.39). The detection of knots was possible
until about 2 mm underneath the surface layer.
- the detection of the nature of the solid wood blocks (oak and maple) glued
under the plywood composed of different veneer sheets (Fig. 3.40). The dif-
ferences between the thermal properties of maple and oak can be observed
for all thicknesses of plywood composed of different laminae.
3.4.3.3 Detection of Defects in Particleboards
Quality monitoring is one of the most important steps in present particleboard

production, with more and more complex equipment, within the framework
of a decrease in quality of raw materials (small diameter logs, chips, wood
waste of different species). To reach the goal of constant quality, the producer
must monitor all the parameters that can effect particleboard fabrication,
using advanced on-line nondestructive testing methods for quality control of
the panels.
The measuring tasks of the on-line production of particleboard are pre-
sented in Table 3.8. In front of each measuring task are the corresponding
118 Thermal Imaging
Diamclcr
-'
.
,,'" ...., • ,• •,*"-'\, •, ,,
'.' '-'
~ ~,
I
(mm)
10
I" \ I
~'\
t
.. \ ,,
".
\ 8
-' -' -'
,-, ,", ,., ,.,
'-' '.' '.' . I 6
(; -
'- I
.:' e:, t)
4
:: :: c c ::
a) I b)
,O. mm 2. mm
A
-.-
ISOmm
Fig.3.38a,b. Detection of artificially induced defects in chipboard through the surface veneer
sheets of various thicknesses. a Geometry of the specimen. b Thermal image. (Wu and Busse 1996,
with permission)
.l
veneered wn d
2.5 u.. " ~I&~ii~

I~'"= (I m,.
a) b)
Fig.3.39a,b. Detection of knots present in solid wood covered with veneer sheets. a Geometry of
the specimen. b Thermal image. (Wu and Busse 1995, with permission)
nondestructive methods used today for quality control. The quality control
operations among which thermal methods have an important role are:
- drying, for non contact infrared reflectance spectroscopy measurement of
the moisture content of wet and dry chips
- gluing, for the measurement of glue distribution on chips with near infrared
spectroscopy as described by Niemz and Sander (1989)
- panel thickness and its surface properties could be detected by phase image.
The thickness accuracy could be of the order of severaillm. (Wu 1994; Wu
and Busse 1995)
Applications 119
a) b)
Fig.3.40a,b. Detection of blocks of different species under the veneer sheets. a Geometry of the
specimen. b Thermal image. (Wu and Busse 1995, with permission)
Table 3.8. Particleboard production measuring task and corresponding methods used for quality
control. (Data from Schulte et al. 1995, courtesy of Tappi J)
Operation Measuring task NDT methods for Thermal methods

evaluation and control
Storage of raw Species mixture No method

material biodegradation
Chipping Chip dimensions Image analysis
Drying Moisture content of Dielectric method Infrared reflectance
wet and dry chips spectroscopy
Gluing Resin content and Image analysis Near-infrared spectrum
distribution Glue color at various frequencies
Mat forming Chip size distribution Gamma ray
mat weight densitometry
Belt scale
Panel quality Thickness Roller head Phase imaging
control Blow detection Ultrasonic attenuation Lock-in thermography
Weight Belts scale
Density profile X ray, nuclear
Mechanical properties radiography
Surface properties Ultrasonic velocity Thermography
- blow detection could be performed with lock-in thermography

- mechanical properties such as the modulus of elasticity of the plate of real
size can be measured using a resonance frequency method. The plate is
tapped with an electronic hammer and the resonance frequency is detected
by a laser probe by means of intensity modulated laser radiation. The width
of the resonance curve at 3 dB is a measure of the vibration damping and of
the corresponding viscoelastic properties of the board.
120 Thermal Imaging
3.4.4 Imaging of Defects in Different Types of Lumber Joints
From an engineering point of view, it has been recognized that there is a per-
manent need for the development of different types of lumber joints for frame
systems of large spans and for members of different widths and shapes for
architectural applications. For structural members it is necessary to insure the
quality control of each element, to optimize the design values and to establish
the strength.
Metal plate-connected wood trusses are largely used in agricultural, com-
mercial and residential constructions. Loading transfer from the metal con-
nectors to the wood member is nonlinear and depends also on the grain
orientation of the lumber. When this complex and heterogeneous structural
member is loaded, it deforms as a whole element. Stress concentration occurs,
and plastic deformations, micro cracking, slippage at component interface
and fracture can be observed in wood or in metallic connectors. A better
understanding of the behavior of the joints is necessary for upgrading the
design procedure. Failure modes of the joint include failure of the wood
member within the plated zone, tension or shear in the metallic connector,
teeth pulling out of the wood, buckling of the plate in gaps between wood
members, etc.
Luong (1996) used infrared thermography with an infrared camera to detect
the damage in splice joints under tension and shear loading before failure, as
can be seen in Fig. 3.41. The infrared detector has a response time shorter than
Ills. The temperature patterns are represented by distinct color hues. Each of
them corresponds to 0.2 °C. The energy emitted as infrared radiation is a func-
tion of thermal conduction, thermoelastic coupling effects and intrinsic dissi-
pations generated by viscosity and lor plasticity. Intrinsic dissipation (Luong
1995,1998) seems to be the most accurate indicator of damage to a mechani-
cal, loaded specimen.
The development of glued laminated timber during the 20th century had a
significant effect on wood use and opened new opportunities for creative prod-
ucts. The capability to make engineered structural products of a variety of
forms became possible with finger joints and with relatively small timber of
high quality.
The thermographic detection of defects in finger-joint lumber was studied
by Masuda and Takahashi (2000) with Douglas-fir lumber specimens of
35 x 75-mm cross section and 600-mm length under bending cyclic loading.
The maximum temperature rise was 3°C for a 1.5 J energy loss. In a defective
sample, local variations of temperature are different in tension and compres-
sion zones.
The defective area has high stress concentration in which heat is generated
by the hysteresis loop of the stress-strain curve. The neutral axis zone has a
rather constant temperature, while in the tension area the temperature rises
from 16.75 to 18.5°C at 1000N repeated load. In specimens with better
Summary 121
Fig.3.41a,b. Infrared images of splice joints under shear loading. a Before loading. b Shear stress
before failure. Color hue corresponds to 0.2°C. c Splice joints under tension. (Luong 1996, with
permission)
adhesion, the dispersion of experimental values around the neutral axis is less
important.
3.5 Summary
Thermal wave imaging that provides a map of surface temperature distribu-

tion is a nondestructive testing technique that uses heat flux to probe the phys-
ical properties of wood with active or passive heating procedures. The main
thermographic methods for solids are: pulse thermography, heating-up ther-
mography and lock-in thermography. Pulse thermography records the tran-
sient thermal response of an object to a thermal pulse, provided by a flash tube.
122 Thermal Imaging
Heating-up thermography utilizes heat sources such as commercial photo-

graphic lighting devices, hot liquids, laser beams, etc. The equipment for
thermal wave imaging uses an amplitude-modulated focused laser beam
directed onto the surface of the sample. At the point of light absorption, some
energy is absorbed by the sample, giving rise to a modulated heat source. This
heat flux propagates as a thermal wave through the material in all directions
and interacts with the structure. The propagation phenomena are influenced
by the thermal impedance of the material, which varies as a function of the
interaction of the thermal wave with the structure.
As an alternative to heating, cooling can also be used with aerosol freezer
spray. The transient character of the temperature field demands a dynamic
tuning of the thermo-vision system. The development of infrared video
cameras has extended the wavelength range of visible light video cameras to
a thermal infrared range between 3 and 12 /lm. Lock-in thermography can
provide three types of images: thermographic, phase and magnitude images.
The development of lock-in thermography allows the utilization of thermo-
graphic equipment to image the average temperature distribution of oscillat-
ing components.
In wood science, thermography is a relatively new field and thermal imaging
of subsurface temperature distribution in wood-based composites to ascertain
the integrity of subsurface structures were first produced less than 10 years
ago.
The active heating procedure, or stress-generated thermal field under
cycling loading, was used in early work on wood thermographic imaging.
The advantage of this technique is that the influence of defects on mechanical
properties of wood can be easily examined. With lock-in thermography, a large
sample depth range can be observed. This method has been used for detection
of delaminations in wood-based composites because of its sensitivity to sub-
surface defects and imaging capability within a short time span.
The passive heating method has a large field of applications for knot detec-
tion, for the detection of the slope of grain, for imaging of moisture content
distribution in lumber, for imaging of wood rupture phenomena, for imaging
of cavities in trees, etc. The thermal image is the result of a very complex inter-
action between the heating source, the material and the defects. The rate of
heat application and the mode of heating by contact or by radiation are factors
of major importance for obtaining good images with correct resolution.
An advantage of passive heating over the active heating procedure is its
ability to produce temperature distributions without resorting to mechanical
loading of the material. The thermal stress or the thermal shock is relatively
low and does not damage the material.
The disadvantage of the passive method is that the thermal images are
transient and require a fast recording system to capture the most interesting
images during the test. Electronic systems (with an infrared camera) or
chemical systems (with liquid crystals) can be used. The most widely used
method is scanning with an infrared camera, which produces an image of
Summary 123
an object through electronic detection of infrared radiation emitted from the

object.
The scanning infrared imaging technique seems today to be the most con-
venient remote sensing method for thermographic inspection of solid wood
and wood-based composites in situ.
4 Microwave Imaging
4.1 Introduction
The microwave scanning technique is a relatively new tool for the characteri-
zation of wood materials. During the last 15 years, considerable progress has
been achieved in the design of scanning probes and in image interpretation
(Martin et al. 1987; Filippini et al. 1990; Chazelas 1991; Yokoyama and
Norimoto 1996; Golosovsky et al. 2000; Kaestner and Baath 2000).
For the correct interpretation of microwave imaging, it is necessary to know
the response of the material to electrical magnetic fields. It is also necessary to
understand the mechanism of contrast and the interaction between the sample
and the probe and to model and to measure the distribution of the electric and
magnetic fields around the probe. Consequently, progress achieved in the basic
knowledge of the dielectric properties of wood and of wood-based compos-
ites will allow new industrial applications of the microwave imaging technique.
Systematic studies on the dielectric properties of different species were
reported from 1948 by Skaar and later by Hearmon and Burcham (1954) and
Brown et al. (1963). The following basic points were discussed:
- the behavior of water and the principal constituents of wood, cellulose and
lignin;
- the determination of the fiber saturation point of wood-water interactions
and the hygroscopic characteristics of wood;
- the measurement and prediction of wood density and moisture content.
The dielectric properties of wood were related to microwave processing. The
advantages of microwave heat processing are numerous, such as: fast process-
ing, small space occupied by the equipment, less temperature degradation of
the products compared with other conventional heating techniques, dimen-
sional stability of the products, selective heating of materials and, lower heat
loss to the surroundings (Dench 1973).
The industrial applications were originally oriented primarily toward
improving of drying and gluing technology. Today, the industrial applications
of the microwave imaging technique are related to the detection of internal
defects such as knots, spiral grain, slope of grain, structural discontinuities, etc.
of logs, lumber, and wood-based composites. Another interesting field of appli-
cation of microwave imaging is the dielectric behavior of vegetative materials,
including leaves, stalks and trunks under various moisture and temperature
126 Microwave Imaging
conditions over a wide range of frequencies (Tan 1981; Ulaby and Jedlicka
1984; Sieber 1985; El-Rayes and Ulaby 1987).
In the first part of this chapter, the dielectric properties of wood are dis-
cussed, including the effects of experimental factors and of physical para-
meters on these constants.
The equipment for laboratory measurements and imaging is presented in
the second part of this chapter. The final part is devoted to the applications for
inspection of forests, logs, lumber and wood-based composites.
4.2 Basic Aspects
The basic aspects of dielectric behavior of wood are treated in several refer-
ence books and articles (James 1989; Skaar 1988; Torgonikov 1990). The dielec-
tric behavior of a solid (Bahr 1982; Zoughi 1990,1996; Ida 1992; Zoughi et al.
1996) is determined by the polarization of atoms and molecules under the
influence of an applied electrical voltage. The polarization can occur at elec-
tronic, atomic, dipole and interfacial levels, and can have different causes that
are strongly dependent on the frequency (Fig. 4.1). Electronic polarization is
determined by the displacement of electrons with respect to the nucleus within
the atoms and occurs in the ultraviolet frequency region. Atomic polarization
is induced by the bending of bonds between the atoms and occurs in the
infrared frequency range. Dipole polarization is caused in wood by the
hydroxyl groups of bound or free water molecules (that are permanent
dipoles), which rotate in the electric field at a relatively low frequency. Inter-
facial polarization is induced by the accumulation of charged ions at interfaces
between different regions within the cell wall of the wood (i.e., between the
middle lamella and the secondary wall, or between crystalline and amorphous
regions).
The mechanism of electrical conduction of wood has been explained by
several theories (Skaar 1988) and it is generally accepted that the electrical
charges are carried by the ions present in the wood cell walls and in cellulose.
In a direct current electric field, cellulose exhibits typical ionic conduction
polarization phenomena. Temperature increase in dielectric solids is produced
by an increase in the energy of the continuous random vibrations of molecules.
When an electric field is applied to a dielectric material, the randomly oriented
dipoles align themselves in a direction opposite to that of the applied external
field. In this configuration, the energy supplied by the field is stored in the mol-
ecules as potential energy. By the removal of the electric field, this potential
energy of ordered dipoles can be converted back into the kinetic energy of dis-
ordered dipoles. The rise in vibrational motions of molecules in a microwave
field is determined by the mechanism of ionic conduction and by dipole rota-
tion, that is, a rotation of the polar molecules under the influence of the exter-
nal electric field. The electric field applied to wood at microwave frequencies
Basic Aspects 127
interfacial
~c
8 ----------
()
.~ dipole rotation
~
:0
------------------
atomic
electronic
frequency
power audio radio infra-red optic
Fig.4.1. Relationship between the frequency and the dielectric constants of solids. (Skaar 1988,
with permission)
induces a dissociation of molecules and a migration of ions, by delivering

ordered kinetic energy. The collisions of the migrating ions with un-ionized
molecules produce heat. The ordered energy is then converted into heat that
is a disordered kinetic energy produced by collisions of the migrating ions with
un-ionized molecules. A continuous dipole rotation can be obtained by an
external alternating electric field. Each time the field is reversed the polar
molecules reverse their position. This transfer of energy leads to an increase
in the temperature of the solid. Dipole rotation is dependent on temperature
and frequency.
Dipole polarization also called the Debye effect depends strongly on several
wood physical parameters, such as species, density, moisture content, temper-
ature and structural orientation versus the anisotropic axes. The relationship
between the dielectric constants in a dielectric material containing permanent
dipoles subjected to a sinusoidally varying electric fields of strength E, are
shown in Fig. 4.2. The dielectric displacement D has two components, the first
noted D' in phase with (Ot, where the angular frequency (0 = 2n f and t is the
time and the second noted D" lagging (Ot by nl2 radians.
The dielectric constants f' and f" are the components of the complex dielec-
tric constant f*. They are defined as:
f'=D'/E* (4.1)
['=t:*sin (wt) Fig. 4.2. Relationship between the complex dielectric displacement,
dielectric constants phase angle e and loss tanD angle. (Skaar 1988,
with permission)
D' - - - - - - - - D*
I
I
I
I
[' I
I
I
I
I
I
I
I
I
I
I
L--_.I.-----',---;;..!
[" D"
and
E" = D"/E* (4.2)
where the strength field E = E* sin rot and E* is the peak amplitude of the sinu-
soidally varying electric field. In complex notation, we have:
[E *] = [E'] - j[E"] (4.3)
In other words, in an alternating electric field the amount of energy that
can be stored by wood is expressed by its dielectric constants [E*], which are
tensors and complex constants composed of a real constant [E'] and an imag-
inary constant [E"] and are related to Eq. (4.3). The imaginary constant [E"]
expresses the loss of the kinetic energy that can be dissipated as heat and is
also called the constant of dielectric absorption. The physical meaning of the
imaginary part of the dielectric constant [E"] is related to the attenuation of
the wave and can be correlated with the moisture content (Tiuri et a1.l980).
The ratio of real and imaginary constants E"/E' is called tan () and is currently
used in laboratory measurements.
It is generally accepted that wood is an orthotropic material and the dielec-
tric tensor [Eij] corresponding to this symmetry (Lin 1973) can be written as:
[Eid = [~:: ~:: ~::l (4.4)

E31 E32 E33
Each term of the tensor is a complex number that can be written with a real
and an imaginary part.
If the principal anisotropic directions (1, 2, 3 or L, R, T) coincide with the
direction of the electrical field, then the tensor [Eij] is written as:
Basic Aspects 129
Ell o
[ (4.5)
[E] = ~
o
and
tan Oil o
[tan 0]=
[
~ tan 0 22
o tan~J (4.6)
Out of the principal directions of elastic symmetry, the dielectric constants

are expressed by Eqs. (4.5) and (4.6):
(4.7)
and
(4.8)
where a L, aR, aT are the projection angles on the axes L, Rand T.
It was demonstrated by James and Hamil (1965) that the coefficients in
Rand T directions are very similar (Table 4.1). Consistent results were
obtained by Chazelas (1991) for spruce at a constant moisture content at 12%
(Table 4.2).
When the transverse elastic symmetry of wood is considered, the values of
the complex dielectric constants (diagonal and off diagonal terms of dielectric
tensor) are given in Table 4.3. The off diagonal terms of dielectric tensor
express the spatial distribution of the depolarization in wood.
The linearly depolarized field E tilted to 45° versus the direction of the fibers
is expressed as:
[E/I]
[E] = E: (4.9)
Table4.1. Dielectric properties of Douglas fir at 8.53 GHz, as a function of structural orientation
and anisotropic directions for different moisture content. (James and Hamill 1965; courtesy of
Forest Products J)
Moisture content Direction L Direction R Direction T

(%) E TanO E TanO E TanO
7 1.9 0.14 1.7 0.07 1.8 0.09

10 2.1 0.17 1.9 0.10 1.9 0.11
12 2.6 0.22 1.9 0.11 2.1 0.13
16 2.9 0.26 2.1 0.18 2.3 0.21
22 4.2 0.45 2.6 0.26 3.0 0.25
Table 4.2. Dielectric constants at 9.6GHz as a function of wood anisotropy for spruce at 12%
moisture content. (Chazelas 1991, with permission)
Specimen Thickness Density E'L E'R E'T £"L e"R E"T

(cm) (kglm 3 )
LR6cm 6.13 471 2.51 1.88 0.52 0.22

6.13 466 2.46 1.87 0.51 0.21
6.12 462 2.44 1.87 0.50 0.21
LT6cm 4.16 462 2.36 1.94 0.46 0.23
6.07 451 2.22 1.85 0.42 0.19
6.07 455 2.20 1.85 0.39 0.17
LR4cm 3.98 468 2.23 1.78 0.44 0.18
4.47 465 2.36 1.83 0.49 0.18
3.98 458 2.31 1.77 0.42 0.18
LT4cm 3.98 462 2.46 1.97 0.46 0.25
4.47 494 2.36 1.95 0.51 0.23
4.48 491 2.38 1.95 0.50 0.21
LR3cm 2.51 475 2.30 1.86 0.47 0.21
3.02 467 2.43 1.84 0.47 0.22
2.00 460 2.34 1.90 0.41 0.19
LT 3cm 2.01 461 2.37 2.01 0.36 0.20
3.01 495 2.47 1.94 0.48 0.25
2.51 495 2.32 1.96 0.48 0.25
Table 4.3. Dielectric complex constants of Douglas fir at 4.8 GHz frequency and 6 and 12% mois-
ture content, calculated with the hypothesis of transverse anisotropy with E22 = E 33 • (James et aI.
1985; courtesy of Forest Products Society)
Moisture content (%) El1
6 2.35 - j 0.15 1.9 - j 0.09 0.0018 + j 0.0075 0.0027 - j 0.001

12 2.73 - j 0.07 2.16 - j 0.031 0.0052 + j 0.0078 0.0123 + j 0.018
where Ell is parallel to the fibers and E.l is perpendicular to the fibers. The depo-
larization is expressed by the ratio D = EII/E.l called also "polarization ratio"
and can be calculated using the Stokes parameters defined as:
D = (Q2 + U 2 + V 2)1,I2 I - 1 (4.10)
where Q, U, V and I are given by the following equation

Q=EIIE*II-E.lE*.l (4.11)
I = EIIE *11 + E.lE *.l (4.12)
U = EIIE *.l + E.lE *11 (4.13)
V = -j(EIIE *.l + E.lE *II) (4.14)
where * denotes the complex conjugate of the quantities. The values of D range
between 0 and 1 [0 ~ D ~ 1]. When D = 1, the wave is completely polarized.
Basic Aspects l31
To express the anisotropic dielectric behavior Togovnikov (1990, 1993)

defined the coefficients expressed by Eqs. (4.15), (4.16), (4.17), for the case of
the transverse anisotropy:
k = £'''£'1-
j (4.15)
k2 = tan o"tan 01- (4.16)
k3 = E';,£''t (4.17)
The coefficients k k2 and k3 are given in Table 4.4 as a function of the fre-
j ,
quency. The increase in frequency determined the decrease in all coefficients

for both softwood and hardwood species. The modification of dielectric
properties with frequency in polar dipole materials is caused by the interfacial
polarization also called the Maxwell-Wagner effect. We shall see later that the
dielectric properties of wood depend in a very complex way on several factors
such as the frequency and power of the electric field, the moisture content of
the specimen, the temperature, the density of the species and the structural
orientation on which the constants are measured.
According to reference books in physics (Hippel 1961; Kittel 1968) for
dielectric materials, the parameters that can be measured are the permittivity
(the complex dielectric constants), the conductivity, the capacitance, and the
susceptibility of the materials.
It is beyond of the purview of this book to refer in detail to all these
parameters.
The dielectric response of wood material in terms of complex capacitance
is expressed by Eq. (4.18)
C * (0)) = C'(O)) - j C"(O)) (4.18)
where C' (0)) is the real part of the capacitance being out of phase with
the applied voltage and C" (0)) is the imaginary or loss component being in
phase with the voltage 0) is the angular frequency. The loss component can be
written as:
Table 4.4. The coefficients of dielectric anisotropy kJ> k2' k3 as a function of frequency, at 20°C,
where k, = E'I/E'.c, k2 = tanol/tano.c and k3 = E"I/E".c, for wood density ranging between 200 and 1000
kg/m3. (Torgovnikov 1990)
Species Frequency kHz
102 103 104 105 106 107 108 109 10'0 10"
Softwood k, 1.60 1.54 1.50 1.46 1.42 1.38 1.32 1.20 1.16 1.14
k2 1.60 1.70 1.75 1.80 1.90 1.80 1.70 1.50 1.30 1.25
k3 2.6 2.6 2.6 2.6 2.7 2.5 2.2 1.8 1.5 104
Hardwood k, 1.38 1.36 1.32 1.28 1.26 1.22 1.16 1.14 1.12 1.11
k2 1.35 1.50 1.50 1.55 1.75 1.65 1.55 lAO 1.30 1.25
k3 1.90 2.00 2.00 2.00 2.20 2.00 1.80 1.60 1.50 lAO
C"(ro) = EoE"(ro)(A/d)+G/ro (4.19)

where Eo is the free space permittivity, A is the area of the sample, d is the thick-
ness and G is the conductance of the sample.
The relationships between the dielectric constants, the permittivity, the
capacitance and the susceptibility of the material are given by Eqs. (4.20) and
(4.21):
C*(ro) = Eo (A/d) [E'(ro)-i E"(ro)]-i G/ro (4.20)
and
C*(ro) = Eo (A/d) [X'(ro)+E(oo)-iX"(ro)]-i G/ro (4.21)
where X'(ro) and X" (ro) are the real and imaginary parts of the susceptibility
and E (00) is the permittivity at infinite frequency.
In recent years, modeling microwave nondestructive testing procedures of
dielectric materials have received new attention because of their suitability for
nonmetallic composites (Poliszko and Hoffmann 1985; Ida 1992; Lebowitz,
Eoughi 1996; Thompson 1996) and ceramics (Schneider et al. 1996). The
models must take into account the wave propagation effects and a combina-
tion of induction and propagation phenomena. For accurate modeling of the
existing environment, Ida and Wang (1996) proposed two models that take
into account the anisotropy of dielectrics, assuming that the media are linear.
A thorough discussion of these models is beyond the purview of this book,
but we can note that the proposed methods can be used for monitoring
production processes.
Another parameter that can be measured is the power of the electric field
dissipated in the specimen as heat. It can be calculated with Eq. (4.22):
P = 21t ro E2 E'tano (4.22)
where: ro is the angular frequency and E is the electric field strength.
Before discussing the principles of defect detection using microwave
techniques it is important to consider first some factors related to the experi-
mental variables and also to analyze the influence of the physical properties of
wood on dielectric constants.
4.2.1 Effect of Experimental Factors
During the propagation of electromagnetic waves in wood, the initial

parameters of transmitted waves are modified by interaction with the wood
structure. As noted by King (1978), the main wave parameters that interact
strongly with wood density and moisture content are the amplitude and the
phase. The complex propagation parameter (y) of a steady-state plane wave can
be written as:
Basic Aspects l33
y=a+j~ (4.23)
where a is the attenuation constant (Nepers/m) and ~ is the phase constant
(radians/s). After corresponding calculation, for tan8« 1, it was conclude that:
a=0.5~oRtan 8 (4.24)
and
~ = ~o R. [1 + 0.125 tan 2 8] (4.25)
where ~o is the phase constant of air, defined by
~o = 2TC/A o (4.26)
where AO is the wave length of the electromagnetic wave in air.
In the case of wood material, it has been demonstrated (James 1975; Wert
et al. 1984; James et al. 1985) that £' depends strongly on wood density and
tan 8 on moisture content.
The loss of wave intensity is reflected by the reduction of signal amplitude
and by the phase retardation that are caused by the absorption of wave energy
and by scattering. In addition, the anisotropic structure of wood induces
depolarization of the incident wave. The electric vector of the incident wave is
decomposed into two components in the directions of the maximum and
minimum dielectric constants. When the wave is elliptically polarized, the
minor axis of the ellipse corresponds to the maximum dielectric constant,
which coincides with the directions of the fibers. The polarization is a func-
tion of the thickness of the specimen. Perfect elliptical polarization is obtained
only when the specimen thickness is optimum. For lesser thicknesses the
polarization angle is smaller than the true slope of the grain by a factor cor-
responding to the depolarization index, as can be seen in Fig. 4.3.
The effect of density on the phase and on the attenuation of waves was
demonstrated by Kharadly (1985) for hemlock (Fig. 4.4) at 11-12% moisture
content. The corresponding regression equations are given in Table 4.5. For
both parallel and perpendicular polarization, an increase in the density
increased both the attenuation and phase shift. These experimental data were
used for the abacus presented in Fig. 4.5, from which it is possible to determine
both wood density and the moisture content of a board by measuring the atten-
uation and the phase shift of a microwave regression equation (r = 0.87) of the
phase shift for different species as a function of the density, ranging from 400
to 800 kg/m3 (Fig. 4.6) at 9% moisture content as was reported also by Martin
et al. (1987). The effect of moisture content on the attenuation, phase angle and
depolarization index was studied by James et al. (1985) with Douglas fir as can
be seen in Figs. 4.7, 4.8 and 4.9. Increasing moisture content caused an increase
in all these parameters.
Values of attenuation coefficients for Indian wood species with nearly 9%
moisture content, as reported by Jain and Sanyal (1996), are given in Table 4.6.
The measurements were performed at 8.2 GHz in the X band microwave region
/ I
X / /
/ / / / grain direction
/ ~
I /
/ /
z
I
a) -t-
X' X
-.....
/ e
I ~
'"
L.h.pol.f
\ \ ,~
E inst
y , ~ain direction
b) \
Fig. 4.3. Slope of grain and the elliptic locus of the instantaneous electric field. a Grain direction
at angle 0 in the plane Y -Z in a specimen of thickness t. b Elliptical polarization of transmitted
instantaneous field. The ellipse is inclined at angle e; the sense of rotation depends upon whether
the X-field component leads or lags the Y-field component in the time phase. The locus becomes
a circle when the two components coincide grain direction and instantaneous E inst. (King 1978;
courtesy of Forest Products Society)
on an ElOmode of vibration. Figure 4.10 shows the linear increase of attenua-

tion with density in the range of 400 to 500kglm3. At a density greater than
510 kg/m3, the attenuation appears to be independent of the density.
purslow (1971) reported attenuation measurements at lOGHz for three
species: redwood, hemlock and whitewood, as a function of moisture content
in the range of 6 to 30%. He stated that the relationship between the moisture
content and microwave attenuation is sigmoid, with three main zones, namely,
zone I between 0 and 8%, zone II between 8 and 30% and zone III over 30%.
Basic Aspects 135
In the first zone, the attenuation increases very slowly and is probably due to
the macromolecular water being very strongly bonded to the wood structure.
In the second zone, corresponding to the presence of "bound water", the atten-
uation dramatically increases with moisture content. Over 30% "free water" is
Table 4.5. Relationships between the variation of amplitude (~A) and of phase (~cp) of the
microwave signals and density (p) of hemlock wood samples at 20°C and with the moisture
content M, constant. (Kharadly 1985; courtesy of Forest Products Society)
Polarizatioin Regression equation Correlation coefficient Standard deviation
Parallel ~cp = 38 + 540.73p 0.95 10.92

~A = 2.1 + 21.16p 0.81 1.02
~cp = 44.51~A 0.87
Perpendicular ~cp = -25.52 + 497.81p 0.93 12.36
~A = -1.07 + 15.67p 0.84 0.67
~cp = 52.19p 0.90
parallel polarization
HEMLOK (# 201 300)
MOIST. CONT.=11-12%
+ NORMALIZED PHASE SHIFT +
-..•
+
«+
XNORMALIZED ATTENUATION ..... +
APHASE SHIFT / ATTENUAT~"
0=0 . ~~
;= 540,7 P + 38,0 x
f >s<~x ~'S<
• x~~)( XX
i{Jx
+ )t~x
x ~
X
c
c4---~--·--,-----;----r----;----r----.----r----.----+1
0,0 0,2 0,4 0,6 0,8 1,0
a) density (glem 3 )
Fig.4.4a,b. Attenuation and phase shift as a function of density. (Kharadly 1985; courtesy of
Forest Products Society)
+
0 perpendicular polarization
-
0 0
0" 0 .+ 0
....
('.l 0 ('.l
M
+
+NORMALIZED PHASE SHIFT •
-
XNORMALIZED ATTENUATIO~
~o" 0
--:--0. .
~HASE SHIFT / A TTENUATI~ +
o '""'
/
~g 1C5~
0"""
~('.l '-'
~- • -90
Cl
~
; - 497,8p + -25,5 ~
-~ -
... f::
[-<
[-<0 ::C: o
0 ++ 0 ~
sf [-<a'i
00
}i.+
<0 ~oo
~oo
00 -
a'i
[-< ~
[-<
<
[-< ~
.".. xxX Cl
< Cl ~
~o
---
[-< 0" ~o + ~ 0
c.o
~
....:I
~~
... 0
<
EeI.O
00
~
X ~
0
00 0 Z
Z
~o
~o
0
0" 0
"""
1.0
X """"
•
0
0
0 0
0
('.l 0"
0,0 0,2 0,4 0,6 0,8 1,0
b) density (glcm 3 )
Fig.4.4a,b. Continued.
Table 4.6. Attenuation of microwaves at 8.2GHz frequency measured on several Indian species.
(Jain and Sanyall996, with permission)
Species Density (kg/m3) Moisture content (%) Attenuation constant"
Abies pindrow 390 9.6 0.321

Jacaeanda acutifolia 440 9.5 0.516
Toona ciliata 480 9.2 0.708
Alibizia lebbek 600 9.4 0.798
Dalbergia sissoo 630 8.3 0.804
Acacia catechu 850 8.9 0.819
"The attenuation constant was calculated with the relationship; a = 2.303 X2-'. (lOg1O F2-log1O F,)
(4.27), where F2 and F, are the corresponding intensities of microwaves for wood samples. The
measurements were performed in the X-band region of microwaves on the E10 mode of vibration.
Basic Aspects 137
¢:::
:.a PI
VJ
C)
VJ
o:l
..c
P2
0-
"0
C)
<-<
;:::I
VJ
o:l
C)
E P3
P = density
P4 M = moisture content
constant M lines
measured attenuation
Fig. 4.5. Abacus for the determination of unknown density and moisture content as a function
of variation of phase and amplitude of the microwave signal for hemlock wood samples at 20 0 •
(Kharadly 1985; courtesy of Forest Products Society)
present, and the slope of the regression relating the attenuation and moisture
content is less steep than in the second zone. The fiber saturation point was
found to be between 26.9 and 28.2%, as the intersection between the regres-
sion lines in the second and in the third zones.
4.2.2 Effect of the Physical Properties of Wood
The basic physical parameters of wood related to the detection of defects with
a microwave technique are: the moisture content, the density and the grain
angle (the slope of grain). These three parameters can be determined from the
dielectric tensor. The relationships between the diagonal terms of the dielectric
tensor and the moisture content, density, grain angle and chemical constituents
o
o
0\
o
o
00
0
0
f"-
6-
<t::
:E
'" °0\0
II)
gj
..<:: c
~
<., c
0
0
c p=107 b=-1,96° r=0,87
11"\ c
c
°0 c
"""
0
0
""0,3 0,4 0,5 0,6 0,7 0,8 0,9 3 3
X 10 kg/m
Density
Fig.4.6. Relationship between the phase shift and the density of different wood species,
with density ranging from 400 to 800kglm3 • (Martin et al.1987; courtesy of Wood Science and
Technology)
0
N
e )
--~
0
......
11"\
~
t;
I':
0
0
0
I': ......
0
.~
6
~
o Parallel
• Perpendicular
Fig.4.7. Attenuation of the signal as
I I I a function of moisture content in
50 100 150 two anisotropic directions, parallel
and perpendicular to fibers. (James
et al. 1985; courtesy of Forest Research
Moisture content (%) Society)
Basic Aspects l39
• Fig. 4.8. Phase shift as a function of mois-

0
ture content in two anisotropic directions,
0
N parallel and perpendicular to the fibers.
(James et al. 1985; courtesy of Forest
au Products Society)
Oil
11)
"0
~
§
t'i
.: 0
0
u 0
11)
~
0;
..<:
0...
o Parallel
• Perpendicular
I , , , , I I , ,
50 100 150
Moisture content (%)
0
0
.-
5",
~t--
;.<
11)
"0
.5
':0
.9 '"
(;j
N
'ta
"0
0..
11)
"0",
.--=: N
.:
=:J
Fig. 4.9. Depolarization index as a
function of moisture content in two
0
anisotropic directions, parallel and
a 50 100 150 perpendicular to the fibers. (James et al.
Moisture content (%) 1985; courtesy of Forest Products Society)
have been studied by many investigators. Values of dielectric constant mea-

surements for various species and at various frequencies (ranging from 1 to
more than 20GHz, and also in the low frequency range from 10-2 to 106 Hz) at
different temperatures, moisture contents and anisotropic directions, were
reported for different wood species, originating in the temperate zone in
,
__+__
I
I
._._~_-G
00
6
t-
O' .,.,'"
§
<:!
<Il
0..
-
"1
'T
,
e 0()
""II
0
<>
e
.Sl
'"o· 'T'
ta <:!
;:l
is
~
.,.,
6
<')
6 300 400 500 600 700 800 900 density (kg/m3)

Fig. 4.10. Attenuation as a function of density for different Indian tree species with density
ranging from 400 to 900kg/m3 • (Jain and Sanyall996; with permission)
Europe, USA and Japan, and also from the tropical zones in India and Malaysia
(Hearmon and Burcham 1954; Uyemura 1960; James and Hamill 1965; Lin
1967, 1973; Norimoto 1976, James 1975; Norimoto et al. 1978; Peyskens et al.
1984, Chazelas 1991; Choffel et al. 1992; Goy et al. 1992; Jain and Sanya11996;
Obataya et al. 1996; Dubey and Deorani 1997; Kabir et al. 1998; Greaves 1998;
Zielonka and Dolowy 1998).
For brevity and simplicity, from this important literature related to the
dielectric constants of wood, the discussion will be limited to considering the
relationship of the dielectric constant to the moisture content, to the structural
orientation that induces anisotropy and to wood density.
Basic Aspects 141
4.2.2.1 Effect of Moisture Content
The effect of moisture content on the dielectric constant of Douglas fir at

3 GHz was reported by James and Hamill (1965). The dielectric constant E',
both parallel and perpendicular to the grain, increased continuously with
moisture content in the range from 8 to 75% (Fig. 4.11). This relationship was
considered linear below the fiber saturation point. The constant tan 0 behaved
differently, increasing rapidly at low moisture levels to a maximum value at
fiber saturation. A concave downward trend of tanO was observed with increas-
ing moisture content in the range of 25 to 75%.
For oven-dry wood Norimoto and Yamada (1972) reported that E' of
Chamaecyparis obtusa varies with the grain angle, as can be seen from Fig.
4.12. In the longitudinal direction the E" value is approximately twice its value
in the radial direction. Very little difference was observed between the e' in the
radial and tangential directions.
Dielectric properties of three softwood species, pine, hemlock and spruce,
were studied by Peyskens et al. (1984) below the fiber saturation point. It was
observed that e' and e" increased with moisture content from 5 to 35%, for all
anisotropic directions (Fig. 4.13). The curve corresponding to tan 0 reached a
0,36 18
0,32
• 1.
16
/ ~ ~In
.....
0,28 / ~ll 1i". V 14
V ~ V W
I ~ /'
-r-" " >

0,24 12
I // §
, t:i
1::
~ 0,20
If ,,)Sl> rse
" 10
I'l
0
V >'
--
OJ
§ .~~ r-.
-"'~ r
.~
] 0,16 I / • V 8
I I{ ~~ V ~"oe V ~
:.a
<l)
0,12 / c~~ V -, ~ 6
• l} _u<, ~
.,co~": ~
0,08
,
¥
~~
/
C1"
l..cf"
/
;r
~pV ._\, ~
~
4
o
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
moisture content ( percent of dry weight)
Fig.4.11. Dielectric constant £', and tan/) of Douglas fir as a function of the moisture content.
(James and Hamil11965; courtesy of Forest Products J)
3
LT
. LR RT
/
....... ---~O
/
/
/
0/
" "- "-
2 r '0
' -
w y-....
........ 0/ .....
.....-C - - - - 0 - ---0--_...,
T J J L L I R R i 1 T
90 60 30 o 30 60 90 60 30 o
grain angle ( deg ) ring angle ( deg )
Fig.4.12. Variation of the dielectric constant in three anisotropic planes. (Norimoto and Yamada
maximum between 20 and 25% moisture content. The increasing values of

tan 8 with a moisture content <25% are probably due to two factors:
- the increase in water within the wood matrix, since water is characterized
by high dielectric values
- the polar components of the cell wall and the cellulose itself acquire more
freedom of rotation at higher moisture contents and contribute to a more
pronounced dielectric behavior.
When the moisture content of the specimen increases in the range >25%, in
the vicinity of the fiber saturation point, the polar groups in the cell wall do
not increase any more because their freedom of rotation is at a maximum and,
consequently, the curve of tan 8 that is indeed E"/E', decreases.
The strong effect of the moisture content on the dielectric constant suggests
that a specific drying pattern can be obtained with microwave energy. High
moisture content zones with high values of E" above the fiber saturation point
will dry first, compared with the surrounding zones containing less moisture.
This is in contrast with conventional drying methods in which the surface dries
more rapidly than does the moist interior.
4.2.2.2 Effect of Anisotropy
The effect of anisotropy can be expressed as a ratio between the dielectric con-
stants, as demonstrated by Makoviny (1988) for two frequencies and four
species: spruce, pine, beech and oak (Table 4.7). For all species it was noted
that E'L > E'R > E'T. In the transverse plane, the ratio E'R/E'T ranges from 1.02 and
1.08 for softwoods and between 1.13 and 1.16 for hardwoods. This means that
from a dielectric point of view the softwood species are less anisotropic than
Basic Aspects 143
6
2'
L,R, T ;
L
5
4 T
R
3
m.e. (%)
1+-----,------r----~----~~----r_----~-
a) 5 10 15 20 25 30 35
2"
2.0 L,R,T:
L
1.5
.. , T
1.0
•. ' R
0.5
m.e. (%)
O~O •• -
b) 5 10 15 20 25 30 35
tan 8
0.4
L, R,T ;
0.3
'" L
.. ' T
R
0.2
0.1
..'
..' m.e. (%)
O.O~----,_----_r----~----_,------r_----~-
C) 5 10 15 20 25 30 35
Fig. 4.13a-c. Dielectric constants as a function of moisture content and anisotropic direction in
pine. a E' as a function of moisture content; b En as a function of moisture content; c tano as a
function of moisture content. (Peyskens et al. 1984; courtesy of Wood Science and Technology)
Table 4.7. The anisotropy of wood expressed by the dielectric

constants of four species (spruce, pine, beech and oak) with
densities ranging from 416 to 677kglm3 • (Makoviny 1988)
Frequency Anisotropy Spruce Pine Beech Oak
1 MHz E' L/E'T 1.72 1.54 1.51 1.42

E'R/E'T 1.08 1.08 1.18 1.14
20MHz E'L/E'T 1.54 1.56 1.33 1.35
E'R/E'T 1.02 1.06 1.16 1.13
Table 4.8. Attenuation coefficients of microwaves at 10 GHz as a function of sample thickness and
of the anisotropic direction. (Chazelas 1991, with permission)
Range of thickness Thickness· (mm) Attenuation in L Attenuation in T

direction (Neper/m) direction (Neper/m)
From 28 to 20 mm 27.9 48.21 26.07

25.15 47.72 25.97
23.11 51.59 29.55
19.57 51.78 28.32
From 30 to 22 mm 30.41 60.33 34.02
27.43 60.41 34.65
24.79 64.78 34.69
22.10 65.35 37.97
• To avoid the effect of wood structure on the measurements, the thickness of the sample was
progressively reduced.
hardwood species in the transverse plane. The influence of the frequency is

more important for the ratio E'JE'T than for E' RIE' T' The highest anisotropy was
observed at 1 MHz in spruce, for which E'JE'T = 1.72 and the lowest anisotropy
was found for beech, for which E'JE'T = 1.33 at 20 MHz. This behavior was
explained by an electrical model of the annual ring.
The effect of anisotropy on the attenuation coefficients as a function of
specimen thickness for spruce is shown in Table 4.8. The samples with a thick-
ness of 30 to 22 mm have a greater attenuation than the 28- to 20-mm samples.
Increasing the thickness tended to increase the attenuation coefficient.
The effect of wood anisotropy on the dielectric constants of a large number
of softwood and hardwood species was extensively studied by Norimoto and
his coworkers at Kyoto University, Japan (Tanaka et al. 1975a,b; Norimoto et al.
1978; Nishino and Norimoto 1990). This research indicates that the dielectric
constants in the longitudinal direction are higher than those in the transverse
plane. Norimoto et al. (l978) noted that this behavior results from the arrange-
ment of the cell wall and lumen, which is in agreement with Lin's (1967) state-
ment that the hydroxyl groups of the cellulose should have more freedom of
Basic Aspects 145
rotation in the longitudinal direction than in radial and tangential directions.

The study of the effect of anisotropy on £" clearly demonstrated that during
drying the absorbed power can be increased by an optimum orientation of the
specimens versus the microwave antenna.
4.2.2.3 Effect of Density
The effect of density on the dielectric properties of wood was extensively

studied by Skaar (1948), Rafalski (1966), Lin (1973) and Vermaas (1973). Skaar
(1948) showed that the density has a consistent effect on the dielectric con-
stants of different species. He noted that the low variability of density within
a single species minimizes the variability in dielectric constants among species.
Rafalski (1966) observed a linear effect of density, ranging between 700 and
1400kg/m3 on £" and tanO.
Lin (1973) studied only one species, western hemlock, using a stepwise
analysis of the logarithmic values of dielectric permittivity, and resistivity as
they varied with moisture content and density. He found that the moisture
content alone accounted for more than 94% of the variability in dielectric con-
stants. Incorporation of density as an additional variable improved the regres-
sion coefficient only by several percent. Vermaas (1973) established regression
equations of the dielectric constants of Pinus pinaster versus density, moisture
content (0; 5; 10; 15 and 20%), frequency (10,30 and 50 MHz), grain direction
and extractives (resin content). The proposed models of regression equations
contained linear, quadratic, and third power terms as well as first order inter-
action terms. He concluded that the dielectric constant £' for all directions
increased with increasing density and moisture content and decreased with
increasing frequency. The loss tangent for all anisotropic directions increased
with increasing density, frequency and moisture content from 0 to 8%. From 8
to 20% an important decrease was observed only in the longitudinal direction.
The extractives had no influence on either £' or the loss tangent. As can be seen,
the effect of density was not clearly demonstrated for all species. The sorptive
capacity and the permeability of each species seems to be more important than
the density for the dielectric properties of wood.
From all dielectric measurements, the parameter able to express in a better
way the molecular structure of wood is the dielectric loss (Norimoto and Zhao
1993; Yokoyama and Norimoto 1996, 1997; Yokoyama et al. 1999, 2000a). The
dielectric relaxation process induced by the motions of methylol groups in the
amorphous zones of the cell wall was demonstrated by Yokoyama and
Norimoto (1996) with absolutely dry spruce wood, within a wide range of
frequencies and temperatures (-160 to +20°C). It was found that the maximum
dielectric loss in the relaxation process was around 10 MHz. The iso-dielectric
loss parameter was calculated for hinoki, from the Cole-Cole equation, as a
function of frequency and temperature. The effect of relative humidity on
relaxation phenomena was demonstrated by varying the relative humidity, in
a) O ~~---r--~---'
0.1
-50 0.Q75
G -100 l--_0.0_5~
~
~ O%RH 1J %RI-J
.g~ -I 500 ~==:::;::===;;::=:::;;:==~
r-
0.1 0.1
Oo
E 0.2
B
-50
0.15
-100 , --~ 0.025
33%RH 43%RH
-\50 + - - - - - - - -- ---'
b) 0-
-50
85 %RH
- 150 3 4 5 6 3
logarithm of frequency
Fig.4.14a,b. Iso-curves of the dielectric loss as a function of temperature and frequency for the
relative humidity, ranging from 0 to 94%, calculated for hinoki. (Yokoyama et al. 2000a; courtesy
which the wood samples were tested at constant temperature (Fig. 4.14). The
distribution of relaxation times narrowed with increasing relative humidity.
4.2.2.4 Effect of Chemical Constituents
The effect of the chemical constituents of wood such as crystalline cellulose,

hemicelluloses and lignin, on the anisotropy of hardwood dielectric constants,
Basic Aspects 147
was studied by Nishino and Norimoto (1990). The theoretical development was
based on models in which the proportion of chemical constituents of wood
was deduced from the rule of mixtures. The dielectric constants of the princi-
pal constituents are given in Table 4.9. These values were used for estimating
the dielectric constants of the cell wall shown in Table 4.10. Closer examina-
tion of these data and of a simplified model of the anatomic structure of hard-
wood leads to the theoretical values of the dielectric constant £' as a function
of moisture content in the range 0-15%, for three anisotropic directions. The
estimated values in the longitudinal direction were higher than those in the
radial and tangential directions and were in good agreement with data
measured with mizume (Betula grossa).
Table 4.9. Theoretical values of dielectric constants of chemical constituents of wood. (Nishino
and Norimoto 1990, with permission)
Moisture E, Noncrystalline E2 Noncrystalline Lignin £p for EM for

content (%) cellulose in cellulose in pentosan mannan
direction 1 direction 2 matrix
0 8.21 8.00 4.04 4.01 4.02

10 10.73 10.51 4.45 6.52 5.46
20 12.87 12.63 4.70 8.64 5.56
30 14.99 14.73 4.95 10.78 7.63
40 17.48 17.24 5.25 13.28 8.86
50 20.64 20.37 5.80 16.49 10.52
60 24.02 23.74 6.30 19.88 12.19
70 28.32 28.00 7.00 24.15 14.37
75 30.58 30.33 7.30 26.48 15.49
Table 4.10. Calculated dielectric constants of cell wall of hard-

woods in longitudinal (ELow ) and transverse direction (ETcw), calcu-
lated for a microfibril angle of 13°. (Nishino and Norimoto 1990,
with permission)
Moisture CLew tTcw

content (%) In longitudinal direction In transverse direction
0 5.60 4.68
10 6.42 5.84
20 7.54 6.72
30 8.62 7.54
40 9.90 8.49
50 11.58 9.72
60 13.34 10.97
70 15.65 12.60
75 16.87 13.64
4.3 Equipment for Dielectric Measurements and for Microwave

Imaging Technique
The technique of microwave imaging of materials was initiated in 1980 with

the development of microwave holography and of radar and remote sensing
applications (Bolomey and Pichot 1990). During the last 10 years, microwave
imaging of penetrable objects in complex environments has been shown to be
a complementary technique to X-ray and ultrasonic tomography.
The recent development of microwave imaging was made possible because
of three factors:
- the development of probing devices of greater flexibility and accuracy at a
reasonable cost
- advanced wave front processing related to the enhancement of image quality,
considering simultaneously the spatial and contrast resolution and the
solution of theoretical problems of inverse scattering.
- the relevant technological progress in microwave and computer
technologies.
The algorithms for microwave imaging are similar to those for ultrasonic
tomography. After the amplitude, phase and polarization probing a of wave
front or of a scattered field have been obtained in the transmission or in the
reflection modes, different procedures can be applied in order to obtain infor-
mation about the structure of an object. The image can be obtained by the
direct use of projection. (Fig. 4.1Sa). The expansion of the scattered field during
propagation is determined by the precise localization of the target in the trans-
verse plane. The domain of validity for the direct use of projections is restricted
b) incident field
surface probing
slice 2
Fig.4.1 Sa,b. Microwave imaging of an object. a With projection produced by direct measure-
ments; b with tomography obtained by numerical processing of the wave front. (Bolomey and
Pichot 1990, with permission)
Equipment for Dielectric Measurements and for Microwave Imaging Technique 149
to planar objects, for which the distance of observation must be within one
wavelength.
In contrast to projection imaging, tomographic imaging is obtained with a
more elaborate processing of a microwave wave front and gives a cross-
sectional representation of the object (Fig. 4.15b). Either a space domain iter-
ative approach or a spectral formulation for diffraction tomography can be
used. In the latter case, using the Born approximation, the image is obtained
after taking the inverse Fourier transform. For inhomogeneous solids, the pro-
cedure of reconstruction is subjected to various artifacts due to diffraction
effects. The most promising approach for microwave imaging is the space
domain iterative formulation, which is very sensitive to the accuracy of local
measurements. The profile of the object is obtained from the complex permit-
tivity values. The domain of validity of this method depends directly on the
signal-to-noise ratio. For industrial applications, cost considerations are of
great importance and a compromise between the cost of the equipment and
the rapidity of the performance must be found.
In the following, we will analyze first the equipment for laboratory
measurements of dielectric constants of wood and, second, the equipment for
imaging the structure of wood during processing under industrial conditions.
4.3.1 Equipment for Laboratory Measurements of Dielectric Constants
There are two basic microwave techniques for dielectric measurements:

the transmission and the reflection methods. In the first configuration, the
dielectric material is located between the transmitting and receiving horn
antennas, while in the second case both antennas are located in front of the
sample.
The instrumentation for laboratory measurement of dielectric constants
and the loss tangent of wood is a resonant method, as described in ASTM D
150-74 (1986). The main components of an experimental device are:
- the source of continuous microwaves (Klystron microwave generator) that
produce the electromagnetic wave at relatively low power and high
frequency.
- the waveguide.
- two pyramidal horn antennas, one acting as a transmitter and the other as
a receiver The distance between the two horn antennas must be kept greater
than 2D2/"A, where D is the horn aperture and "A is the operating wave length.
- the specimen, located between the transmitter and the receiver horns.
- the detectors of the signal amplitude, phase and degree of depolarization.
The equipment for measurements of the variation of amplitude and phase
of the electromagnetic wave as described by Kharadly (1985) is shown in
Fig. 4.16a. The system works as a bridge circuit balanced in two steps, without
co-polar branch V1
o
wood specimen I I
-
s:::
;:;.
orthomode
transducer is
elliptically Ol
linearly OQ
polarized polarized S·
OQ
incident transmitted
a) wave (45°) wave
wood specimen
I microwave ~ orthomode
source transducer
linearly ~ elliptically
polarized polarized
b) wave wave
cross-polar branch
Fig.4.16a,b. Microwave systems for the location and detection of different parameters of wood (Kharadly 1985). a System for the
measurement of attenuation and phase shift for detection of wood density and moisture. b System for measurements of polariza-
tion of transmitted waves for the detection of the slope of grain.
wood specimen
l;n~c1y 'V\J'-.-~"VIf'-'
polarized
lin=ly
polarized
wave wave
(c)
(d) (a) strip line (b) split waveguide (c) free wave
Fig. 4. 16c,d. c System for the detection of the discontinuities introduced by the presence of knots
(courtesy of Forest Research Society). d Sensors for transmission measurements (Thompson
and with a sample, by adjusting the variable attenuator and phase shifter.
This system was designed to measure the moisture content and density of
wood.
The slope of grain measurements are based on the measurements of the
variation of the polarization of a transmitted wave that is elliptically polarized
by the wood structure (Fig. 4.16b). The orthomode transducer separates the
components of the transmitted wave into a co-polar component and a cross-
polar component, which are alternately sampled with a high-speed microwave
switch before processing of the corresponding signal by the detector.
For the detection of defects, such as knots, both the variation of density and
the slope of grain must be measured. For this purpose, a combination of the
previous systems can be used, as shown in Fig. 4.16c. The incident wave is
polarized at 45° versus the direction of the fibers, which generates an ellipti-
cally polarized wave with two components (the co-polar and the cross-polar),
which are detected separately. In this way, a typical response signature is
obtained, indicating the presence of the defect.
Based on a similar approach, Forrer and Funck (1998) used a dielectric spec-
trometer operating in a frequency domain at 10 MHz, for detection and 3-D
imaging of knots. The electrode of the spectrometer was 19.5 mm in diameter
and was mounted on a translation table. The scanning line was chosen to
large loss knot (LKL423), 10 MHz, 25%M

00 0, 16
r- 0, 14
~
V>
-0 0. 12
<:
8
<>
"., 0, 10
~
.;:
0,08 §
..,
<:t
U V>
~
...... 0,06 ..Q
V>
"Q
N 0,04
... dielectric constant ... loss tangent
0,02
0 0
0 I 2 3 4 5 6 7 8
distance along scan li ne (em)
Fig. 4.17. Signature of a loose knot expressed by the dielectric constants and by the loss tangent.
Above The location of the scanning line on a loose knot. Below Dielectric constants as a function
of the position on the sample. (Forrer and Funck 1998; courtesy of Holz als Roh und Werkst.)
~o
=-~
f~\0~'
~/(~,~~)'\~\
'''~
~:~~~~*\~\""
~d ' '; ;\ ~
~ ~~) ,
I ~ \ FigA.18. Three-dimensional images of a
tight knot 14mm in diameter. (Forrer and
Funck 1998; courtesy of Holz als Roh- und
Werkst.)
cover a zone of approximately 5 cm 2 in the grain direction with the electric field
applied through the specimen in the radial and tangential directions as can be
seen in Fig. 4.17. The location of the loose knot is precisely determined by both
the dielectric constant and the loss tangent. A 3-D image of a tight knot of
14mm diameter is shown with good resolution in Fig. 4.18.
The limitation of microwave through-transmission or reflection methods is
caused by the low sensitivity of this technique to differentiate between differ-
ent types of defects, as can be seen in Fig. 4.19. Forrer and Funck (1998) studied
the dielectric constants of knots, pitch pockets, clear wood, blue stain and open
holes as functions of moisture content and frequencies (Fig. 4.20). From this
grey levels
255
threshold
\
Fig.4.19. Microwave signals for different types of defects. (ChoffeI1999, with permission)
graph, it was impossible to define precisely a typical defect, but a general sep-
aration in three main groups (air gaps, pitch pockets and tight knots) occurred
with increasing moisture content and frequency. In a large cloud cluster of
data, open holes and loose knots are in opposite position to tight knots.
Between them, clear wood, blue stain, and pitch pockets can be observed.
Forrer and Funck (1998) noted that under laboratory conditions, the main
sources of errors associated with dielectric measurements of wood containing
defects is induced by the moisture content variation in the defect zone com-
pared with the area of surrounding sound wood and by the variability of the
signal introduced by the wood sensor interface.
4.3.2 Equipment for Online Imaging of Wood Structure
Microwave imaging is a quantitative nondestructive evaluation method for

characterizing surface and subsurface defects of a wood sample. The defects
can be imaged by scanning in the proximity of the samples with a resolution
that depends on the size of the aperture. The images can be obtained in the far
a) S! • knot
c) s:
- n;lrh ch
o :1""" 'ood
, I ilch lrea.k
pen Olcs
• blue . tain
MC=6% MC=6% MC=6%

f=1.4MHz f= IOMHz f=20MHz
,.
7-
~~:
. 11
l
,
'1'/
i r
~~
" J~
,v:.
if.:::l>
~
o O.os 0. 10 0, 15 0)0 0.25 0.30 o 0,05 0. 10 0.15 0.20 0,25 0,30 o 0,05 0.10 0,15 0.20 0,25 0.30
a loss tangent b loss tangent c loss tangent
d)=: • kno~
-
o clear ood
I ilch streal -
••
1 ppen o les
• blue lain
;; ""
,I' ,0' ,I' ~
~~
lI!J I~ ,\I'!J
.~" r+
,
M 25% t~r+ MC=2S% .1It· MC=2S%
f,!l f~ L4MHz _ I(IJ~ 10MHZ ~N
f=20MH2
N
r;i
N
1.)
N
ri I
:i! ,< '"
o 0,05 0, I0 0,15 0,20 0.25 0.30 h 0 0.05 0, 10 0, I5 0,20 0.25 0.30 0 0.05 0, I0 0.15 0,20 0.25 0.30
g loss langenl loss langen. loss langenl
g)=: I knOI
h) 0
-
o clear 'ood
I pitch Irca
1 open o les
• blue lain
~ I
~ Jt ~.
II
.2iI
1i~ ~~
,~ . ~
~~
' )Ii
i MC= IO% ;~ MC=IO%
_11 .~~A
MC= IO%
i' f=20MHZ
1-.,.. \.
f=1.4MHz Jfl:( : f= IOMHZ F'
:l1'li '
o 0,05 0, 10 0. 15 0,20 0.25 0.30 o 0.05 0.10 0, 15 0.20 0,2 5 0,30 o 0,05 0, I0 0,15 0,20 0.25 0,30
d los tangent e loss tangent f loss tangent
or in the near field. For far-field imaging the smallest detectable defect is deter-
mined by the ratio d/A >1, where d is the length of the defect in the plane
normal to the microwave propagating vector and A is the microwave wave-
length. This condition is known as the Rayleigh criterion, which is applicable
to all electromagnetic imaging in the far field. For near-field imaging, the
resolution can be as high as A/IOOO, because in this case Rayleigh's criterion
does not apply.
The probes used to evaluate the properties of the samples can be scat-
terometers or radars that measure the scattering properties of the medium, or
can be reflectometers that measure the reflectivity of the sample due to its
inhomogeneities and defects. The probes can operate in the proximity of the
medium as open-ended coaxial lines, cavity resonators, antennas, etc. By
moving the position of the transmitter across the sample, an image of the
structure can be obtained. The development of microwave systems for online
measurements of the moisture content in wood and wood-based composites
is directly related to the economic benefit in the sawmill and particle board
industries for front-end processing, in-process control and output quality
control.
For online wood processing, the microwave equipment must satisfy several
criteria, including:
- the recognition, location and mapping of defects, such as knots, reaction
wood, slope of grain and other major discontinuities of wood structure;
- the capability of operating over a wide range of temperature, wood mois-
ture content and density of different species, with sharp resolution;
- the normal speed of operation required by the processing;
- a non contact and nonhazardous technique, with no susceptibility to vibra-
tions and with simplicity of operation;
- inexpensive training for the operators.
The actual microwave equipment for the inspection of logs, lumber and
composite boards is discussed below.
The inspection of logs is performed with the experimental device
(Kaestner and Baath 2000) shown in Fig. 4.21. In front of a wide-band antenna,
the log is rotated and translated. The frequency band used for measurements
was between 4 and 8 GHz. More details about the experimental results obtained
Fig.4.20a-i. Relationships between the dielectric constant and loss tangent for knots, pitch
pockets, clearwood, pitch streaks, open holes and blue stain at moisture contents ranging from
6.6 to 25% and for frequencies ranging from 1.4 to 20 MHz. a Moisture content 6% and frequency
1.4 MHz. b Moisture content 6% and frequency 10 MHz. c Moisture content 6% and frequency
20 MHz. d Moisture content 10% and frequency I.4MHz. e Moisture content 10% and frequency
10 MHz. f Moisture content 10% and frequency 20 MHz. g Moisture content 25% and frequency
1.4 MHz. h Moisture content 25% and frequency 10MHz. i Moisture content 25 and frequency
20 MHz. (Forrer and Funck 1998; courtesy of Holz als Roh- und Werkst.)
w ide band antenna
----'x
n knot
b)
Fig.4.21a,b. Scanning oflogs with a microwave device. a Components of the device. b Refraction
indices at the interface between log and knot. I10g and Ikno, are the distances from the antenna to
the log and the knot; nai, and nkno' are the refraction indices at the interface. (Kaestner and Baath
2000; courtesy of Forest Research Society)
with this equipment are given in Section 4.4.2. Technical details concerning the
equipment used for the measurement of phase and amplitude for the inspec-
tion of lumber are given by King (l978), King and Yen (l981) and Martin et al.
(l987).
While microwave techniques have shown a great potential for inspection of
lumber, their applicability seems to be limited by their inability to recognize
the nature of the defects. To overcome this drawback (Choffel and Martin 1966;
Choffel 1999), it was proposed to supplement the equipment with a CCD
camera, thus providing a coupling of a vision and a microwave device (Fig.
4.22). In this system, wood structural data are correlated with lumber strength
and compared with classification grades. The assignment of a grade to a piece
of lumber was based on the detection and location of defects and on three
physical and mechanical parameters: the density, the strength and the stiffness.
The camera provides the profile to calculate the dimensions of the board and
to detect the presence of defects, while the microwave sensor shows the inter-
nal structure of the wood. The network is described by Beatty et al. (l999) and
is based on the identification of the characteristic microwave signature of
lumber that distinguishes one piece from another. The accuracy of the in-
spection of the internal structure depends on lumber thickness, operating
frequency and the input power level.
The system described here has some advantages, such as: relatively low cost,
satisfactory design complexity, good spatial and measurement resolution, no
extensive need for signal processing and no radiation hazard. The system
requires no contact and has a rapid response to automatic process control.
lights
vision part microwaves part

defects detection knots detection
a) and strength estimation
Linear Camera
1728 pixels
256 Grey levels
Knots sensor
Halogens
Conveyor
b)
Fig.4.22a,b. A vision and microwave-coupled device for detection of defects in lumber. a The
device used for lumber inspection. b Details of the system of microwave probe and vision camera.
(ChoffelI999, with permission)
For the inspection of composite boards (fiberboard, flakeboard and parti-

cleboard), King and Bausel (1993) developed a homodyne instrumentation
with a continuous-wave source of 10GHz. (Fig. 4. 23). The microwave signal
received is amplitude modulated at 455 kHz and then mixed in the homodyne
receiver with a portion of a continuous wave carrier signal. The wave ampli-
tude and phase are converted to the 455 kHz base band signal for detection.
Based on all of the collected experimental data, the computer processes
the moisture content, density and dielectric constants. More recently, V6lgyi
(2000) has proposed a nondestructive method based on a microwave free-
md}
m", m
ref. ....
rp 8 me
phase and
amplitude A
e-< :>
0-
E p", }p
Pd
0
detectors u
e'
e" }e
Fig. 4.23. System for inspection of composite boards with a homo dyne instrumentation. (King
and Bausel 1993; courtesy of Forest Products J)
space/double transmission reflection for the measurement of the dielectric

complex tensor for continuous quality prediction of particleboard using
monostatic transmission sensors in reflection-aperiodic-open configuration.
Images of moisture distribution in real-size particleboard were obtained yield-
ing an average value of 4% with a standard deviation of 0.48%.
4.4 Applications
This section is intended to cover a wide range of topics related to the applica-
tion of microwave imaging techniques for the inspection of forests, logs,
lumber and wood based composites. The microwave parameters used for
image reconstruction are: the amplitude, the phase and the polarization. The
system involves transmission of microwave energy through the samples or
objects and measurements of reflected, refracted or diffracted fields. Two- or
Applications 159
three-dimensional images of the objects can be obtained without physical

sectioning.
As with other imaging techniques, inspection of large areas are possible. At
the same time, wood structural information can be obtained on a scale ranging
from meter to centimeters. The breadth of the discussion is intended to aid in
deciding which approach would be best to make a specific series of measure-
ments for the purpose of a well-defined imaging purpose.
4.4.1 Microwaves for Inspection of Forests
Microwave dielectric properties of plant materials were extensively studied at

Kansas University in Professor Ulaby's laboratory. As noted by Ulaby and
Jedlicka (1984), "from the standpoint of wave propagation, a vegetation canopy
is a dielectric mixture of discrete dielectric inclusions (leaves, fruits, stalks,
etc.) distributed within a host or background material such as air:' On the other
hand, the canopy is an inhomogeneous and anisotropic medium. The size of
the canopy inclusions compared with the wavelength in the microwave region
determines the absorption and scattering of microwaves.
The absorption and scattering microwave coefficients depend on the both
wave polarization and the angle of incidence of the microwaves to the canopy
and are related to the dielectric constant, volume fraction and geometry of the
canopy inclusions (Ulaby et al. 1990; Richards 1990; Baker et al. 1994). It was
demonstrated by Ulaby et al. (1983) that the canopy absorption coefficient can
be calculated from the canopy equivalent dielectric constant ee where ee is a
tensor with two components associated with the horizontal and vertical polar-
ization directions. The waveguide transmission method can be used for the
measurement of this dielectric tensor.
The two main factors influencing the dielectric tensor of plant materials are
the dielectric constants of the water component of the plant material and the
temperature of the media. EI-Rayes and Ulaby (1987) used a broadband mea-
surement technique to determine the dielectric properties of vegetation (leaves
and trunks) over a wide range of frequencies, from 0.2 to 20GHz. As can be
seen from Fig. 4.24a, the dielectric spectrum of a balsam fir tree shows an
important effect of frequency on 10' between 0.5 and 6 GHz. A smaller frequency
effect was found for elf.
To evaluate the radial variation of dielectric constants, measurements were
performed from the pith to the bark on a freshly cut trunk of balsam fir, as can
be seen in Fig. 4.25. The variations of the complex dielectric constant 10* values
are represented by concentric circles as a function of two parameters, the posi-
tion on the transverse section of the tree and the corresponding moisture
content. The values reported for each circle represent the average of 16 sample
measurements. From the above figures, we observe a decrease of the 10* values
with increasing moisture content from the pith to the bark. The complex
00 balsam fir
trunk
Mv=0,166
.." .. .
...
t'11
o Co o c
o
o 2 4 6 8 10 12 14 16 18 20 22
frequency (GHz)
Fig.4.24. Dielectric spectrum for f' and fN of a balsam tree trunk in a frequency range from 1
to 22GHz. (EI-Rayes and Ulaby 1987; copyright 2001 IEEE)
5,5
x(cm)
® denote measurement made wi th probe pointing

downward again t the trunk cross-section
Fig. 4.25. Radial variation from pith to bark of the complex dielectric constant f*, of a fresh cut
balsam fir tree. (El-Rayes and Ulaby 1987; copyright 2001 IEEE)
dielectric constant in the pith region is E* = 32 - j 8 and in the bark region is

E* = 8 - j 1.
The applicability of microwave techniques for detection and classification
of forest vegetation was possible by using radar imaging in the microwave fre-
quency range (Sieber 1985; Kessler 1987; Kasischke and Melack 1997). It was
Applications 161
demonstrated that the estimation of the woody plant biomass and the
monitoring of different temporally dynamic processes are possible by using
spaceborne synthetic aperture radars (SAR). The best performance for
biomass estimation was achieved with the lower frequency (P- and L-bands)
radar systems.
The penetration measurements at the L-band for coniferous and deciduous
trees are given in Fig. 4.26, for horizontal (HH) and vertical (VV) polarization
directions using the experimental arrangement shown in Fig. 4.27. The data
were collected at I-m distance along blocks of ten trees. The mean backscat-
tering cross section was calculated as a function of the height, ranging from
the ground to 6m. For coniferous trees (spruce) the backscattering between 1.5
and 4m is relatively constant at -20dB for both HH and VV channels and
increases dramatically between 4 and 6m. The attenuation is higher for hori-
zontally (HH) polarized L-band waves than for vertically polarized waves
(VV). The behavior of hardwood trees is different, and the hypothesis that the
radar acts as a shape filter was advanced by the authors. They proposed that,
for hardwood trees, the volume of branches is large enough to act as prime
scatterers. For coniferous trees, however, the tree area projected in the direc-
tion of the incident radar wave seems to be the most important parameter. The
backscattering characteristics of spruce trees were investigated as a function
of polarization angle, ranging from 0° to 100°. At 0° the electrical field vector
was oriented parallel to the trunk. The signal obtained was different, depend-
ing on whether or not the trunk was illuminated (Fig. 4.28). The stem strongly
scatters the incident wave, and the increasing polarization can be related to the
decrease of its cross section with height. The slope of the curve between 0° and
35° can be related to the diameter of the trunk. At a 43° polarization angle, a
maximum of the curve was observed. This angle was presumed to correspond
to the preferred orientation of the major branches. The peak at 43° was also
observed for a tree whose trunk was not illuminated. For polarization angles
larger than 43°, both curves (HH and VV) have the same appearance. This
aspect was related to the high probability that other branches were oriented at
an angle of 60°.
Hardwood trees have a different backscattering behavior because of the dif-
ferent size and distribution of their branches (with or without leaves) in the
canopy. The signature of a deciduous tree is very different from that of a conifer
tree, as can be seen for a walnut tree (Fig. 4.29), a very asymmetric tree. In this
case, a minimum of -13 dB was observed at a polarization angle of 60°, fol-
lowed by an increase to -9 dB at 85°. The decreasing part of the curve, between
0° and 60°, is due to the thick branches down to 45° from nadir. The second
part of the curve for polarization, above 60°, was related to the connection
angles between the trunk and branches. The modification of illumination
geometry indicated a different orientation and branch structure (Fig. 4.29).
Sieber (1985) concluded that the selection of an appropriate polarization with
an X-band and an L-band radar permitted the detection and identification of
deciduous and coniferous trees. The experimental data reported by Sieber
decidous forests
oQ(dB)
-10
~ -12
0
-;;; -14
-=
.:2 -16
ta;::: -18
-=Q) -20
~ -22
-24
-26 height (m)
-28
0 1.25 2.50 3.75 5.00 6.25
a) horizontal polarization
decidous forests
oQ(dB)
-10
~
-12
0
-14
-;;;
-16
-=
.:2 -18
ta
-20
~ -22
~
-24
-26
height(m)
-28
0 1.25 2.50 3.75 5.00 6.25
b) vertical polarization
coniferous forests
oQ(dB)
-10
-12
~ -14
;:: -16
~
-=0 -18
·~-20
2-22
Q)
~-24
-26
height (m)
-28
0 1.25 2.50 3.75 5.00 6.25
C) horizontal polarization
Fig.4.26a-d. Penetration measurements at L-band radar for coniferous and deciduous forests.
a, c Horizontal HH polarization. b, d Vertical VV polarization. (Sieber 1985, with permission)
Applications 163
con iferous forests Fig.4.26a-d. Continued.
-10
-1 2
~
0
-14
x -1 6
c - 18
0
.~
-20
..,c'" -22
t::
ro -24
-26
height (m)
-28
0 1.25 2.50 3.75 5.00 6.25
d) vertical polarization
a)
tree
"
//1
radar antenna
height of radar
above gro und
. '"" f I
vertical heighr of
bcamltrce intersection
b) polarization of electrical field

radar antenna
height of radar
above ground
\LI'1 antenna beam

\ . I
c)
\ 'I
IV Fig.4.27a-c. Experimental arrangement
I
for penetration measurements (Sieber
~\"""'
1985; courtesy of ESA Journal). a The posi-
tion of radar antenna and of the tree.
b Position of the radar central beam and of
the polarization direction versus the tree.
[C5S
c The INSCAT (Institute for Navigation's
'\
,,).=;)
Scatterometer) sensor mounted on a truck.
""-==(E- -~~~ (Ulaby and Jedlicka 1984; copyright 2001

IEEE)
Fig.4.28a,b. Backscattering characteristics for a spruce tree as a function of polarization angle.
a Tree illuminated by radar beam. b A tree not illuminated by radar beam. (Sieber 1985; courtesy
of ESA Journal)
(1985) were collected in the Rhine Valley, Germany, with the INSCAT (Institute
for Navigation's Scatterometer) sensor mounted on a truck (Fig. 4.27c).
4.4.2 Microwaves for Internal Inspection of Logs
The inspection of the internal structure of logs by the microwave technique

has been reported by Kaestner and Baath (2000) The detection of knots in logs
Fig.4.29a-d. Backscattering characteristics for a walnut tree as a function of polarization angle.

a Tree illuminated by radar beam. b Tree not illuminated by radar beam. c Idealized profile of the
tree. d Detail of branches. (Sieber 1985; copyright 2001 IEEE)
Applications 165
0° (dB)
-5.0
-6.0
-7.0
-8.0
-9.0
- 10.0
- 11 .0
-1 2.0
-13.0
- 14.0
(a) 0.0 25.0 50.0 75.0 100.0
0° (dB)
-5,0
-6,0
-7,0
-8,0
-9,0
- 10,0
-11 ,0
-12,0
- 13,0
po larization angle (degree)
- 14,0 I I I I I
(b) 0,0 25,0 50,0 75,0 100,0
c) d)
Fig. 4.30. Distribution of knots on the

transverse section of the bolt. (Kaestner
and Baath 2000, with permission)
of various species was performed in the frequency band between 4 and 8 GHz.
In this section, we discuss the imaging technique used for the detection of
knots in a spruce bolt of 17.5 cm diameter and 18 cm length. The actual dis-
tribution of knots is sketched in Fig. 4.30 where we observe three large knots
designated as a, b, and c. They are disposed in a radial position at about 900
between knots a and b and at 600 between knots band c. Knot b is externally
visible.
Figure 4.21 shows the experimental device for knot detection. It is composed
of three main parts: the microwave device with a wide band antenna, the con-
veyer for bolt rotation and translation, and the computer for equipment control
and for storage of experimental data. The image reconstruction is based on the
following dielectric parameters of the wood: the polarization ratio and
the phase difference of two components of the electrical field, one parallel to
the fibers and the other perpendicular to the fibers. The wave propagation phe-
nomena that take place in the sample are very complex and a simplified
hypothesis must be introduced to facilitate the understanding of these phe-
nomena. In a first approximation it is assumed that the received signal at the
antenna is the sum of two signals, the first emitted from the log and the second
emitted from the knots. Secondly, it is assumed that for the continuous case,
the sum can be replaced by an integral and the dielectric constants by a tensor
function.
Continuous wave radar measurements of the depolarization of reflected
waves were used to reconstruct the image. In the frequency range from 4 to
8 GHz, it was possible to obtain a discrete frequency spectrum. Using the
inverse Fourier transform, the discrete spectrum measurements were trans-
formed into a time delay spectrum. This spectrum shows the spatial distribu-
tion of the electrical field and of the polarization-related parameters that are
used for image reconstruction.
Applications 167
The images obtained with the polarization ratio (Fig. 4.31a) have less reso-
lution than those images obtained with phase difference (Fig. 4.31b), on which
the knots are better defined. An improvement of the image resolution was
obtained using iso-surface topographic slices calculated for the polarization
ratio and for the polarization angle between the two components of the elec-
tric field (Fig. 4.32a,b). Knots a and b are readily visible in both cases, while
knot c is readily visible only in Fig. 4.32b, where data based on the polariza-
tion angle were used.
Further research is needed for improvement of the image resolution and for
a better understanding of possible correlations between the dielectric tensor
Fig.4.31a-d. Iso-surface on a topographic slice. a Ipi The amplitude; b ythe argument (y, arg p).
c E The ellipticity angle which gives the phase difference between two wave components and
describes the degree of elliptical polarization. d r The angle between the two components of
the wave, corresponding to the tilt of the linear polarization. (Kaestner and Baath 2000, with
permission)
a) real image
c)
Fig.4.32a-c. Microwave imaging of a knot in softwood lumber. a Real image of the board.
b Longitudinal-tangential microwave imaging of the section through AA. c Radial- tangential
microwave imaging of the section through BB. (Chazelas 1991, with permission)
and the physical parameters of the wood, such as the slope of grain and the
variation of moisture content. Today the technology is available to build a
microwave scanner that can operate at industrial rates. The only important step
to be fulfilled is the improvement of image resolution.
Applications 169
4.4.3 Microwaves for Mechanical Grading of Lumber
Microwave nondestructive techniques have increasingly attracted the interest

of wood industry practitioners for many years because of the ability of
microwaves to penetrate the wood and to inspect the internal structure without
contact. The microwave measurement systems are relatively inexpensive and
can operate in industrial environments. The major applications were originally
oriented toward the development of continuous systems for drying of lumber
with microwave energy (Barnes et al.I976). However, microwave techniques are
also being considered today as an "emerging technique" for quality control of
lumber. The technical approach when conducting microwave inspection of
lumber consists of using the wave properties in the far-field of the transmit-
ter. In this case, access to both sides of the lumber is not required. Coherent
phase information and the ability to polarize the waves as a function of wood
structure or wood defects have significantly improved the capabilities of the
microwave method for the evaluation of lumber quality.
King and Yen (1981) proposed a microwave analog and digital system for
nondestructive testing of dimensional lumber. The versatility of the system was
obtained by using a combination of two elliptically polarized waves, that take
into account the reflection, refraction and diffraction of waves by wood, which
is an anisotropic medium. The elliptically polarized waves were measured
following the major and minor axes of the field. The ratio of the amplitudes at
the major and minor axes ranged between 15 and 40dB. Lumber thickness,
slope of grain and moisture content are critical parameters for lumber flexural
strength. In a microwave field wood becomes very attenuating material when
it has a high moisture content or is too thick.
The system proposed by King and Yen (1981) has several advantages:
- the low cost of the components;
- the signal processing is done at audio frequency;
- a single probe antenna is used to measure polarization and there is no need
for a probe calibration;
- amplitude, phase and polarization angles are available in real time, simulta-
neously and independently.
The main disadvantage of the system is that at high speed inspection rates
of lumber, the mechanical vibrations disturb the measurements of the polar-
ization ratio and increase the noise in the electrical field.
The implementation of microwave scanning techniques for defect detection
in the wood industry is only 20 years old (Szymani and McDonald 1981; Glos
1982). The scanner produced in Finland, for an inspection speed ranging from
60 to 300 mIs, used the polarization of microwaves for detection of the slope
of grain, and the phase velocity of microwaves for detection of knots in soft-
wood lumber.
Automatic radio frequency scanning of lumber with knots, disordered slope

of grain, clear wood, and wood with different defects has been demonstrated
by many scientists (McLauchlan et al.1973; Samson 1988; Niemz 1989; Choffel
et a1.1992; Rice et al. 1992; Steele et al. 1992,2000; Tremblay 1995). The slope of
grain indicator produced by Metriguard (USA) has the ability to evaluate the
defects on unplaned and planed hardwood lumber. All defect types can be
evaluated with the exception of rot. The capacitance head of the device has an
18-mm diameter and can be positioned at 3mm from the wood surface.
The evaluation of dielectric properties of wood by the device is made with
3600rpm and provides 60 grain-angle measurements per second.
Since in a sawmill the profit crucially depends upon choosing the right
sawing patterns, it is important to be able to determine the size of the lumber
accurately. Microwave measurement systems are capable of operating in in-
dustrial environments without requiring access to both sides of the lumber,
the microwave attenuation being exponentially related to the thickness.
Numerical values of attenuation of microwaves at 10GHz in the longi-
tudinal and tangential directions of wood, for spruce samples of different
thicknesses are given in Table 4.8. It is informative to note the important
variability of these values and to mention that interface reflection phenomena
were not considered. The 3-D imaging of a knot in a spruce sample with
a laboratory device is shown in Fig. 4.33. The position of the knots is clearly
located and the deviation of the slope of grain around the knot is well
represented. A microwave imaging of a spruce board containing different
defects (a sound knot, a pitch pocket, cracks in a knot) and sound wood is
presented in Fig. 4.34. The defects are well identified but their nature is not
quite clear.
4.4.4 Microwaves for Inspection of Wood-Based Composites
In this section we will analyze the microwave nondestructive technique for

testing of wood-based composites, such as fiberboard, oriented flakeboard and
particleboard. Relevant contributions have been reported by Musial (1988),
King and Basuel (1993), Wissing and Welling (1995) and Volgyi (2000). The
dielectric parameters selected for the inspection of wood-based composites
were the dielectric constants of the material and the amplitude and phase of
the wave. These parameters were correlated with the moisture content, density,
temperature and mechanical properties of the boards.
The moisture content measurement in wood-based composites using
microwaves is based on their large water permittivity compared to that of other
constituents. In Fig. 4.35 the variation of the relative permittivity of pure water
at 25°C versus frequency and wavelength are presented. We note the inverse
variation of both E' and E" versus frequency in the range from 1 to 10 GHz. At
a frequency greater than 20 GHz , both E' and E"have the same decreasing ten-
Applications 171
Fig.4.33a,b. Images of knots reconstructed with a the polarization ratio; b the phase differ-
ence between the two components of the electric field, parallel and perpendicular to the fibers.
(Kaestner and Baath 2000, with permission)
dency. It is well known that the moisture in wood is composed of both water
and dissolved ions. For a hypothetical liquid containing both water and ions,
the variation of 10" versus frequency is different than that of pure water as can
be seen from Fig. 4.36. The bound water does not have the same behavior
as free water. The bound water reached a maximum value between 108 and
length length
real image microwave image
Fig. 4.34. Microwave imaging of a softwood board (100 X 40 X 1000mm) (Choffel 1999, with
permission)
wavelenght (em .)
0
30 3 0(3
0
S'
0
.;;
:€
~o
P.-
/
<l)
;;.
/
'.g /
V.... /
S"/
/
10 100
frequency (GHz)
complex permittivity of water at 25°C
Fig. 4.35. Relationship between the permittivity and frequency for pure water at 25°C.
(Thompson 1989, with permission)
Applications 173
('
loss
t
factor
total
50 ionic conductivity
free water
bound water
/
I
........ -<.
1- - _-I'. -- ....
,I
Fig.4.36. e" as a function of frequency for a hypothetical dielectric material containing water
and dissolved ions. (Thompson 1989, with permission)
109 Hz, in the same domain in which free water reached a minimum. The influ-
ence of free water is more important than that of bound water at a frequency
higher than 1 GHz. This graph (Fig. 4.36) can help in selecting the appropriate
frequency for wood drying because of the capability of the electromagnetic
radiation, directed through the bulk material to quantify the water content in
real time (Thompson 1996).
From a large number of references we can see that the dielectric constants
were extensively used for measurements of the moisture content in wood and
wood-based-composites (Nanassy 1970, 1972; Norimoto and Yamada 1972;
Tiuri et al. 1980; Skaar 1988; Kuroda and Suzuki 1996; Yokoyama et al. 1999,
2000a,b).
Thompson (1989) found that the volume of water present in the dielectric
material can be deduced from the complex permittivity of the sample E*, and
can be expressed as follows:
(4.27)
where: Vm is the fraction volume of water; e,.* is the complex permittivity of
water, and Ed* is the complex permittivity of the dielectric material such as dry
wood substance. In this equation, all parameters are known except the volume
fraction of water, Vm'
For in-process quality control, King and Basuel (1993) developed a non-
contact microwave transmission technique for measurement of the moisture
content of fiberboard, based on the assumption that the variation of the atten-
uation and phase of the microwave signal are linearly dependent on the dry
masses of solid components and of water.
The absolute moisture content, MC~w, deduced by the microwave technique,
is calculated as the ratio of the partial basis weights of the water present to the
partially dry wood weight.
(4.28)
where MC~w is the absolute fractional moisture content of the sample, mw is the
weight of water, and md is the weight of wood on the basis of the absolutely
dry fraction.
The total weight mtot.l can be deduced as the sum of the components:
(4.29)
The variation of the amplitude and of the phase as functions of md and mw
are given by:
(4.30)
(4.31)
where aj and az are experimental calibration constants for amplitude, and, a3
and a4 are experimental calibration constants for the phase.
After calculation, the expression of the moisture content as the ratio
between mw , the weight of water, and md, the weight of wood, on the basis of
an absolutely dry fraction, the moisture content is expressed by:
MC~w =mw/md = (a3LlA-ajLl<!»/(azLl<!>-a4LlA) (4.32)
The values of the coefficients a j az a3 and a4 are given in Table 4.11.
Figure 4.37 gives the variation of MC~w as a function of the gravimetric
moisture content measurements in the range of 0 to 20%. A breakpoint in the
variation of amplitude and phase were observed at 6% moisture content. Under
Table 4.11. Values of the coefficients al> a2, a3, a4 calculated by a least square regression for fiber-
board at constant temperature of 24°C. (King and Basue11993; courtesy of Forest Products J)
Moisture content md (g/cm2)' a 1 (dB cm2/g) a2 (dB cm2/g) a, (deg cm2/g) A4 (deg cm2/g)
>6% 0.614 -0.936 53.15 76.94 469.62

0.777 -0.181 48.90 77.36 513.68
0.948 0.031 43.59 78.28 515.64
Mean value -0.268 47.39 76.75 515.39
<6% 0.614 0.594 27.63 81.16 377.22
0.777 0.905 30.04 83.42 408.05
0.948 1.204 29.04 85.17 454.66
Mean value 0.962 27.81 83.65 409.34
'md (gtem2) is the weight of wood per unit area of board, on the basis of absolute dry fraction.
md = Pdt; where Pd (gtem') is the dry bulk density of the board of thickness t (em).
Applications 175
20
md=614kglm 3
16
~ me>6%+0.43%
e.... 12
~
~
], 8
a
()
4
me<6%+0.45%
0
0 4 8 12 16 20
me (by weighing) [%]
Fig. 4.37. Relationship between the moisture content determined with a microwave technique
and the routine moisture content measurement for a fiberboard of 614kg/m3 density, at 24°C.
(King and Basue! 1993; courtesy of Forest Products J)
6% moisture content, the water molecules are strongly bound to the chemical
constituents of wood and the attenuation of the microwaves is reduced and
exhibits a nonlinear behavior. Increase in the moisture content from 6% to the
fiber saturation point activates weaker bonds of hydroxyl groups of the hemi-
celluloses. At the fiber saturation point, free water is present in the cell lumina.
From the electromagnetic point of view, there appears to be little or no differ-
ence between the polarizabilities of weakly bonded and free water, so that I1A
and 11<1> are linearly dependent on the moisture content above 6%.
The influence of moisture content on the complex dielectric constants of the
fiberboard is shown in Fig. 4.38. The real dielectric constant £' is linearly
dependent on the moisture content, whereas the dielectric constant £" is non-
linearly dependent on the moisture content. During the processing, the
temperature of a fiberboard varies over a very large range (from 200 e after 0
emerging from the press, to 20 0 e in the final stage). The influence of temper-
ature on the amplitude and phase variations at different levels of moisture
content, ranging from 0.3 to 10.4%, is shown in Fig. 4.39. Both the amplitude
and phase variations increase linearly with the temperature. The phase
variations seem to be more sensitive to an increase of temperature when the
moisture content increases from 5 to 10%.
Microwave attenuation was used by Musial (1988) to study the degree of
flake orientation of OSB under the assumption that the physical characteris-
tics of flakes are similar to those of solid wood from which the flakes had been
produced. On the other hand, Musial (1988) found that the influence of resin
2.6 Fig.4.38. Relationship between the

components of the complex dielectric
constants and the moisture content of
3.2 fiberboard at 10GHz and a constant
density and temperature of 24 DC. (King
and Basuel 1993; courtesy of Forest
2.8 Products J)
2.4
2.0
0 4 8 12 16 20
1.0
0.8
0.6
0.4
0.2
4 8 12 16 20
mc(%)
and water repellent on the dielectric properties of the board can be neglected.
The third point is that there is no electrical interaction between the board and
the individual flakes. The dielectric properties of the flakes are not significantly
modified during board production. The anisotropy of the boards is expressed
by the ratio between the attenuation parallel and perpendicular to the direc-
tion of board formation. Considering the coordinate system x, y and z for the
board and x', y' and z' for the flakes as shown in Fig. 4.40a, the attenuation
anisotropy ratio is a linear function of flake orientation (Fig. 4.40b). This linear
relationship can be used in further development of a continuous nondestruc-
tive monitoring technique in the processing of boards.
The results previously described are based on the microwave transmission
technique. More recently, the development of monostatic transmission sensors
in reflection-aperiodic-open configuration (King 2000) permitted the con-
tinuous control of large dielectric boards. Volgyi (2000) proposed the utiliza-
tion of a multifrequency permittivity monitoring system, in which the
attenuation and the phase are measured and correlated with the moisture
Summary 177
mc=1O,4% Fig. 4.39. Influence of temperature at a
1 ~I%
5 constant moisture content on amplitude
and phase variations, measured on a
fiberboard. (King and Basuel 1993;
courtesy of Forest Products J)
- 1,7%
1 0,3%
o !
30 40 50 60 70 80 90
mc=lU,4%
130
120
8,1%
110
~ 100
5,1%
~ 90
(3. 80
<1 70
60 tt~~~::======= 0,3%
1,7%
50
30 40 50 60 70 80 90
temperature [0 C]
content and density of the board and with the corresponding mechanical prop-
erties. Imaging of the distribution of the moisture content, resin proportion
and other parameters can be obtained for each board. Numerical analysis of
microwave measurements was based on the implementation of Kasa's circle fit
procedure (Corral and Lindquist 1998) that minimizes the errors of antenna
pattern measurements.
In the future, the holographic technique of a multifrequency microwave
method as generally described by Drobakhin (2000), for dielectric material,
could be an important advancement in the nondestructive control of large
wood-based composite boards.
4.5 Summary
The imaging of wood structure using microwave techniques is based on the

determination of its dielectric properties. This chapter has presented an
overview of microwave imaging applications for wood, describing theoretical
and experimental results. Systematic studies of the dielectric properties of
various wood species have been reported by many authors who have discussed
wood-water interactions and the measurement and prediction of wood
density and moisture content.
x' Fig.4.40. Microwave attenuation anisotropy

ratio and the degree of flake orientation in
board plane OSB at 13.6GHz (Musial 1988). Above Ori-
entation angle of flakes in the board. Below
Linear negative relationship between the
z
degree of flake orientation and the attenua-
y' tion anisotropy ratio. (Courtesy of Wood
y Science and Technology)
o right
• left
o~--~----~----~--~~
o 0,2 0,4 0,6 0,8 rad
degree of flake orientation
There are two basic microwave techniques, the transmission and the reflec-
tion technique. Signal analysis of probes is relatively simple and is related to
the measurement of amplitude, phase and the polarization of the waves. These
parameters are used for image reconstruction of wood structure. Today, the
areas that may benefit from microwave imaging techniques are related to the
internal defect detection of logs, lumber and wood-based composites, and, also
to the imaging of vegetation including leaves, stalks and stems under various
moisture and temperature conditions over a wide range of frequencies.
The imaging of wood structure can be obtained by scanning in the prox-
imity of the sample with a resolution given by the aperture size. The images
can be obtained in the far field or in the near field. In the far field, the small-
est detectable defect size is determined by the ratio d//.. >1, where d is the length
of the defect in the plane normal to the microwave vector and /.. is the wave-
length. The probes used to evaluate the properties of the medium can be either
scatterometers or radars, which measure the scattering properties, or reflec-
tometers, which measure the reflectivity of the medium due to its inhomo-
geneities and defects. The probes can operate in the proximity of the medium
as open-ended coaxial lines, cavity resonators or antennas. By moving the
position of the transmitter, an image of the medium can be obtained. While
microwave techniques have shown great potential for lumber inspection, their
applicability seems to be limited by the impossibility of identifying the nature
Summary 179
of the defects. To overcome this difficulty a vision system has been coupled to
microwave antennas.
The development of the microwave imaging technique is based on several
advantages such as:
- a noncontact operating system and relatively small size of antennas,
determined by the wavelength at microwave frequencies;
- fine resolution of measurements that can increase the ability of this
technique to detect defects;
- wave parameters, such as amplitude, phase and polarization can be
measured in real time.
The difficulties of applying microwave imaging to wood material arise from
inherent material properties, such as the anisotropy, heterogeneity, and the
presence of natural defects in wood. The main disadvantage of the microwave
system is that at a high rate of inspection the mechanical vibrations of logs or
lumber disturb the measurements of polarization. Technical limitations,
associated with the advantages of microwave imaging techniques, illustrate
the potential and capability of this method for new practical applications.
5 Ultrasonic Imaging
5.1 Introduction
Ultrasonic imaging is a technique widely used to reconstruct the properties of

the materials under inspection from global wave propagation data. Acoustic
tomograms provide an excellent noninvasive means of obtaining information
about inhomogeneous media as noted by many investigators (Kak 1979;
Herman 1980; Greenleaf 1981; Crawford and Kak 1982; Devaney 1982, 1986;
Dean 1983; Denis et al. 1986; Tomikawa et al. 1986; Dumoulin and de Belleval
1987; Falls et al. 1989; Frick et al. 1997; Douglas Mast 1999).
As for X-ray computed tomography, ultrasonic tomography refers to the
cross-sectional imaging of an object from either transmission or reflection
data collected by illuminating the sample from different directions.
Ultrasonic tomographic reconstruction techniques can be classified as:
- techniques based on the projection-slice theorem (filtered back-projection

and direct Fourier transform), which are fast, but restricted to projection
data that are sets of straight rays.
- techniques based on iteration procedures (algebraic reconstruction tech-
niques and simultaneous iterative reconstruction techniques) that are
relatively slow, but may be used with complex sampling geometries and a
bending ray path.
Different types of waves can be used for ultrasonic imaging of solids

(Schomberg 1982; Schechter et al. 1994, 1996; Sulivan et al. 1996). The most
common are the longitudinal waves but shear waves and surface waves like
Rayleigh or Lamb waves can also be used (Hutchins et al. 1993; Pei et al. 1995;
McKeon and Hinders 1999). First, a precise analysis of the wave type launched
into the sample must be made to understand the propagation phenomena in
an inhomogeneous solid wood specimen, with structural defects. Second, the
possible mode conversion must be studied and avoided, because of the strongly
induced degradation of ultrasonic images by such phenomena.
The resolution of the ultrasonic imaging techniques is very much limited by
the wave length and by the size of the transducers. However, unlike X-rays, ultra-
sonic rays do not travel in a straight line through heterogeneous and anisotropic
materials. For this reason, the computational requirements of this technique are
much more important than for X-ray computed tomography and an increase in
speed and accuracy of the reconstruction algorithms is necessary.
Ultrasonic tomography applied to wood is an important challenge for wood

scientists because of the natural variability, the anisotropy and the inhomo-
geneity of this material. Ultrasonic imaging techniques applied to wood must
be able to distinguish between the natural structure of the material and its
pathological features. Ultrasonic velocities and attenuation in different
anisotropic directions, the reflective properties of wood surfaces and the back
scatter of ultrasonic waves from the inhomogeneities must all be considered.
Proper signal processing methods must be chosen according to the struc-
tural characteristics of wood at macroscopic and microscopic scales. The com-
putationally most intensive part of the ultrasonic tomography technique, based
on the ray theory, is the tracing of the acoustic ray paths through the medium.
Commonly, ray tracing approaches are based on Snell's law (Lytle and Dines
1980). The ray bending is based on Fermat's principle of minimum propaga-
tion time (Kline and Wang 1992; Wang and Kline 1994). In anisotropic media
like wood, Snell's law is difficult to formulate because it is associated with the
wave normals while the wave paths are defined by the energy propagation.
Figure 5.1 shows the energy flux deviation for oak in three anisotropic planes.
Each wave has an energy flux that deviates from the wave normal. If an arbi-
trary oscillatory displacement is applied to the surface of a wood specimen,
the displacement will be resolved into three components which are then prop-
agated with individual phase velocities and energy fluxes. The variation of the
polarization vector when the propagation direction is outside of the principal
directions of the material symmetry is shown in Fig. 5.2. To display the prop-
agation phenomena in three dimensions, the numerical modeling methods
suggested by Lanceleur et al. (1998) were used. In Fig. S.2a (with three color
codes), the propagation vector is along r as defined by the local reference basis.
At the same time, three polarization directions are represented, corresponding
to three waves deduced from the Christoffel equation, such as one quasi-
longitudinal, very rapid wave, having the propagation direction parallel to the
polarization direction and two quasi-shear waves (the first one fast and
the second relatively slow) having the polarization directions perpendicular to
the propagation direction. From a theoretical point of view, the polarization
direction of these three waves is mutually perpendicular. Figure 5.2b shows the
variation in the acoustic properties of oak. The slowness surfaces correspond-
ing to the three types of waves are represented. The inner surface corresponds
to the quasi-longitudinal wave and the outer surface corresponds to the slow
quasi-shear wave. The smaller slowness value corresponds to the velocity mea-
sured in the fiber direction (L). The inner slowness sheet is colored in red and
exhibits a flat ellipsoidal shape. In the transverse plane, the color varies from
red to yellow. This means that the polarization direction varies more and more
when the angle approaches the third anisotropic axis of wood (axis labeled as
T, tangential). The variation of the polarization direction of the waves as a func-
tion of the propagation direction is indicated by the color code. It should be
noted that, for the quasi-longitudinal wave, the polarization vector is always
Introduction 183
IA
::'" ' ;" "
?- ..
..... ... ...
' . '.~" " "'-"',
.'
'.
, -,
..•.. "
:" ~;>::;'\
0,8 0,8
.'
.'
0,6 ." , '.
"
,,'
"
0.4
... - .. . ..,-
{" '~'
. ...
~
~
0.2
...• ... .. . ...... .. ....... ... ..

o 0,2 0.4 0,6 0,8 1.2 1.4 o 0.2 0.4 0.6 0.8 1.4
a) slO\~ ness (p~m m ) b) s.lowness (p:slmm)
" ,
I .•
1.2 .... " ... : ..... :-.

" "
" >':,
. ~.
."'
0.8 . ~\
Fig.S.1a-c. Energy flux deviation shown on
" slowness curves, as a function of direction of
' -" .
0.6 ultrasonic wave propagation in oak, in three
anisotropic planes (Berndt, pers. comm.). a
Plane LR; b plane LT; c plane TR. The slowness
curves correspond to the quasi-longitudinal
waves (the internal curve) and to the quasi-
transverse curves (two external curves)
deduced from Christoffel's equation. The
o 0.2 0.4 0.6 0.8 1.2 L4 numerical data were selected from Bucur and
C) , 10""", (pslmm) Archer (1984)
perpendicular to the propagation direction. This is not the case for quasi-shear
waves, for which there is an important variation of color. This means that the
polarization vectors of quasi-shear waves are more sensitive to the interaction
with wood structural elements than is that of the quasi-longitudinal wave. The
examination of the interaction between wood structure and the propagation
phenomena has been discussed in detail by several investigators (Bucur 1980,
1995; Berndt and Johnson 1994; Berndt et al. 2000).
For media of interest to the geophysics community, the modeling of veloc-
ity propagation (Kline et al. 1994; Wang and Kline 1994) was performed using
either finite difference or finite element formulations. In wood science, pio-
neering works on wood structure imaging reconstruction by scanning, from
a) X Polari ation
along 8
(tran verse wave)
Polarisation
along r
(longitudinal, a c)
Polari ation
along IjI
(tran erse wave)
b)
Oak
,.' .
.::.'
( ... ."
....
0.5
......
0
-0.5 'r'
: ....
-1
"·1
Fig. S.2a-c. Three-dimensional
representation of acoustic
properties of oak (c). a Local
basis and color code. b
0.5 Variation of polarization angle
on slowness surface for oak.
0
(Bucur et al. 2001, with
C) -0.5 permission)
Basic Aspects 185
ultrasonic data as velocities and stiffnesses have been reported by McDonald

(1978), Chazelas et al. (1988), and Biernacki and Beall (1993). More recently,
high resolution ultrasonic imaging has been reported by Tomikawa et al.
(1986), Biagi et al. (1994), Berndt et al. (1999), Comino et al. (2000), Rust (2000),
Socco et al. (2002) and Martinis (2002). Stress wave tomography in standing
trees was reported by Rust (2000).
This chapter is intended to cover the treatment of the subject of ultrasonic
imaging of wood structure on a wide range of topics related to image pro-
cessing and to the algorithms for the reconstruction of the internal structure
of solid wood and wood-based composites. The breadth of the discussion is
intended to aid the reader in deciding which technique would be the most
appropriate for a specific purpose.
5.2 Basic Aspects
Ultrasonic tomography is a diffraction type tomography and is an important

alternative to straight ray tomography with ionizing radiation. The main ben-
efits of imaging techniques using acoustic and electromagnetic radiation are
that they are noninvasive and safe at a low energy level. The accuracy of tomog-
raphy with acoustic energy is affected by the diffraction or refraction of the
energy field. In contrast to X-rays, acoustic waves do not travel along straight
rays and the projections are not line integrals. The energy flow is described
with the wave equation and can sometimes deviate from the straight-line
trajectory.
There are three main types of algorithms that can be used to form tomo-
graphic images from ultrasonic data: transform techniques, iterative methods
and direct inversion techniques. The parameters used in ultrasonic imaging
are: time of flight, amplitude, frequency spectra of the waveforms, the inte-
grated energy of the spectral peak making possible an estimation of attenua-
tion and the "central value" of the spectrum, and the phase shift. The peak
amplitude and centroid of each spectrum are sensitive parameters for the
detection of significant structural differences. The evaluation of wood behav-
ior in the frequency domain (Halabe et al. 1995) gives more information about
its structure than it does in the time domain. Energy distribution and energy
flow are good parameters for enhancing the contrast of the images.
In practice, approximate formalisms are employed to allow the use of the
theory of homogeneous medium wave propagation for media with slight
inhomogeneities. Different approaches can correct the way paths errors, such
as:
l. Algebraic reconstruction algorithms of the image that use an initial approx-

imation of the refractive index to estimate each ray path. For weakly refract-
ing objects, the correction of the refractive index distribution is obtained
measured forward .. ' .••.•...................

..........
scanered field •...,-
~
_..>.....-_--
Fourier trans/om
space domain frequency domain
Fig. 5.3. Schematic representation of the Fourier diffraction theorem (Kak and Slaney 1988),
which relates the Fourier transform of a diffracted projection to the Fourier transform of the
object along a semicircular surface in the frequency domain. (Copyright 2001 IEEE)
after a few iterations. One can assume that the cross section of the object
consists of an array of unknowns (Fig. 5.3). After setting up the corre-
sponding algebraic equations for the unknowns in terms of the measured
projection data, the image can be reconstructed.
The algebraic algorithms used most frequently are represented by the
following acronyms:
- ART - algebraic reconstruction technique - on which each equation cor-
responds to a ray projection. The computed ray sums are a poor approx-
imation of the measured ones and the image suffers from significant
noise.
- SIRT - simultaneous iterative reconstructive technique - reduces the
noise of ART by relaxation and produces better images than ART. The
relaxation parameter becomes progressively smaller with increasing
number of iterations. A comparison of ART,SIRT,LSQ (least-squares) and
SVD (singular value decomposition) tomographic inversions for geo-
physical applications is given by McGaughey and Young (1990). ART and
SIRT are iterative inversion techniques, while least squares and singular
value decomposition are direct inversion techniques.
- SART - simultaneous algebraic reconstruction technique - combines
ART and SIRT and yields image reconstructions in one iteration, using a
Basic Aspects 187
model of the forward projection process. This method is recommended

in complex image reconstruction with curved rays with overlapping and
nonoverlapping ray strips. The reader is referred to Herman (1980) for
an exhaustive explanation of the algebraic reconstruction and to refer-
ence articles (Gilbert 1972; Herman et al. 1973; Dines and Lytle 1979;
Andersen and Kak 1984; McGaughey and Young 1990).
2. Reconstruction of the image with an algorithm based on the Fourier dif-
fraction theorem, when the sizes of inhomogeneities in the object are com-
parable to the wavelength. This theorem states that a projection yields the
Fourier transform of the object over a semicircular arc in the frequency
domain (Fig. 5.4). This is the fundamental algorithm of diffraction tomog-
raphy. The wave equation and the Born and Rytov approximations are used
to derive a mathematical expression that relates the scattered field to the
object (Fink 1983). The mathematical approach to this subject is presented
very clearly by Kak and Slaney (1988) and by Kline et al. (1996), and the
algorithms in the frequency domain and the back projection method are
discussed.
Diffraction and scattering in solids become important when the size of

the object is comparable to the wavelength of the radiation (Delsanto et al.
1998). In this case, first-order Born or Rytov approximations for scattering
are used. The Born approximation assumes that the scattered field is small
compared to the incident field, whereas the Rytov approximation assumes
that the phase change is small compared to the wavelength recommended or,
in other words, that the phase gradient is small. The Rytov approximation
is for large scatter, while the Born approximation works better with small
scatters.
The factors that limit the accuracy of the images obtained with diffraction
tomographic reconstruction are related to the theoretical approach of the
approximations in the derivation of the reconstruction process and to the
experimental limitations. The mathematical limitations imposed by the Born
and Rytov approximations are severe, because the reconstruction process is
limited (Norton and Linzer 1980) in the range of the objects that can be exam-
ined (cylinders, spheres, etc.). The experimental limitations are related to the
finite amount of data that can be collected. Up to the limit introduced by the
possible presence of the evanescent waves, it is possible to improve the recon-
structed image by collecting more data. The effect of the Born or Rytov approx-
imations can be estimated if the exact acoustic field is calculated or measured.
The difference between the reconstructed image and the real object is a
measure of the quality of the approximation. The Born reconstructions are
appropriate at a large refractive index, as long as the phase shift of the inci-
dent field is less than 1t. The Rytov approximation is very sensitive to the refrac-
tive index and produces very good reconstructions for objects as large as 100A
(A is the wavelength).
P(t,B2)
Fig.SA Reconstruction of the image with fan beam projections collected in SI> S" etc. (Kak and
Slaney 1988); x,y coordinates related to the object in the space domain, ~1 and ~2 angles related
to the position of SI and S,; profiles PI (t, ~1) P2 (t, ~2)' etc. as a function of position and time t.
(Copyright 2001 IEEE)
The experimental limitations are related to the equipment limitations and

are caused by:
- ignoring the evanescent waves, that is, those that are attenuated over a
distance of several wavelengths. The limitation is quantified by the highest
received wave number kmaximum = 21t/'A;
- the finite received signal length, that is, a physical limitation on the amount
of data collected;
Equipment for Ultrasonic Imaging 189
- the number of sampling data along the receiver line, because the sampling
is limited and a nonzero sampling interval must be chosen. With the Nyquist
theorem, a low pass filtering operation can be introduced and the highest
measured frequency can be calculated from kmeasured = relT, where T is the
sampling interval.
- the limited views of the object under inspection, not estimated by the
Fourier transform, can degrade the image in certain directions.
It can be noted that an enhancement of acoustic imaging of polymers can

be obtained by colling (Yamanaka et aI1991).
High resolution images for wood of transmitted and reflected energy were
presented by Berndt et al. (1999) and the group in the Geophysics Department
at Politecnico di Torino, Italy, under the direction of Sambuelli and Socco
(Martinis 2002) and the group from the University of Torino, under the direc-
tion of Nicolotti. Ultrasonic tomographic images were obtained with living
trees, such as, for example, the tomography at 54kHz by a direct transmission
technique, shown in Fig. 5.5, which corresponds to the transverse section of a
tree (Platanus acerifolia) of 40cm diameter. The central zone of the section,
about 10cm in diameter, is degraded by a fungi. In this zone (Martinis 2002),
the values of ultrasonic velocity are very low (600-1000m/s) because of struc-
tural degradations induced by the fungi.
Berndt et al. (1999) reported images obtained with C scan by reflection and
transmission with southern pine samples of 100 x 70 x (3-70)mm, with 1 MHz
unfocused and 5 MHz focused transducers immersed in a tank filled with
water. On the ultrasonic images it was possible to recognize the earlywood
(Fig. 5.6) and the latewood.
5.3 Equipment for Ultrasonic Imaging
There are three traditional ultrasonic experiments related to three visualiza-

tion methods known as A-scan, B-scan and C-scan (Papadakis 1999). A-scan
assumes a constant velocity and maps reflections where the depth of the reflec-
tion is proportional to the time of detection. Mapping of equal velocity zones
permitted the detection of defects like knots in real size lumber pieces
(McDonald 1978; Schmoldt et al 1996b; Niemz et al. 1999; Kabir et al. 2000a;
Iancu et aI2000). B-scan attends the visualization with a set of waveforms along
a line. This technique is preferred in medical applications. C-scan visualizes a
set of line scans covering an area of the sample. This procedure is performed
by an immersion technique or with air-coupled transducers and is largely used
for nondestructive testing of materials for the aircraft industry. Work pieces of
complex curved shape and very large size can be inspected automatically in
correspondingly large tanks equipped with multi-axis manipulation devices.
The experiments can be carried out on different sound paths, by varying the
35
2.50
25 2.00
......
~
..... 200 1.50
1.00
0.50
[mm]
Fig. 5.5. High resolution images of the transverse section of a tree (Platanus acerifolia)
(Martinis 2002, with permission)
location of the transmitter and the receiver. C-scan may be either of a reflec-
tion type or a through-transmission type. Modern digital computers can
extract the signal corresponding to each point of the work piece and produce
two dimensional images. Two-dimensional images in C-scan obtained with
modern digital computers require advanced signal processing. Volumetric
imaging can easily be obtained from a set of line scans covering an area of the
sample for the benefit of understanding propagation phenomena.
Fig.5.6a,b. Earlywood and latewood in pine, imaged by ultrasonic reflection. a Ultrasonic and
photographic image superimposition at a 5-MHz area scan. b a at I-MHz area scan. (Berndt et
al. 1999, with permission)
5.3.1 Equipment for Contact Scanning
In this section, we consider "contact" methods, immersion methods and non-

contact methods such as procedures with air coupling transducers or similar.
The equipment for direct contact scanning in A-mode was developed first
(Harpole and McDonald 1981), As an example we have chosen the equipment
designed for green and rough sawn lumber scanning in through transmission
mode (Fig. 5.7). A pair of transducers of 1 MHz inspected the lumber piece
immersed in water and the time of flight was measured. The scanning speed
was about 50m/min, which corresponds to 9000 board ft/h. The main disad-
vantage of this system was the relatively low sampling speed of lumber and the
large quantity of water lost around the board at the input gate.
More recently, Schmoldt et al. (1996b) and Kabir et al. (2000a,b) at USDA
Forest Service have proposed a scanning ultrasonic system equipped with
rolling transducers, which operates in a range of frequency between 90 and 180
kHz. In view of the in-line detection of the specific differences between defects,
several ultrasonic parameters (Fig. 5.8) were selected for the signal processing:
time of flight centroid, time of flight energy, time of flight amplitude, pulse
length energy value and energy/pulse value. These parameters were used for
the detection of knots, cross-grain and bark pockets in yellow-poplar deck-
boards. Energy value and energy/pulse value were the most sensitive parame-
ters for defect detection. The scanning rate had little influence on the data
b)
Fig.5.7a,b. Ultrasonic scanning of lumber by immersion in a water tank.. a Device for lumber
scanning. b Ultrasonic and real images of a lumber board (McDonald 1978; courtesy of Forest
Products Society)
collection and the authors noted that "scanning at relatively high industrial
speeds is feasible."
For direct transmission measurements, the direct coupling of the transducer
requires couplants that can be liquids (Birks 1972), grease, different adhesives,
air, etc. The air-coupled transducers were used in line production of particle-
board and plywood for the detection of blows and blisters with a speed of 90
m/min. The equipment for ultrasonic tomography available today is only for
laboratory measurements and consists of an ultrasonic generator, transducers
and an oscilloscope. The ultrasonic signal obtained from field measurements
require modern and complex processing (Berndt et al. 2000).
a) b)
1 4 0 . . . - - - - - - - -- - ---, o
-10
120~---~------4
-20 .. .... - EV (dB)
1 00~~~·~~~~~~-4 -30 1-- - -EPV (dB) - - -
80 -40
60 i--- -
40~~~--_.~---~~
-50
-60
-70
~
);.I',~
·V- ~
.
-80
-90 .-
-100
\0 20 30 40 50 0 10 20 30 40 50
c) d)
350 , . - -- - - - - - - - - - - , o
300
---·----EV(dB) _
-+---------i--~--l ·20
--Energy - ---. EPV (dB)
250 ' -'-'- ' - Pulselenglh -t------:c---l -40
----- TOF·cenlro id :
200 -+---------:~--i--_l ·60
\ ...~
-v
ISO -j--- - - -- I I - -I/-- -I -80
~r ~
·100 I
I ~
·120
·140
o 10 20 30 40 50 o 10 20 30 40 50
BOlrd length (Inch)
Fig.5.8a-d. Parameters selected for ultrasonic scanning of board: time-of-flight centroid (TOF),
time-of-flight energy, time-of-flight amplitude, pulse length energy value (EV) and energy/pulse
(EPV). a Measurements of the TOF centroid through a yellow poplar sound knot. b Measurements
of the energy value and energy/pulse value through a yellow poplar sound knot. c Measurements
of energy, pulse length and TOF centroid through a decayed red oak board. d Measurements of
energy and energy /pulse value through a decayed red oak board.
Berndt et al. (1999) reported images obtained with relatively high frequency
ultrasonic transducers such as: 1 MHz focused transducers and 5 MHz unfo-
cused transducers using the device depicted in Fig. 5.9a. The line scan in the
radial direction of the sample is shown in Fig. 5.9b, in B-scan technique. The
e) f)
450 - . - - - - - - - - - - - , o -r-~-~-..-----,.-...,
..... ... EV (dB)
400 +-~- --Energy ·20 .
--EPV IdB)
350 ---- .. -- Pulse Length
300 --- - - TOF -cenlloid
·40 -1- - - - - - - - -1
250 +-- t--- -- - --j

200 +-- f -- - - - ----l ·80 -f--+-t+- -¥--------1
150 +---;-~------j
100 ,\...,~y.,........"__ _- ---l

.120 + - _J...-_ _ _ __ -j
50~~~~~~~~----l
·140 - ' - - - -- - - - - - - - '

10 20 30 40 SO 0 10 20 30 40 50
BOlrd length (Inch)
Fig. 5.8e,f. e Measurements of energy, pulse length and time-of-flight centroid through a crack
in red oak lumber. f Measurements of energy and energy Ipulse value through a crack in red oak
lumber. (Kabir et al. 2001, with permission)
contrast function is given by the instantaneous reflected energy. The overlay-

ing of the total reflected energy with the echo arrival time in radial direction
shows that the time is relatively constant, whereas the energy is more sensitive
to the structural features. It is generally accepted that the ultrasonic energy
incident on wood is split into three components, and that the energy propa-
gates in a direction normal to the slowness surface. The deviation of energy
flux can be calculated and can provide useful information about the sample
state (Berndt et al. 2000). Collecting signals at multiple locations or using mul-
tiple receivers at different positions and frequencies can improve the results of
experimental ultrasonic measurements.
5.3.2 Equipment for Noncontact Scanning
The development of equipment for noncontact scanning is related to the pro-

gress achieved in producing noncontact probes, for which air is the coupling
agent. Noncontact probes are also called as air (or gas) coupled ultrasonic
transducers. Coupling efficiency related to energy transfer (Berndt et a12000)
depends on impedance matching. Ultrasonic velocity in air is 330m/s. Wood
exhibits surface acoustic waves ranging from 385 to 1450m/s (Bucur and
Rocaboy 1987). Under these conditions, air can be a perfect coupling medium
for the detection of relatively slow acoustic waves.
a) wood sample
"
.. '.
-~
b) mllial poSition (mm)
10
c) radial posi1ion (rnm)
Fig.5.9a-c. Equipment for contact scanning with immersion technique. a Ultrasonic scanning
setup. b Scan line in radial direction at 5 MHz. c Total reflected energy and peak reflection arrival
time vs. scan position. (Berndt et al. 1999, with permission)
Several types of transducers are available for non contact measurements,

e.g., piezoelectric transducers unfocused or spherically focused, electromag-
netic acoustic transducers (EMAT), capacitive transducers, and laser beam
optical generators and detectors. EMAT probes have been used successfully to
detect defects in metallic pipes, tubes and plates and for studies of the
anisotropy of materials. These probes are inefficient ultrasonic generators and
require large power supplies for proper operation. Capacitive probes cannot be
used as ultrasonic generators. They require little space between the sample and
the probe. Laser beam ultrasound function avoids these difficulties. They
operate in both metallic and nonmetallic materials and can be used in hostile
environments. The ultrasonic waveform is generated by a pulsed laser incident
normal to the sample, and the resulting waveform is detected by a piezoelec-
tric transducer located on the opposite side of the sample. For non contact char-
acterization of composite materials, laser beam interferometers as well as laser
generation and detection of ultrasonic waves have been used (Green 1987).
The use of air as a coupling medium has an important advantage related to
the avoidance of permanent damage and decontamination of the specimen
tested (Luukkala et al. 1971). On the other hand, because of high acoustic
impedance contrast, air-coupled transducers exhibit very high sensitivity to
surface roughness.
On paper webbing in motion, air-coupled measurements were performed
at 150kHz (Jarti and Luukkala 1977). Panametrics transducers (type T7) have
been commercially available since 1996 for measurements on composites,
metal sheets, etc. The frequency of air coupled transducers for industrial appli-
cations ranges from so to 500 kHz as reported by Lynnworth et al. (1997) and
Lynnworth (1989). The internal characteristics of a noncontact probe and its
operating modes are shown in Fig. 5.10. The impedance matcher is placed
before the piezoelement, and the structure is protected with a thin metal
window.
The use of airborne ultrasound is affected by different factors, such as the
transmission and reflection coefficients at material boundaries, where a large
mechanical impedance mismatch exists, which leads to a large dynamic range
between the incident reflection and the echoes generated within the sample.
In air, the impedance is 410kgm-2 s-I, while for wood in the fiber direction it
is about 2 X 106 kgm-2 s- 1 and in metals about 20 x 106 kgm-2 s- l • The large
acoustic impedance mismatch at the transducer/air boundary makes the trans-
mission of energy across the interface difficult. About 99.9% of ultrasonic
energy is reflected from material surfaces. Table 5.1 gives the transmission
coefficients and energy transfer in various materials. Array transducers for
nondestructive application and imaging of solids were developed for the
improvement of image quality (Papadakis 1999).
The experimental arrangement for the examination of wooden plates
with air-coupled ultrasonic waves and with spherically focused piezoelectric
transducers is shown in Fig. 5.11. Two experimental configurations with I-MHz
transducers were used to measure the ultrasonic velocities in wooden plates
a) b)
Q
c)
~,;,
-
e e
~
\.
!- .•• ,
»1' • • • ')
~".II.,
_I.....I-i'''''\!'_...............-
~. ~ ...... "." . . '\.
normal through shear wave plate wave

transmi sion
Fig.S.l0a-c. Transducer for non contact ultrasonic measurements. a Transducer's flange is sup-
ported between two silicone attenuating rings (Lynnworth and Magory 1999). b Internal details:
1 thin metallic window; 2 AJ4 impedance matcher; 3 100-kHz piezoelectric element; 4 potting
(Lynnworth and Magory 1999). c Different experimental arrangements for thorough transmis-
sion mode, with longitudinal and shear waves and with Lamb waves. (Buckley 2000; Academic
Press, with permission)
TableS.I. Ultrasonic energy transfer in various materials in non contact mode. (Bhardwaj 2000; with
permission)
Material Impedance Transmission Transmission Energy Energy Loss Loss energy

(Mrayl)' coefficient in coefficient in transfer transfer energy at at interfaces
air water in air in water interfaces in water
(dB) (dB) in air (dB) (dB)
Steel 51 34 x 10-6 0.11 -89 -19 178 38

Aluminum 17 10-' 0.3 -79 -10 158 20
Acrylic 3.5 5 x 10-4 0.84 -66 -1.5 132 3
Silicone rubber 18 x 10-3 0.96 -35 - 0.35 70 <1
' The impedance is expressed in Mrayl (1 RayJ = 1 kg m-' S- I). The impedance of air is 440 MrayJ and the im-
pedance of water is 1.5 Mrayl.
" .....
a) b)
Fig.5.11a,b. Air-coupled ultrasonic waves in a wooden plate. a Two spherically focused piezo-
electric transducers coaxially aligned. b The focused transducer is coaxially aligned with a fiat
piezoelectric transducer. (Fortunko et al. 1991; copyright 2001 IEEE)
with a thickness between 3 and 25mm. The first configuration using two coax-
ially aligned, spherically focused transducers, having the axes of symmetry
parallel, is shown in Fig. 5.11a. The central plane of the sample contains the
foci of both transducers. The second configuration (Fig. 5.11 b) used only one
focused piezoelectric transducer and one unfocused transducer. The focal
point of the spherically focused transducer was located at the surface of the
wooden plate in front of the unfocused transducer. In both cases, the angle of
incidence (9) is variable to allow the measurement of the time of flight and of
the amplitude of the ultrasonic signal as a function of this angle. It is impor-
tant to note that transmitting transducers must be driven by a high-power,
unipolar pulse generator (450 V output voltage, 3Q impedance and 2000 V
peak-to-peak voltage applied to the transducers). A low-noise preamplifier and
four precision attenuators (total range 132 dB) were used to minimize the noise
factor.
The design of higher frequency noncontact probes (>500kHz) is difficult
because of the short wavelength and high absorption in air, and only two solu-
tions were considered efficient: piezoelectric ceramics elements modified to
optimize its transduction in air and capacitive transducers.
Recent advances in high frequency piezoelectric transducers (Bhardwaj
et al. 2000), in the frequency range of 200 kHz to 5 MHz, coupled with a
high dynamic range-nanosecond accuracy ultrasonic analyzer that is able to
provide 100-dB extra gain, 30 dB of which is used to compensate the non con-
tact mismatch and 70dB to overcome air-material acoustic impedance mis-
....--.--.. •
•,
~
•.
•
"'''' 1 ' '"
Ii
Ii
Fig. 5.12. Different ultrasonic scanning modes of sheet products. (Buckley 2000, with permission)
match, made it possible to obtain performances comparable with conventional

contact mode transducers. The accuracy of the time of flight measurement was
±50ns. Tomographic images with this equipment were obtained for an eight-
ply graphite fiber re-enforced plastic composite 1.5 mm thick, in pulse-receiver
arrangement. The transducers were 5 mm away from the sample. The scanned
area was 25 x 25mm, with 2-MHz frequency and 64-dB gain. The pulse was
produced by 4Q square wave, the scanning was performed with a 3-mm-
diameter focused receiver above the sample and with a 25-mm unfocused
transmitter underneath it.
Most applications of air-coupled ultrasonic probes for wood composites are
for single channel C-scan systems, used for testing flat panels (particleboards,
veneer, plywood) less than 20mm in thickness (Buckley 2000). Multichannel
systems were developed for large composite panels, for rapid inspection at the
rate of 1 m 2/min. The scanning of sheet products for industrial quality control
can be performed as shown in Fig. 5.12 with a bulk wave transmission mode
on a central line, in a zigzag course or by inducing Lamb waves with a beam
incident angle to the surface of the specimen.
In the lumber industry, air-coupled transducers are used for:
- quality assessment of bulk lumber for internal decay and voids prior to pro-
cessing. The transducers are used in conjunction with dry pressure coupling
for laminated lumber.
- the detection of delaminations in particleboard, veneer and plywood. By
using Lamb waves, it is possible to inspect the entire width of the boards.
- the detection of cracks in structural beams and utility poles for in-service
inspection of wood products.
Noncontact tomographic imaging of aluminum was reported by Hutchins
et al. (1993). They used a pulsed laser to generate Lamb waves in metallic sheets
that were detected by electromagnetic acoustic transducers (EMATS).
For wood-based composites, Grimberg et al. (2000) reported the utilization

of Lamb waves for imaging of veneer structure. The equipment for Lamb waves
launched by Hertzian contact and the operating mode of the transducer are
shown in Fig. 5.13. The buffer rods used for emission and reception were iden-
tical and were made from an aluminum-magnesium alloy with a radius of 3
mm. Relationships between the amplitudes of P and S waves generated into the
plate-inspected specimen and the elastic constants of the wood composite were
established. When a delamination between the veneer sheet and the particle-
board exists, the Lamb waves split into two beams, one propagating in the
veneer layer of thickness hi with velocity CI and the other one propagating in
the particleboard of thickness h2 and velocity C2. When delamination exists,
the propagation velocity along the y-axis, corresponding to the plate thickness,
is smaller than the corresponding velocity in a sound zone. The results pre-
sented here were obtained using the assumption that the sheets are homoge-
neous and isotropic. A better understanding of propagation phenomena can
probably be achieved in the future with more complex calculations, based on
the assumption that wood in an orthotropic material, with corresponding
Young moduli and Poisson ratios.
5.4 Applications
Topics covered in this section include imaging at macroscopic level of the inter-
nal structure of standing trees, of lumber, and of defects in wood-based com-
posites. Acoustic microscopy imaging is not discussed in this book.
In considering the application of ultrasonic imaging, it is important to note
that for wood the transmission technique seems to be the most appropriate.
Since resolution in an ultrasonic image is dependent on frequency, it is impor-
tant to note that the size of the sample to be inspected, the type of the trans-
ducer and the band width must be chosen with care, keeping in mind the
propagation phenomena and the anisotropy of the material.
Most ultrasonic inspection and imaging are done at frequencies ranging
from 50 kHz to 1 MHz. As the frequency increases above 1MHz, the image res-
olution increases but the signal attenuation increases also. Ultrasonic images
are degraded by blurring due to practical limitations imposed by beam diam-
eter and pixel size. Digital processing methods may be used to improve the res-
olution. The spatial resolution, at which the original data were acquired, limits
the improved resolution in the processed image. The resolution inherent to
acoustic images is basically determined by the beam diameter and by the pixel
size. Nyquist's theorem explains the effect of the pixel size on the image reso-
lution. The pixel size is inversely proportional to the resolution. In order to
support the spatial resolution, the beam must be spatially sampled at less than
half the dimension of the pixel size or, in other words, the pixel size of the
image must be one half of the ultrasonic beam diameter.
Applications 201
_--4
RF amplifier ....
P Texas
Instruments
Fig. 5.13. Inspection of a veneer sheet with Lamb waves. Above Experimental setup for Lamb
waves launched by Herzian contact in a multilayered structure composed of a veneer sheet glued
on a particle board. Below Transducer for Lamb waves measurements. (Grimberg et al. 2000, with
permission)
5.4.1 Imaging of Internal Structure of Standing Trees
Ultrasonic tomographic imaging of the internal structure of standing trees has

been reported in several recent articles by the group at Politecnico di Torino,
Italy (Comino et al. 2000; Socco et a12000, 2001; Martinis 2002) using the veloc-
ity calculation with SIRT. The ultrasonic velocity was measured by the direct
contact transmission method, using exponential probes operating at 54kHz.
The transducers were located on 16 equidistant points around the trunk
perimeter selected for the tomography. By changing the reciprocal position of
the transmitter and receiver for each measurement, 120 independent acquisi-
tions were obtained. The distance for the tomographic data was calculated
assuming a straight path between the two transducers and was determined
with an uncertainty of 0.1 mm which was supposed to be equal to the radius
of the transducer cross section.
Tomographic image resolution is related to the pixel size and from a math-
ematical point of view the number of cells cannot exceed the number of mea-
surements, and the size of the cell should not be smaller than the radius of the
first Fresnel zone. To define the parameters that rule the iterative processing,
the average time uncertainty (O.l/..ls) and the spatial resolution (5 cm) were
defined by a theoretical approach.
The software used for this purpose was developed for interpretation of
seismic borehole data (Friedel et al. 1992). The algorithm is based on the
relation:
R 1 R
T= f-dl= fPdl (5.1)
sv s
where T is the travel time of the ultrasonic wave from the source (S) to
the receiver (R ), I is the ray path length, v is the velocity and p = l/v is the
slowness.
The mathematical background of the inversion algorithm is treated in
reference books and articles and will not be discussed in this book because
of space limitations. As noted by Socco et al. (2001), the procedure can be
described as follows: "the velocity calculation was performed with SIRT,
starting from an initial arbitrary velocity distribution model, computing the
forward model travel times, comparing the calculated travel times with the
measured time, and correcting the initial velocity model. These steps are
repeated until the fitting between calculated and measured travel times is
considered good enough:'
The flow chart for data acquisition, analysis, and interpretation is shown in
Fig. 5.14. After the processing of ultrasonic data, the tomographic image was
compared with the real (photographic) image of the transverse section of the
tree. For this purpose, a disk of 40 cm thickness was cut from the tested tree.
From this disk, cubic specimens were cut, on which ultrasonic velocities were
measured and used to construct an alternative of velocities distribution. This
Applications 203
Trees
cutdown
Good
signal/noise
ratio
Fig.5.14. Flowchart for the tomographic imaging of the transverse section of a tree. (Socco et al.
reconstruction was compared with the first tomographic image obtained

directly on the tree.
For field and laboratory measurements, the signal was observed on an oscil-
loscope and stored on a personal computer for processing. Spectral analysis
was performed with FFT and Hilbert transforms of the signal obtained. The
best resolution obtained was 5 cm and was determined taking into account
simultaneously the mathematical and the physical resolution, for which the
first Fresnel zone (RF) ray distribution was calculated with Eq. (5.2), which
gives the relationship between velocity (v), frequency (f) and distance (d) for
the ultrasonic signal
(5.2)
25 ~-----------------------------------------------------,
20 +--------------------------------------------
~
ii
~ 15+------------------------------------
!
10+-------------------------------- - - - -
o
2.00 2.26 2.53 2.79 3.06 3.32 3.59 3.85 4.11 4.38 4.64 4.91 5.17 5.44 5.70
Pre nel r.y classes ( em )
Fig.S.IS. Fresnel ray distribution calculated at S4kHz on the hackberry trunk (Celtis australis)
shown in Fig. S.14. ( Socco et al. 2000, with permission)
The Fresnel ray distribution is given in Fig. 5.15 for measurements per-
formed on a 63-year-old hackberry tree (Celtis australis) with very complex
trunk shape caused by previous wounding and pathological attacks (Fig. 5.16).
The most probable Fresnel ray class is at 5 cm, which corresponds to the best
image resolution. A good agreement was observed between the three images,
that is, the photographic image (Fig. 5.16), the tomographic image of the trans-
verse trunk section obtained from 120 independent velocity measurements
(Fig. 5.17), and the reconstructed image with measured velocities on cubic
samples (Fig. 5.18a). It should be noted that the sharp velocity contrasts are
smoothed by the tomographic inversion.
For the illustration of the simultaneous variation of the velocities in the L,
Rand T directions, the specimens presented in Fig. 5.18b were used to measure
the corresponding velocities. The variation of velocity values versus the posi-
tion of the specimen is illustrated in Fig. 5.19. Comparing these results and the
values of the velocities employed for the tomographic imaging, it can be con-
cluded that the tomographic images match the images constructed from veloc-
ity measurements on the individual blocks.
Another example is shown in Fig. 5.20, in which the tomographic image of
the transverse section of a standing tree (Platanus acerifolia) of 450mm diam-
eter with an important decay zone is presented. The velocities measured in the
RT plane range between 1000 and 1800m/s. In the central, decayed zone, the
velocities are relatively low compared to sound wood. This is probably due to
the low density of the tissue affected by the fungal attack. The low velocity
peripheral zone observed on the tomographic image is caused by the experi-
mental configuration and is likely to be an artifact of the inversion algorithm,
because this zone lies outside all measured ultrasound paths.
Applications 205
Fig. 5.16. The complex shape of the transverse section of the trunk of hackberry (Celtis australis)
(Comino et al. 2000, with permission)
600 2.00
1.80
500
1.60
400 1.40
E 1.20
§. 300
1.00
200 0.80
0.60
100 D.40
0.20
100 200 300 400 500 600
[mm]
Fig. 5.17. Tomographic images of the transverse section of the trunk in Fig. 5.16 obtained with
120 independent velocity measurements. (Comino et al. 2000, with permission)
a)
km/s
600
1.80
500
1.60
400 1.40
E 1.20
.§. 300
1.00
0.80
0.60
0.40
100 200 300 400 500 600
b) [mm]
Fig. 5. 18a,b. Reconstructed image with velocity values measured on cubic specimens and posi-
tions of the specimens in the trunk section. Above Reconstructed tomographic image. Below
Cubic specimens. (Comino et al. 2000, with permission)
Applications 207
2
1.8
1.6
'OJ'
~ 1.4
Co
1.2
~
Q
l 0.8
0.6
0.4
~ Z~ ~ ~ ~ ~ ~ ~ ~ ~
sample labels
si Uj
z z
fLI
Fig.5.19. Measured ultrasonic velocities and calculated velocities for two orthotropic directions
Rand T on selected specimens shown in Fig. S.lS. (Socco et al. 2000, with permission)
1.80
1.60
1.40
1.20
1.00
0.80
0.60 Fig.5.20a,b. Ultrasonic
0.40 tomographic reconstruction of
the transverse section of a
standing tree (Platanus
acerifolia) 40cm above ground
level. (Martinis 2002, with
b) ImmJ permission)
5.4.2 Imaging of Lumber Structure
The development of the methodology for the characterization of wood with a

direct transmission technique, that is, the calculation of the off-diagonal terms
(Bucur and Archer 1984; Bucur 1986) of the stiffness matrix, supported the
exact calculation of elastic constants of lumber with knots. Using an array of
cubes of 2 x 2 x 2 cm (Chazelas et al. 1988), it was possible to determine the
variation of the stiffnesses at each point of the sample. The specimen was
divided into numerous cubes, on which ultrasonic velocities were measured
and corresponding stiffnesses were calculated.
The mapping (Fig. 5.21) of the variation of elastic constants in three
anisotropic planes of wood was possible, and in this way the imaging of
knots and of slope of grain deviation was obtained with a resolution of 2 cm
(Chazelas et al. 1988).
High resolution imaging of wood, at a millimeter scale, was possible when
image reconstruction from projection was developed using very sophisticated
signal processing for ultrasonic tomography for materials associated with the
development of the corresponding ultrasonic transducers.
Neuenschwander et al. (1997) have published ultrasonic images of lumber,
corresponding to a I-mm resolution of pine lumber knots as shown in Fig.
5.22. The image was obtained in C-scan mode, by an immersion technique in
water by reflection with 2.25-MHz broadband transducers. The image of the
radial section (Fig. 5.22a) clearly shows the typical pattern of latewood and
earlywood as well as the deviation of the slope of grain around the knots. In
the transverse section of the specimen (in the TR plane), the circular pattern
of the annual rings is well defined. The presence of compression wood can be
observed in the central zone of the figure in the 16th and 21st rings. Cracks
several millimeters wide are visible in the first annual ring.
5.4.3 Imaging of Defects in Wood-Based Composites
As noted by Maloney (1996) in the family of wood-based composites, the mate-

rials described under the label "engineered wood products" (particleboard,
medium density fiberboards, different structural panels, etc.) are manufac-
tured annually in millions of tonnes all over the world. Effective internal defect
detection for these materials takes full advantage of the development of non-
destructive imaging methods for quality assessment. In this section, we present
the methodology developed recently for the imaging of the most frequently
observed defects: delaminations and voids.
Imaging of delaminations in wood composites was performed with the
equipment described by Grimberg et al (2000; Fig. 5.13). The inspected
structure was composed of a 20-mm-thick particleboard, on which two beech
Applications 209
.. lifTncss." image of knots
e22
~
a) e33
real image of knots
-
- VI.--' r-::;
--
<
V :/
..... ..-' 11\
r-
~1 l\'
-
-_
1/
-t- ........ r, r-; k '- 1--
-f- ...... =' f- ......
I-
--
-.....: ::--;:
b) loose knot right knot
Fig.5.21a,b. Reconstruction of the image of boards with loose and tight knots in a pine sample,
based on the measurements of stiffness CLL = Cll CRR = C22and Cn = C33 and related angles on
cubic specimens compared with the real image of boards. (Chazelas et al.1988, with permission)
veneer sheets of I-mm thickness were glued onto the two faces. The delami-
nations were simulated by inserting rayon foils of 50 J.lm X 0.5 mm as shown in
Fig. 5.23a. The density of the wood composite was 780kg/m\ the Young's
modulus in the longitudinal direction was 1.4 x 10 1O N/m 2, and the corre-
sponding Poisson ratio was 0.14. The velocity in sound zones was 4300±90m/s.
In a zone with delamination, the velocity measured in the veneer layer was
significantly lower, ranging from 1100 to 2128 m/ s. The map of the delamina-
tions is shown in Fig. 5.23b. The location of each delaminated zone is clearly
visible.
a) I 20mm
200mV l100mV
SOOmV l100mV
Fig.5.22a,b. Ultrasonic images obtained in C-scan mode by immersion and in reflection mode
of the structure of a pine specimen. a Longitudinal-radial section with knots, the pattern of the
earlywood and latewood and the slope of grain deviation around the knots. b Transverse section
with typical pattern of the annual rings and the compression wood. (Neuenschwander et a1.1997,
with permission)
For better understanding of the differences between the velocities measured

in veneer and in particleboard layers, we suggest that the initial hypothesis
should be improved. It was originally suggested by Grimberg et al (2000)
who considered the layers as isotropic solids. A better approximation can be
obtained with the hypothesis of transverse isotopy for the mechanical pro-
perties of the veneer. For industrially produced veneer sheets, the reference
anisotropic plane is the plane LT. The same hypothesis can be used for the
Applications 211
a) 6cm
'n
<:>
5 10 10 10 4 (em)
b)
•••
:."
:-"
:.•
:.•
.•
'.
:.'a
.
:•
.,.'
."
""
"
:•
"
,.."
.
"
.'0-
"
•
:
"
"
"
."•"
Fig. 5.23a,b. Imaging of delamination in a composite with Lamb waves, The composite was
produced with two beech veneer sheets glued to the two faces of a particleboard layer. a Position
of the delaminated zones. b Image of the delaminations obtained with Lamb waves. (Grimberg
et al. 2000, with permission)
20mm
1000 mV 5100 mV
Fig. 5.24. Imaging of holes in medium density fiberboard of 18-mm thickness in C scan mode,
by the immersion technique with O. S-MHz transducers. The image was superimposed on
the geometry of the specimen. The position of the second hole from the right is well defined
(Neuenschwander et al. 1997, with permission)
particleboard, where the elastic constants in the direction of the board for-
mation and in the corresponding perpendicular direction are similar, and only
the elastic constant thickness direction of the board is relatively low (Bucur
1992).
The ultrasonic imaging of voids in medium density fiberboards of 18-mm
thickness is shown in Fig. S.24. The voids were simulated by holes of different
diameters bored in the fiberboard. The image was obtained in C-scan mode,
by a transmission technique, with the specimen immersed in water and with
broadband transducers of O.S-MHz central frequency. In the future, for on-line
quality assessment, research on the development of dry or noncontact scan-
ning techniques in a large frequency range must be performed.
5.4.4 Imaging of Defects in Wooden Poles
The nondestructive inspection of wooden poles in service for electrical power

distribution, for communication and for housing is a very important safety
measure. The traditional inspection method is tapping with a hammer to gen-
erate a sound. The interpretation and quality of the decision about the service
life of the pole depends on the ability and experience of the inspector. X-ray-
computed tomography can also be used, but is not easy to handle on a large
scale and is hazardous for the operator and the environment. As an alternative,
Summary 213
ultrasonic inspection is of great interest for electrical companies all over the
world because of its handling ease and safety. The ultrasonic propagation char-
acteristics measured on poles (Arita and Kuratanik 1984) and especially the
tomographic imaging (Tomikawa et al. 1985), readily interpreted by the field
operators of electrical companies, are excellent alternatives to the old percus-
sive hammer method.
The attempt to reconstruct ultrasonic tomographic images is based on the
assumption that the ultrasonic waves propagate in a straight line and that the
sound speed greatly decreases in zones where decay is present. The fan-beam
capture of ultrasonic time-of-flight was used and 324 points were employed to
reconstruct the image using 78-kHz transducers, for which the wavelength was
2 cm. This experimental arrangement was completed with 97 projection data
of parallel beams, obtained by interpolation of the fan-beam arrangement. This
large number of data was deconvoluted using the technique of filtered back-
projection. The computational time for producing an image was about 1 h,
which is too long for practical use. To reduce the time requirements, the recon-
struction of the images must be done in hardware. It is necessary to develop
an on-line system that enables rapid reconstruction for in-situ pole inspection.
5.5 Summary
The main purpose of this chapter has been to identify the breadth of applica-
tions of ultrasonic tomographic reconstruction techniques of wood at macro-
scopic levels. Ultrasonic tomography is a diffraction type tomography that is
noninvasive and is safe at low energy levels. Different types of ultrasonic waves
can be used for wood imaging, but the most common are bulk longitudinal
waves. Ultrasonic images can be reconstructed from all characteristic para-
meters of the wave: time of flight, amplitude, frequency spectra of the wave-
form, the phase, etc. The energy distribution and energy flow are important
parameters for enhancing the image contrast. In this chapter, we have dis-
cussed very recent results related to tomographic imaging with the time of
flight of ultrasonic waves.
There are three main types of algorithms that can be used to form tomo-
graphic images from ultrasonic data: transform techniques, iterative tech-
niques and direct inversion techniques. For the algebraic reconstruction
algorithms (ART), each equation corresponds to a ray projection. The sums of
the computed rays are a poor approximation of the measured ones, and the
image suffers from significant noise. The SIRT (simultaneous iterative recon-
struction technique) reduces the noise of ART and produces better images. The
factors that limit the accuracy of the images obtained with diffraction tomo-
graphic reconstructions are related, firstly, to the theoretical approach to the
approximations in the derivation of the reconstruction process and, secondly,
to the experimental limitations. High-resolution images have been obtained
for small, clear specimens of different species, for standing trees, for lumber
and for wood-based-composites. Two-dimensional imaging requires advanced
signal processing and modern digital computers.
The equipment for ultrasonic imaging can operate in mode A, Band C in
contact and non contact (or air-coupled) scanning, using frequencies ranging
from 50 kHz to 5 MHz. As the frequency increases above 1 MHz, the image res-
olution increases, as well as the signal attenuation. The resolution inherent in
acoustic images is basically determined by the beam diameter and by the pixel
size.
The piezoelectric transducers used for ultrasonic tomographic imaging can
operate in direct transmission mode, inducing bulk or surface waves. Conven-
tional piezoelectric transducers in the range of 50 kHz to 1 MHz are commonly
used for data acquisition. Air-coupled ultrasonic transducers are mainly used
for inspection of wood based composites like veneer or low thickness fiber-
board and particleboard. Ultrasonic tomography has a very large field of appli-
cations in decay diagnosis of standing trees, in parks and public gardens in
the imaging of lumber structure to reveal knots, grain deviation, cracks and
compression wood and in imaging of defects in wood composites, such as
delaminations between layers and voids.
6 Nuclear Magnetic Resonance
6.1 Introduction
Nuclear magnetic resonance imaging was first developed for medical applica-
tions. However, since 1980, an increasing number of researchers have reported
applications of this technique to materials science using standard NMR spec-
trometers, in which modifications had been made, and computing facilities had
been adapted to produce images. Every measurement in conventional NMR
spectroscopy, such as chemical shift, nuclear spin relaxation times, dipolar cou-
plings and the spectrum of translational motions can, in principle, be trans-
lated into a spatial resolved contrast and used to produce images. The major
impediment to the extension of this technique was and still is the price of the
equipment.
Nuclear magnetic resonance imaging offers to wood science and wood
processing a nondestructive, noninvasive and noncontact technique to study
wood structure characteristics, such as growth rings, the sapwood/heartwood
boundary, rays, reaction wood, resin canals, knots (Bucur 1990), wounds,
wetwood, decay by fungi, etc. Images obtained from magnetic resonance pro-
perties of iH, \3e or other nuclei provide a tool for in vivo analysis of trees.
This technique is useful in evaluating normal events in the life of trees as well
as different perturbations induced by external factors, such as climate, diseases,
etc. Tree vitality as indicated by the sapwood/heartwood ratio can also be
studied. Water distribution within the growth ring, in roots, branches and
leaves, diurnal or annual periodicity of the water content in the stem can be
determined in vivo. Wood drying is an important field of application of mag-
netic resonance imaging. Another research aspect is related to the distribution
of preservatives in wood or to the decomposition of wood by fungi and/or
other pathologic agents able to induce a modification of the moisture content
of wood (Birkeland 1990; Kucera 1990). When combined with localized spec-
troscopy, nuclear magnetic resonance imaging provides a unique experimen-
tal tool for the study of the structure of wood.
Knowledge of the fundamentals of nuclear magnetic resonance technique
is critical to the productive use of this technology for wood, which is a porous
material. The physics of magnetic resonance phenomena is well understood
and is described in various reference books. Amongst these books and review
articles, one can mention the books written by Abragam (1961), Talpe (1971),
Slichter (1978), Gadian (1982), Mansfield and Morris (1982), Schau (1984),
Callaghan (1991), Gil and Neto (1999), and Lambert et al. (2000). The princi-
pal phenomena are identified in the literature by their initial letters such as:
NMR for nuclear magnetic resonance and MRI for magnetic resonance
imaging. This chapter reviews methods and applications of the NMR technique
to wood considered here as an anisotropic porous material, with emphasis on
imaging and spectroscopy techniques developed specifically for wood.
6.2 Basic Aspects of NMR Imaging
NMR is an important technique for measurement of wood properties that pro-

vides a number of bulk parameters (moisture content distribution, wood
porosity, proton diffusion and absorption as a function of time, chemical com-
position, chemical shifts that gives information about the position of protons
in the samples, etc), that cannot be provided by any other methods. In addi-
tion, by using pulse NMR techniques, one can obtain additional important
information about the diffusion and flow of moisture in wood.
A detailed presentation of the theoretical background of the NMR technique
is far beyond the scope of this book. In this chapter, we shall give a brief
description of the basic properties of atomic nuclei, explaining their interac-
tion with the applied magnetic field.
In this book, the term nuclei is used synonymously with nuclear magnetic
moment.
The NMR properties of the nuclei commonly used in wood science are given
in Table 6.1. High resolution experiments utilize IH, 13C and 31p which have a
spin quantum number equal to 1/2. The distribution of hydrogen or other
atoms and their relaxation time in a strong magnetic field lead to differences
in the NMR signal used for the reconstruction of the images.
Some atomic nuclei, such as IH, 13C and 31p, possess an angular momentum
or spin and an associated magnetic moment. The spin can be visualized as the
spinning motion of the nuclear magnetic moment about its own axis. The mag-
netic moment determines the couple on the nuclei that tends to align it with
Table6.1. NMR properties of some nuclei (Gadian 1982) (© DG Gadian 1982, reprint from
Nuclear Magnetic Resonance and its applications to living system by David G Gadian (1982) by
permission of Oxford University Press)
Nucleus Spin-quantum Resonance frequency Relative sensitivity

number (MHz) at 5 tesla at constant field
IH 1/2 213 100

l3e 1/2 53.5 1.6 x 10-2
31p
1/2 86.2 6.6
Basic Aspects of NMR Imaging 217
the applied magnetic field. In the absence of a magnetic field, the average
sample magnetic moment produced by the nuclear magnetic moments is zero
because the spins are randomly oriented.
Quantum mechanically, a spin 1/2 nuclei can only align parallel or anti-
parallel to the field. Boltzmann statistics gives the fraction of nuclei in each
state with more nuclei in the lower energy state, with the nuclei magnetic
moment parallel to the field, and anti-parallel to the field direction There will
therefore be a net sample magnetic moment in the direction of the field.
Because of their angular momentum, the nuclei are subjected to a torque and
they precess like a gyroscope around the direction of the field. The frequency
of precession is the Larmor frequency 0)0 and is proportional to the gyromag-
netic ratio of the nucleus "I and field Bo as defined by:
(6.1)
The frequencies associated with conventional NMR techniques are usually
in the range of 100 to 600 MHz. NMR frequencies can be identified with the
absorption of microwave photons of energy h 0) (where h is the Plank's con-
stant = 6.6255 x 10-27 erg. s = 6.6255 x 10-34 N ms), causing transitions between
the energy levels associated with the quantum mechanical states of the nucleus
in a magnetic field as:
~E=h "(Bo (6.2)
where ~E is the difference in energy between the two states.
The nuclei resonance frequency is therefore directly proportional to the
local magnetic field experienced by the nuclei. However, the applied Bo field
also induces electronic currents in the surrounding atoms, producing a further
small shielding or screening field O"Bo. The effective field Beff at the nucleus is
therefore given by:
Beff =B o (1-O") (6.3)
0" is known as the shielding or screening constant and depends on the
atomic environment of the individual nuclei in a chemical compound.
The NMR frequency for an individual nuclei is therefore
0)0 = "I Bo (1- 0") (6.4)
where 0" has typical values from 10-6 to 10-3 •
Different nuclei in a chemical structure will therefore have slightly
different NMR frequencies, depending on their chemical environment.
Resonant frequency is defined by the chemical shift 0 (in ppm = part per
million) and is expressed as a fractional change in the frequency measured.
For example, the NMR spectrum of acetic acid CH 3COOH gives rise to two dis-
tinct frequencies for the protons in the CH3 and COOH parts of the molecule.
The zero value of the chemical shift for IH and 13C spectra is defined by
comparison with measurements on tetramethylsilane. The chemical shift is
independent of the magnetic field strength. The application of a small radio
frequency field at the Larmor frequency, perpendicular to the static, applied

field causes the spin to be excited between the energy levels of the nuclei in
the static field. This leads to the absorption of power at the Larmor frequency.
For IH nuclei in a field Bo of 9.3 tesla an ac Bl field at a Larmor frequency of
400 MHz is required to produce a resonance spectrum.
As can be seen from Fig. 6.1, in a stationary coordinate system, the momen-
tum Mo has two components, Mz and Mxy rotating in the XY plane at frequency
(0. If a new magnetic field is applied in a direction perpendicular to Mxy, the
whole system will rotate and will have the new coordinates X', Y' and Z'. The
rotating magnetic field is produced by a radio frequency (RF) oscillator. No
effect is seen on Mxy, but a torque is exerted on Mz that will thus precess about
the X' -axis and about the applied field Bo direction. Rotation of Mz will induce
a voltage in a detector coil situated along the Y-axis, and resonance will be
detected.
The maximum induction signal, induced in the detector coil, will occur
when the pulse of RF energy at the Larmor frequency is sufficiently long to
rotate Mz at 90 into the XY plane. On removal of the Bl field, the nuclei will
0
precess with their spin perpendicular to the applied static field Bo. However,
the nuclei will experience small local fields that alter Bo. These local fields are
both positive and negative, so that some spins precess faster and some slower
and are no longer in phase. Their magnitude reduces to zero as exp( -t/T z ),
where Tz is the transverse relaxation time.
The thermally induced fluctuation in the local field associated with the
motion of atoms also produces components of the local field at the Larmor fre-
quency. This causes net magnetization. Mz returns to its equilibrium value as
Mz = Mo[1 - e(-tlT1)j, where Tl is referred to as the longitudinal relaxation or the
spin-lattice relaxation time.
The characteristic relaxation times for a decay of the Z and XY components
of the magnetization T I and Tz are strongly dependent on the environment of
the excited spins. The effects of relaxation and precession are combined in the
Bloch equations describing NMR (Abragam 1961). In general, Tl differs from
Tz because Tl involves a transfer of energy from the spin system to the sur-
roundings, while Tz describes the energy concerning loss in phase coherence
within of the spin system.
Fig.6.1a-d. Macroscopic description of magnetic moment of the nuclear spin system. a Nuclei
are in equilibrium with the static magnetic field Bo. Mo is the overall magnetic field vector; Mz
is the static magnetic field along the axis Z; Mxy is the rotating magnetic field in the XY plane;
e is the angle of precession. b Rotation of Mz in the presence the of Bj magnetic field generated
by the Larmor frequency. Mz is tipped away from Z direction by B field. c Loss of phase coher-
j
ence of spins after the extinction of the Bj field, leading to a reduction in precessing Mxy as
MXY(o)e-tiT '. d In practice, decay is observed because nuclei experience different fields in different
molecules and shift the effect that can be measured. The local variation in the field Bj is observed
due to the spin-spin interaction. T, is generated by sample properties (pore size, chemical
bonding of H,molecules)
a) b) Z'
Z
I
, ,.Mo
Bo
static field
Bo 1 Mx
r
o I
~ y I )h/ X,
X X'
d) t:C
c) 1 ~
Bo r;"
90° ~
!4.
'"
j '"o
...,
'" T"i FID Z
s::
:;.::I
time
8
X' i·
OQ
spins components ABC FID=free induction decay
tv
....
\0
Fig. 6.2. J3Carbon cross'polarization/magic angle spin·

89~ 84
ning (CP/MAS) NMR spectroscopy spectra for pine
(Pinus sylvestris) and beech (Fagus sylvatica), determined
105 I! at 75.47MHz I3C frequency, 4.5kHz rotor spinning
148 / t frequency and 1.5-ms contact time for cross-polarization.
~". . /v . . (Gilardi et al. 1995, with permission)
Pinus sylvestris '_."-"'....'-
In the case ofNMR spectroscopy, the investigation ofNMR relaxation mech-

anisms in inhomogeneous solids involves the characterization of decay curves
as a sum of exponential functions. The estimation of TI and T2 is relatively
accurate for two or three exponentials (Menon et al. 1987, 1989; Flibote et al.
1990).
Figure 6.2 shows a typical NMR spectrum of pine and beech with major
differences in the lignin and hemicellulose components. The peaks corre-
sponding to chemical shifts 0 of 105 ppm correspond to proton resonances in
cellulose and are present in both spectra. The peaks of higher intensity at 84
and 89 ppm show that in pine the cellulose crystallinity is higher than in beech.
The lignin peaks are at shifts of 148 ppm (phenolic carbon, with methoxyl
groups attached) in pine and in beech at 154ppm (with methoxylated pheno-
lics). The acetate groups present in the xylan in beech give rise to peaks at 22
and 174ppm.
NMR techniques can be used to determine Tl - the spin-lattice relaxation
time - and T2 - the spin-spin relaxation time in wood. The factors that influ-
ence TI and T2 are the physical state of the wood sample and the presence of
water molecules in the cell wall and lumina (Araujo et al. 1992). The quality
of the NMR signal for wood samples is strongly dependent on the amount of
water present. A significant difference can be seen between free water, water in
the cell lumina, and the bound water, in the cell wall. The contribution of each
subsystem to the total magnetization is called abundance and is proportional
to the amount of protons in the governed subsystem. The abundance increases
with the moisture content in wood and is 100% at the fiber saturation point.
The relaxation time TI of protons in pure bulk water is 3 s while electrically
bonded protons have relaxation times of about 10-3 s.
The spin-lattice relaxation time Tl and the spin-spin relaxation time T2 are
determined by fluctuating magnetic fields and by molecular motions. The time
free water in lumen
I
I T2
I
I
I Tc=lIwo
10- 5 L,-_ _ _ _ _ _-lII..-_ _ _t_im_e_(S_ec_On_d_es-,-l
10- 12 10-9
Tc (s)
Fig.6.3. Relaxation time as a function of molecular correlation time Teat typical frequencies
associated with spin motion. (Pfeffer and Gerasimowicz 1989, CRC Press, with permission)
scale, as shown in Fig. 6.3 for these relaxation processes, can be quite different.
Slow motions in the kHz range cause the spin to lose the precessional phase
coherence determining T2, while protons are associated with fluctuations of the
magnetic field, at the Larmor frequency (typically many hundreds of MHz) are
required to restore the spin population to equilibrium along the Bo direction
determining Tl'
Polymers of molecular weight between 10,000 to 25,000Da may have relax-
ation times on either side of the minimum of the curve, depending on
the Larmor experimental frequency. For molecules having relaxation times
Tc >10-9 s the relaxation times T1 and T2 are different. In contrast to the
relaxation of proton spin magnetization in wood samples, on which one finds a
multi-exponential decay time constants ranging between less than 1 ms to
several tens of ms, relaxation in bulk water is a single exponential process: T1 =
T2 = 2.4 X 10-9 m 2/s at 25°C. Diffusion measurements are of particular interest
in wood science and industry. For this purpose, we have to study the effects of
field gradients and of static inhomogeneities in field on spin-spin fluctuations.
By using field gradients perpendicular to the X, Y or Z axes, one can make
measurements on slices in all directions, using familiar computed tomographic
techniques. Three-dimensional information can be obtained from all points of
the space available for object imaging.
The field gradient is also important in determining the diffusion time
because during the measurements the water molecules diffuse in and out of
the measured section. Only molecules remaining with the section contribute
to the magnitude of the pulse echo in a 90°-180° sequence. This enables the
diffusion coefficient in the direction of the field gradient to be measured. In
practice, for imaging improvement, Chang et al. (1999) demonstrated the utility
of the sophisticated sequence of switching of pulse gradients in all directions.
Interestingly, any reduction in precessing the Mxy component due to field
inhomogeneities can be reversed by applying a 180° pulse as indicated in Fig.
6.4. As the spins represented by the components ABC continue to precess rel-
ative to each echo by the same amount, they all come together at a time 2 t, to
give what is called a spin echo signal. However, alternatively, the effects of a
static inhomogeneity in the field can be reversed. Random processes such as
the spin-spin fluctuation affecting T2 and any diffusion of spins from one
region to another in an inhomogeneous field cannot be reversed. In a homo-
geneous field, this technique enables T2 to be measured, while in a field gradi-
ent the decay enables the diffusion content to be measured along the field.
For spins diffusing in a field gradient, the pulse echo sequence is determined
by the spin diffusion length (2 D t)1I2. If the length is known from the struc-
tural features, we can measure T2* to determine the diffusion coefficient D, or,
if D is known, the diffusion length, which can be related to the dimensions of
cell structure.
The NMR technique can also be used to investigate the diffusion of water
both along and perpendicular to the fibers. This is also achieved by NMR mea-
surements in a field gradient. In practice, during the sequence, the water mole-
cules can diffuse out of the structural region. If an interaction takes place
between the water and the cell wall, the alignment of spins can be destroyed.
This limits the time the spin can remain in a coherent state (the T2 time) so
that the diffusion time can be deduced and the pore size can be determined.
Moreover, a long diffusion time can be related to the diffusion measurement
along the fibers and can be used for the calculation of porosity.
Brownstein (l980), using NMR data published by Hsi et al. (l977) for
wood specimens at 38% moisture content of Thuja occidentalis, explained the
observed nuclear relaxation rates in terms of a strong relaxation. The deduced
relaxation times are shown in Fig. 6.5. For these specimens, the relaxation time
T2has two components, a slow relaxation time T2s of the order of a few tens of
ms and a fast component T2f of the order of a few ms. T2f is identified to be the
same as T\. Similar data were reported by Byrne et al. (l986) for Pinus radiata,
with two values ofT2, 0.8 and 10ms measured at 60 MHz. When, the specimens
were saturated with water, T2s increased considerably, while T2f leveled off at
about three times its initial value. To explain the phenomenon, Brownstein
(1980) proposed a model entailing (we suppose) diffusion of free water present
in the lumina and a monomolecular layer of water near the cell wall (corre-
sponding to a 6% wood moisture content), which is illustrated schematically
in Fig. 6.6.
The model consists of two water layers of thickness a and b with diffusivi-
ties D] and Du, where D] « Du. The strong magnetic relaxation is supposed to
be present only in region I. Calculations demonstrated that in region II the
90° pulse
--~~------------~~~~~--------~r-~ Bt - O
180 180' pulse
____~-4________~~____________~--~------------------~----~r_----~~r;n~
21
"pin ceho ;81101
linle
Fig.6.4a-c. Relaxation time as a function of pulse sequence time. a Reduction of precessing Mxy
component due to field inhomogeneity can be reversed by applying a 1800 pulse. b The spins rep-
resented by the components ABC continue to precess, and they come together at a time 2 t, cor-
responding to the spin-echo signal. c For spins diffusing in a field gradient, the pulse-echo
sequence is determined by the spin diffusion length. If the diffusion length is known from geo-
metric structural features, one can determine the diffusion coefficient from T,* measurements
Fig. 6.5. Theoretical and experimental data
80' o
of TIf and T" as a function of moisture
o content. (Brownstein 1980, Academic Press,
70
T2s(ms) o with permission)
60 o
50
o
40
o
30
o
. data
20 • p
o theory
10
OL-------------------------
t3
T1f(ms) 000000
i:,.O"d 0 c
2
.....'19-
00 0.2 0.4 0.6 0.8 1.0 1.2 1.4
gram Hplgram wood
free water in cellular lumen
a) bound water in cellular wall
b)
water in lumen cellular material
--+II--~~_-_-_-_-+I=a=====.I-;-b-- X
Fig.6.6a,b. Physical model proposed by Brownstein (1980) for a wood cell, to explain the diffu-
sivity of water in wood, with 1-D geometry of the density of proton magnetic moment. a Free water
and bound water in a cell. b Diffusion model. Region I is defined by the thickness x = a and diffu-
sivity D] and corresponds to free water in the wood, which corresponds to 0.38 g H,O/g wood. When
only a monomolecular water layer is present, we have D]« Dn; a2/D] = 25.8 ms and a $0.26 /lm. The
diffusion process is inhibited within this layer. Region II is defined by the thickness x = b and the
diffusivity Dn and corresponds to the bound water in the cell wall. The proton distribution is con-
stant and Dn» D] and Dn = 2.4 x lO-scm'/s. (Academic Press, with permission)
Table 6.2. Diffusivity (D) in some species determined with the NMR technique, compared with
classic techniques
Species Units D parallel D Perpendicular Ratio Reference
Douglas 1O-9 m 2/ s 1.7 0.88 1.93

Maple 1O-9 m 2/ s 1.2 0.36 3.3 MacGregor et al. (1983)
Fir 1O-lOm2/s 3.0 0.28 10.7 Choong (1963)
Theoretical 1O-lOm2/s 5-40 Byrne et al. (1986)
proton density distribution was a constant for each mode under consideration
and that the relaxation time was dependent on a2/DI> corresponding to the dif-
fusion term for water molecules across the thickness a. This parameter deter-
mines the time scale and "does not affect the form in which the relaxation time
varies with the ratio b/a:' The model requires that the ratio of the two diffu-
sivities DdDn = 1/900. If Dn is taken to be 2.4 X 1O-9 m2/s, for water, the para-
meter a2/DJ has the value of 26ms and consequently a < 0.26~m. This means
that region I consists of rather a thin layer of water molecules.
Like all physical parameters of wood, the diffusion of water is anisotropic.
MacGregor et al. (1983) calculated the diffusion coefficient of water molecules
under equilibrium conditions (static or self-diffusion) as a function of the
angle e between the longitudinal anisotropic axis corresponding to the cell
axis and the field gradient (Fig. 6.7).
The coefficients parallel and perpendicular to the cell axis are given in Table
6.2. The ratio between the two coefficients reflects the anisotropy of wood.
Water diffusion in sugar maple is more anisotropic than in Douglas fir. In the
direction parallel to the cell axis, the diffusion coefficient is smaller than that
of bulk water, probably because the water in wood (sap) contains different
magnetic ions (Mn, Fe, eu) and is different from bulk water and also because
the cell walls restrict diffusion.
Lindgren (1994) summarized in a very concise manner, the principles ofthe
NMRmethod:
1. the random orientation of the magnetic moment of hydrogen nuclei in the
absence of a field;
2. the lining up of nuclei when a magnetic field is applied;
3. the 90° tips on magnetization on applying a radiowave field;
4. the removal of the radiofrequency signal, and subsequent generation of a
radiowave by the precessing tipped magnetization frequency and amplitude.
Several methods for inversion recovery to measure Tl and T2 are used,
including saturation recovery and progressive saturation. The following
describes the inversion recovery pulse sequence as given by Hailey et al. (1985).
It was noted previously that the magnetic moment precesses at a frequency
proportional to the local magnetic field strength. The tipping at 90° and at 180°
depends on the strength and duration of the pulsed radio frequency signal. A
l.5rr---,----,---.,---~~____,
,-.,
1.2
OJ
0)
en
N---E 0\
~0.9
-
tr)
0
><:
0
0.6
a)
0.3
0 40 80 120 160 200
8 (degrees)
2.0 , , , 1
Sugar maple
1
~ ;-
-6 1.2 r
0
o
1
.6
0.8 u''--_-'-'_ _ _.l.....-I_---1'_ _ _-'--

b)
,_---I
-
o 40 80 120 160 200

Fig.6.7a,b. Diffusion coefficient vs. the orientation angle of the specimen for a Douglas fir and
b sugar maple. The theoretical values were deduced for Douglas fir with DI/ = (1.7±O.1) 10-5 cm'/s.
D~ = (O.88±O.OS) 1O-5 cm'/s and for sugar maple with DI/ = (1.2±O.l) 1O-5 cm'/s D~ = (O.36±O.OS)
1O-5 cm'/s. (MacGregor et al. 1983, with permission)
particularly important sequence of pulses CPMG is named after the inventors

Carr and Purcell (1954) and Meiboom and Gill (1958), cited by Abragam
(1961). This sequence (Fig. 6.8) consists of a 90° RF pulse, tipping the Mz into
the XY plane (Fig. 6.8b) magnetization precess, around the field direction (Fig.
6.8c) growing an "induction signal" in an RF pick-up coil with its axis in the
plane. However, the fluctuations in the local field result in small local changes
in the local field, leading to a slightly different time, varying precessional fre-
quency in different microscopic regions, so that the individual components
lose their phase coherence resulting in a reduction in amplitude amounting to
J~
;' t= T
y
x F
('
;kt
__
\ .....
X
'y
- ,,/
--- +x
G
-x
a)
!r~O'
I
,~ ,' ,
I,' 10,I, Il'r I'ru liT'
1' 0
, ' , i~ ' ~i • t
A 0 D l' H 21' 31' 41' 51' 61' 71'
l ~Jll:-~--Ir---'-----
o~,4 T 2T 3T 4T 71'
• t
B F J
900 x -~ -180 0 y-1' -1800 y-1' -180 0 y b)
Fig.6.8a,b. Carr-Purcell/Meiboom Gill sequence. a Spin isochromatic representation. b Pulse

sequence and NMR signal for T,measurements. (Hailey et a1.l985; courtesy of Forest Products
Society)
This is a random process and cannot be reversed. However, in practice,

e(-t/T, ).
the applied fields are never completely homogeneous and therefore result in
dephasy, giving an induced signal recovery as e( - tlT *, ) where T* 2 depends on the
homogeneity of the field.
The free induction decay (FID) curve describes the decay of the induced
signal arising from free precession of the nuclei in the field Bo. The Fourier
en
~~~~~--+---+---+---~--~--~---r--~§
~.~~--~~~--+---+---~--~--~---r--~1 .-5~
~~~~~~~=r~~~~~~~I~ , 0)
~~--~~--4-~~~~~-+--+-~I~
I
r- 1"0..
: ~
. . . . . . . . . . . . . . . . . . .time
2 3 4 5 6 7 8 9 x SOO/lS
Fig. 6.9. Free induction decay (FID) amplitude of 'H nuclei in maple with different moisture
content as a function of time following a 90° pulse. Note that the signal increases in amplitude
and the relaxation rate decreases with moisture content. (Sharp et al. 1978; courtesy of Wood and
Fiber Science)
transformation of this decay gives the conventional NMR spectrum, in which

the signal amplitude is plotted as a function of frequency. If the nuclei are
present in different environments, the resulting FID contains contributions
of different frequencies, each component having its own characteristic time
constant.
An example of free induction decay versus time is given in Fig. 6.9, for maple
at different moisture contents above and below the fiber saturation point.
Figure 6.9a shows a very short relaxation time <20 fls for the sample at 6%
moisture content and a long relaxation time >500 fls for wood at 30% moisture
content. Different moisture contents give signals of different sizes and differ-
ent relaxation times can be measured. The wood moisture content therefore
can be related to the FID signal. Sharp et al. (1978) showed that the wood mois-
ture was proportional to the FID, as shown in Fig. 6.10. (It should be noted that
the amplitude of the free induction signal just after a pulse is simply propor-
tional to the number of protons.)
The amplitude of the free induction decay signal can be used to obtain data
relating to the mobile and less mobile components of the protons. An example
is given for western red cedar, for which the FIDs were acquired with a 90° RF
pulse sequence and with 5 s between pulses. Note that the measurements
depend on the time between pulses. The free induction decay curve for three
levels of moisture content in western red cedar is shown in Fig. 6.11. One can
describe these results in terms of the FID from protons directly associated with
1.0
.el
'\::
::l
.rj
ta
"@
.§
en
0.5
Q
~
.......
0 o maple
"0
(1)
• spruce
::l
.'=
0..
§
0 50 100 150 200

Fig. 6.10. Amplitude of FID signal measured at 50~s as a function of moisture content. (Sharp
et al. 1978; courtesy of Wood and Fiber Science)
western red cedar

.g FID
.€ ~horter '2
1 92% M.C.
~
lager T2
.~
"@ - - - - - -
signal from wood and water 54% M.e.,
o
~ \ ~ signal from cellular wall 0% M.C.
I Ii """"------- mobile protons (H20) - - - - - - - l..~ Time

less
mobile
protons
Fig.6.11. Free induction decay at three moisture content levels. (Hailey et al. 1985; courtesy of
the wood structure with a superimposed additional signal from the absorbed
water. No mobile protons were detected on the signal corresponding to 0%
moisture content. The mobile protons of water in a sample of western red cedar
of 25% moisture content can be quantified on the graph as the intersection of
the profile of the mobile protons with the axis of the signal amplitude.
Menon et al.(l987) measured the IH NMR free induction signal for western
red cedar sapwood at three moisture contents: 216,92 and 54%, and observed
three relaxation times (Fig. 6.12). The very rapid component at 54% is related
to Tl and is <lO/-1s. Two components were measured for T2, a fast one, decay-
ing at T2 fast < 100/-1s and a slow one decaying at T2 slow < 200/-1s. The intensities
lOIr-------------------------------~
'-" "
Western Red Cedar
\.
p ....
'~lOo \. ....
<J) •
.5 "" '216%
tl '.
:::s
~
u 92%
WI
y 54%
I I I I
0 100 200 300 400 500

a) time (msec)
Douglas Fir
., •• '99%
47%
o 100 200 300 400 500

b) time (msec)
Fig.6.12a,b. CPMG decay curves. a Western red cedar at 216, 92, 54% moisture contents. b
Douglas fir at 99 and 47% moisture contents. N.B. The logarithmic intensity scale that could give
a straight line for a single component exponential decay. (Menon et al.1987,Academic Press, with
permission)
of CPMG data normalized versus the solid proton signal from FID at each
moisture content were plotted versus time for western red cedar and for
Douglas fir. The authors detected three components of the relaxation time on
visual inspection of the slope of the regression line. The spin lattice relaxation
curves are decaying nonexponentially except at the highest moisture content.
It should be noted that the protons near the surface interact with the cell walls.
T2 becomes a measure of pore size.
The spin-lattice relaxation time T J can be determined from an inversion
sequence illustrated in Fig. 6.13. Immediately following the magnetization a
180 0 pulse is reversed. The magnetization then reverts to its equilibrium value
as Mz(t) = Mo[1-2 eH/TIl ]. This defines what is known as the longitudinal relax-
z z z
a) +~
~Y 1 - - -- Y A-"!,- - -Y
\ :
\'
x t=O x x J
A C M0
z z
_."MO
/ "1
, Y
..r--"t-- Y
\ I
x ~ x
D Mo F
b)
I
'" 180° x
1!~ ';
~~c:%~·~--~-
I
90 0 -x
"....,"">"l;me
------=--------'
~r~--~r~----~---
~ I'" I"'",
180° x
~),~·~--~~~II~'m-~
90 0 -x
A E A E
Fig. 6. 13a,b. Inversion recovery pulse sequence and NMR signal for TI measurement. a Inversion
recovery spin isochromatic representation. b Pulse sequence. (Hailey et al. 1985; courtesy of Forest
Products Society)
ation time T 1• This time can range from ms in metals to several hours in insu-
lators at low temperature and is determined by local fluctuation in the field
experienced by the individual nuclei but at the Larmor frequency. This decay
depends on the nuclei concerned with larger fluctuations in a liquid than in a
solid. For example, the nuclei in the water of wood comes into equilibrium
faster than nuclei associated with protons in the cellular structure.
Observation of the amplitude immediately following a subsequent 90° pulse,
enables T 1 to be measured and is convenient for varying the recovery time 'to
This assumes that at the start of the sequence M z(t) equals M equilibrium, but
if T 1 is very long this may not be true. To make measurements on a specific
nuclei subsystem, the time between pulse sequences 't must be longer than the
time Tl (Tl > "[). If the time between pulse sequences is shorter than the relax-
ation time ("[ < T 1), no signal from the nuclei is observed.
For the measurement of T2, more sophisticated pulse sequences (90°, 180°,
180°, etc.) than for TI are used. The presence of the component relaxation times
T I and T2 is attributed to the different mobility of water molecules in different
structural regions, as for example the free water in the lumen and the bound
water in the cell wall. Molecular processes and the total signal intensity depend
on these two times as well as on spin density and proton mobility.
Since the FID signal is proportional to proton density, the moisture content,
MC, in wood can be deduced using
(6.5)
where MC is the moisture content of wood measured with the NMR technique;
PHwood is the proton density in wood calculated from the amount of cellulose
and lignin ; PH water is the proton density in water; Mo is the zero intercept of
the FID curve of the mobile component So is the zero intercept of FID of the
total signal component. The proton density is expressed as the number of
protons per gram. The ratio PHwood/PH water is 0.549 for western red cedar and
0.538 for Douglas fir.
The agreement between the NMR moisture content measurements [MC]
and routine oven-dried moisture measurements [MC] has been shown with
the regression equation calculated for the sapwood and the heartwood of
western red cedar and of Douglas fir, which is R2 = 0.9916
MC = 0.9791 MC-0.7561 (6.6)
This agreement is a milestone in the further application of NMR techniques
in wood science since it demonstrates that it is possible to measure directly the
moisture content of wood without oven-drying the sample. From such data an
upper limit on the difference in moisture content between the heartwood and
the sapwood of the same species can be established.
Because nuclei in different environments decay with different characteris-
tic time constants, Olek et al. (1994) used NMR protons to distinguish bound
water from free water in birch. The spin-spin relaxation time T2 was measured
with a pulse spectrometer at 30MHz and with the CPMG method. The time
between It pulses was 200).is for samples with a moisture contents close to the
fiber saturation point and 5000).is for samples with an even higher moisture
content. The repetition time between pulse sequences was 5 s. The multi-
exponential spin echo decays were calculated as a function of a two-
component relaxation functions, such as:
A(t) =AAexp(-t/T2A )+A Bexp(-t/T2B) (6.7)
where A(t) is the amplitude of spin echo at time t; T2A, T2B are the spin-spin
relaxation times of the spin subsystems and A Aand AB their amplitude con-
tributions (abundances) of the subsystems.
10~~~~~~~~~~~~~~~
8 77 % moisture content
:> 6 "
o
~4 0 00
.€
-a 00
00
§2 o
000
o 0
o
1
0 10 20 30 40 50 60 70 80 09
a) time (ms)
6.0
0
4.0 0 10% moisture content
o 0
:> 2.0 0
OJ 080
~ b:g ao
i
0
o 0
O.6 0
0.4 000
0.2 0
0.1
0 2 3 4
b) time (ms)
Fig. 6. 14a,b. Spin-echo amplitudes versus time in birch above (a 77%) and below (b 10.4%) the
fiber saturation point. (Olek et al. 1994, with permission)
The spin echo amplitudes versus time for a saturated specimen (77%) and
dry specimen (10%) are shown in Fig. 6.14. The spin echo's decay with time is
in the range 0-90 ms. The sample with 77% moisture content clearly shows two
distinct relaxation components corresponding to T2A, 60 ms and to T2B, 100 ms.
For the dry specimen the decay is also non exponential and characterized by a
relaxation time «4ms).
The moisture content MC above and below the fiber saturation point can be
expressed as a function of the corresponding relaxation time T2 (respectively,
T2B and T2A ) and is then calculated with:
T2 = MC/(a+ b MC) (6.8)
where a and b are empirical constants determined for birch as: a = 2.481 and
b = -0.0124. above the fiber saturation point and a = 18.67 and b = -0.1503
below the fiber saturation point and MC is the gravimetric moisture content
of wood, determined by routine measurements.
The relationships between the abundances AB and AA and the moisture
content above the fiber saturation point are given in Fig. 6.15. The abundance
100 <=:C",~~;-r-~;-;-~~;-r-;-r-r,---,~,~,~,~,,-,'~
90f 0 j
:
80[ 8 ~
~ ~gt
0..:r: 50 _ o
..:r: 40 AA =-1.281·MC+136.9 o
o
30
i~ t ,,,r~~.,9~, I , , , , I " ,I "
30 40 50 60 70 80 90
a) moisture content (%)
100~~r,---~r,---;-r-r,---;-;-r,---,~,~, " ~,;-r-r,---~~

90
80 AB=1.281·MC-36.91
70 r=0.96
o
~ 60 o
° 50-
..:r:~ 40
30
20
10 o
O~~~~~~~~~~~~~~~
30 40 50 60 70 80 90
b) moisture content (%)
Fig. 6. 1Sa,b. The abundance AA of the subsystem A and the abundance AB of the subsystem B vs.
the moisture content in the range 30 to 90% moisture content. a AA of the subsystem A vs. mois-
ture content. b AB of the subsystem B vs. moisture content. (Olek et at. 1994, with permission)
AB deduced from the fast decaying component of spin-echoes decay, T2B ,

increases with the moisture content (the correlation coefficient is R = 0.96),
and the regression line passes through o. The abundance AA deduced from TA
decreases with moisture content. By extrapolation, the maximum moisture
content of birch can be associated with a relaxation time T2B of 75ms. This is
much less than the relaxation time of bulk water, which is several seconds. The
abundance AA increases while AB decreases with increasing moisture content.
Above the fiber saturation point, the NMR signal is therefore produced by the
protons of bound water.
Above the fiber saturation point, T2A (Fig. 6.16) and T2B increase with mois-
ture content. Water molecules of the subsystem with relaxation time T2B (> TA)
have a restricted mobility compared with the molecules of bulk water. The
same subsystem has also the longest observed spin-spin relaxation time and
it is expected that this system corresponds to magnetic relaxation of the water
molecules filling the cell lumina (free water) albeit with a reduced relaxation
time due to the cell wall. Probably, the capillary forces restrict the mobility of
the water molecules with the relaxation time T2B. The time T2A is shorter than
14~~~~~~~~~~~~~~
12 MC o
~1O
T2A 18.67-0.1503'MC
'" 8': r =0.99
S
~ ';
..:;
f-;<'1 6
4
0
30 40 50 60 70 80 90
a) moisture content (%)
70
60 MC 0
T2B 2.481-0.0124·MC (1'0
50 000
r =0.98
40
30-
g'" 20'
~
E-< 10
0
30 40 50 60 70 80 90
b) moisture content (%)
Fig. 6. 16a,b. The spin-spin relaxation time T2A and T28 vs. the moisture content above the fiber
saturation point. a T2A vs. the moisture content. b T28 vs. the moisture content. (Olek et aI. 1994,
with permission)
Table 6.3. Spin-spin relaxation time (ms) expressed by fast, slow and intermediate components
in birch wood as a function of the moisture content. (Data from Olek et aI. 1994, with
permission)
Moisture content Fast component Slow component Intermediate component
Above fiber saturation point 2-14 15-60

Under fiber saturation point 0.4-2.8 2-10
T2B and varies in the range 2 to 17 ms and also increases with the moisture
content above the fiber saturation point (Table 6.3). The low spin-spin relax-
ation time T2A indicates that the very low mobility of the water molecules asso-
ciated with this system, is mainly related to bound water.
Below the fiber saturation point, we have seen that the decay of spin echoes
is nonexponential. Two relaxation times were measured, the first T2A, ranging
between 0.4 and 4 ms and a relatively slow time called T2C, ranging from 2 to
10ms. This time is shorter than the time involving capillary water. Both T2A
(Fig. 6.17) and T2C (Fig. 6.18) increase with the moisture content. For the rela-
4.4 ,-r-rrl,,--r---..,,--.-r.--r,,'-'.,---rrl-.--r---..-'-'-,
4.0:" MC
3.6 . T2A l4.76-0.2544.MC
~3.2. r =0.87
S 2.8 :-
<2.4
(',12.0
E-< 1.6 a
0 00
1.2 o
0.8
o
0.4
0.00Ld::.J-"-L--L5-'-'-.L-Ll'-0L.L.J-"-L15-'-'-.l....L2'-0'-'-L..L..1
2 5.-L-L--L..l....J
30
Fig. 6.17. The spin-spin relaxation time T'A vs. the moisture content below the fiber saturation
point. (Olek et al. 1994, with permission)
16
14 MC
T 2C
12 4.9l5-0.1009·MC
~1O r=0.81
8
u8
~6 co
4 0
0 OJ
2
00 5 10 15 20 25 30
Fig. 6.18. The spin-spin relaxation time T,C vs. the moisture content below the fiber saturation
point. (Olek et al. 1994, with permission)
tionship between T 1C and the moisture content, the correlation coefficient is

positive and relatively high r = 0.81. At zero moisture content, the relaxation
time was estimated to be zero. The relationships between different relaxation
times are: T 1C < T1B and T 1C > T 1A , and for all terms, we have T1B > T 1C > T 1A •
Because T 1C > T 1A , it was supposed that the mobility of the molecules of the
system is much higher than that of the molecules of bound water and con-
sequently, the corresponding NMR signals could be produced by condensed
water molecules in very thin anatomic elements of wood. Another explanation
was advanced by the authors who suggested the existence of a very thin
monomolecular layer of water on the cell wall. This water determined the
increase in all mechanical properties of wood in the vicinity of 5% moisture
content. From all experimental data (T 1B > T 1C > T 1A ) presented by Olek et al.
(1994), the weight of evidence suggests that bound and free water relax
separately and that the corresponding relaxation times can be precisely
measured.
The NMR Imaging Technique 237
6.3 The NMR Imaging Technique
In this section, we will outline the NMR imaging technique, with special stress
on what the images represent and on the algorithms used to identify and isolate
particular structural features of wood.
6.3.1 Techniques for Imaging
The magnetic resonance technique can image nuclear spin density, NMR relax-
ation times and diffusion of water molecules at various points throughout the
wood sample. The values of relaxation time T1 and T2 are important in deter-
mining the quality of an image and can be used in selective imaging of bound
and free water molecules in a wood sample under certain conditions. The
intensity can be proportional to the product of the proton density and the ratio
of the relaxation times. High resolution NMR imaging is very sensitive to the
more mobile protons. The dependence of the image intensity on relaxation
times is very important in differentiating between wood tissues with the same
proton density but different values of the ratio of the relaxation times. In
general, the quality of an image depends on many parameters. It is also possi-
ble to vary the experimental conditions in such a way that the image is par-
ticularly sensitive to a specific chosen parameter. The spatial discrimination is
dependent on the magnetic gradient field.
The techniques currently used in NMR image reconstruction are: back pro-
jection, similar techniques used in X-ray computed tomography, rotational
frame, echo planar for two-dimensional images and two-dimensional Fourier
transform techniques for three-dimensional imaging of the object. The Fourier
transform technique, which encodes the spatial information in phase with the
NMR signal, is optimal with respect to signal-to-noise ratio. The imaging of
porous solids like wood requires particular considerations. The solid matrix
material has an important effect on the relaxation, which can limit the resolu-
tion of standard imaging methods. The different water phases in wood must
be observed separately if possible. If not, the proportion of measured signals
corresponding to various phases must be distinguished.
For all imaging methods, there is a strong dependence of imaging time on
spatial resolution. The imaging time depends on spatial resolution as (resolu-
tion)4. However, the spatial resolution tends to decrease as the size of the object
increases. Typically, resolution is currently of the order of 1-2mm. In the
future, improvements of the technology will be directly related to the improve-
ment of acquisition time. The usual approach to obtain 3-D images is to use a
2-D image of a selected thin slice of a 3-D object, and then to move to the next
slice and so on until the whole object is reconstructed A thin slice is defined
by using frequency-selective radio frequency pulses, or a time-dependent field
a)
b)
Fig.6.19a,b. Outline of the NMR system. a Block diagram. b Position of the log in the center of
the magnetic field. (Wang et al. 1990, with permission)
gradients. The back-projection reconstruction method is the most popular

method for NMR imaging using scanners.
The major components of an NMR scanner are the magnet, three gradient
coils (for X, Y and Z directions), an RF transmission and receiving coil and
computers for control of pulse sequencing, data analysis and 3-D reconstruc-
tion (Fig. 6.19). The magnet provides a uniform static field. Both electromag-
nets and superconducting magnets are available for imaging. To avoid
problems with attenuation or phase shifts of the radiofrequency fields within
the sample, it is usual to work at relatively low fields of a 0.1 tesla, which cor-
responds to a lH frequency, which is about 4MHz.
Three sets of gradient coils modulate the magnetic field to spatially encode
the object. Two-dimensional images can be obtained by choosing one of the
three orientations of the three gradient coils as the slice selection. The other
two coils can then be used for readout and phase encoding gradients to map
out information on the plane. The gradient provided by these coils has a direct
effect on the spatial resolution of the image. The stronger the gradient, the
better the spatial resolution. The RF coil delivers a radiofrequency power to
flip the magnetization. The same coil picks up the NMR signals as a receiver.
The signal detected by the receiver coil is amplified, filtered, digitized and fed
into an array processor or a computer for image reconstruction. The computer
also controls the execution of the RF pulse sequence, the field gradients, data
acquisition, image reconstruction and display function.
An NMR image of the water content in a log (Fig. 6.20) can be obtained with
the following views: coronal from top to bottom, sagittal from left to right and
cross-sectional. The pixel is usually 128 x 128 or 256 x 256. The size of the
Fig. 6.20. Three NMR imaging views of a log. a Coronal view, from top to bottom. b Sagittal view,
from left to right. c Transverse view. (Chang et al. 1989; courtesy of Forest Products J)
sample imaged can typically be about 50 x 50 cm 2 and can be imaged with a

spatial resolution of order 1 mm. The imaging time depends on the pulse 90°,
180° sequence repetition or "recovery" time, the number of phase encoding
steps and the number of averages in the sequence. Initially, NMR systems used
repetition times of 500-800 ms. Faster repetition times are required for indus-
trial applications, but there are fundamental limits in repetition times set by
physical parameters.
In modern imaging facilitation, "clever tricks" are used to maximize the rate
with which the data can be obtained by sophisticated pulse sequences and
switched into field gradients in all three directions. Typically, Chang et aI.
(1989) using modern scanners in a saw mill, were able to acquire ten images
per second for log inspections.
Another interesting imaging technique called "current density imaging
(CDI)" involves the passing of a current to the object to be imaged. This current
produces a local magnetic field reflecting the current distribution through the
sample. In this way, it is possible to distinguish regions of high electrical con-
ductivity from those of poor conductivity. An interesting example of this tech-
nique is shown in Fig. 6.21, which provides additional quality information to
that obtained from conventional imaging alone. This imaging technique, which
generates maps of the spatial distribution of electric currents through differ-
ent wood tissues, superimposed on a conventional spin-echo technique, was
reported by Beravs et al. (1998) in a study of 2-year-old trees. One current pulse
is applied between a radiofrequency 90° pulse and a 180° pulse and a second
between a 180° pulse and a signal pulse echo. Both have the same magnitude
and duration, but are in opposite directions. During imaging, the electric
pulses produce phase shifts in the proton image, proportional to the magnetic
field produced by the current and to the duration of such pulses.
a) b) c)
Fig.6.21a-c. Electric current density imaging of a young oak tree. a Cross section in light
microscopy. b Nuclear magnetic resonance image. c CDI (current density image). A Cambial zone
+ differentiating xylem + complete phloem; B differentiated secondary xylem of the current
growth ring; C first annual ring; D pith. (Beravs et al. 1998; courtesy of Holzforschung)
)-r:,;, ho'~
x
, /
;0'01
RFcoll
- wood
_'''-.. glass tube
RF input/output
Fig. 6.22. The NMR probe containing the wood sample. (Quick et a1.l990; courtesy of Wood Fiber
Science)
Two-dimensional images of sapwood and heartwood, growth rings, knots,

reaction wood, resin pockets and other features have been reported by Chang
et al. (1989b), Wang et al. (1990), Niemz et al. (1998a, 2000b,c).
One-dimensional images can be obtained more rapidly and more easily
than 2-D images, using a modified solid-state spectrometer, as noted by Quick
et al. (1990). The NMR probe containing the sample is shown in Fig. 6.22. This
device was designed for measurement of the radial profile of the moisture
content during drying of western red cedar wood above the fiber saturation
point. One-dimensional images for pine specimens was reported by Lindgren
(1994). The drying of a 10 x 10 x 10 mm pine sample was imaged over 25-11 %

moisture contents.
6.3.2 Algorithms
Specific algorithms were developed for data interpretation in NMR spec-

troscopy and nuclear magnetic imaging. The most common method is to fit
the decay curves with the smaller number of discrete exponential terms, which
provide a satisfactory representation of the experimental data. This approach
fails when more than three exponential components are considered. A second
approach is to analyze relaxation times in terms of their discrete or continu-
ous distribution (Whittall and MacKay 1989). This procedure requires no a
priori assumption about the relaxation behavior and provides a more stable
fit to the experimental data. The application of linear inverse techniques to
characterize decay curves in terms of discrete or continuous distributions of
relaxation times is presented in Fig. 6.23. The amplitude of the nonnegative
least-squares spectra is plotted versus T2 • Three peaks were observed: one at
0.006 s due to the bound water in the cell wall, and two peaks at 0.030 and
0.200 s due to the free water present in cells of smaller and larger diameters
respectively, probably in earlywood and latewood. As a comment on this exper-
I:: I::
0.0 0.10 0.20 0.30 0.40 2 .....;::l
time (s) U
'l)
U
'l)
0..0..
en en
10000
0.006: '
8000 I
'l) I
I
]6000 I
I
~OOO I
§2000 b I
I
I
0
10- 3 10-2 10- 1
T2(s)
Fig.6.23a,b. Linear inverse technique to characterize decay curves in terms of discrete or con-
tinuous distributions of relaxation times. a Amplitude vs. time. b Amplitude vs. T2 with discrete
and smooth spectra. (Whittall and MacKay 1989, Academic Press, with permission)
imental approach, we note that an essentially continuous spectrum introduces

more adjustable parameters to fit the results than does a discrete spectrum.
A fine example of the power of algorithms for the NMR technique to image
the cross section of a whole log is shown in Fig. 6.24, in which one can clearly
see the individual rings, the core region and, in diametrically opposed posi-
tions, the first indication of the growth of side branches (top of the figure) and
the more fully developed side branches at the bottom of the figure.
6.3.3 Deduction of Relaxation Times from Measurements
For simple systems involving protons in water, the relaxation of Mz and Mxy are
given by simple exponentials with relaxation times T1 and T2' However, wood
is a naturally occurring composite material, where the water molecules in dif-
ferent macroscopic regions have very different relaxation times, where the
values of relaxation times from different regions are significantly different.
In some circumstances it is possible to extract T 1 and T2 for the protons in
different regions of the sample by fitting the relaxation to two exponential
terms. Some authors (Menon et al. 1987) have even attempted to describe the
relaxation process with three separate relaxation times. However, to obtain reli-
able information in such a way with any confidence would require an extremely
high signal to noise ratio and very long relaxation times to enable the various
regimes to be differentiated. For example, with an observation value for T2 of
6 and 200 ms for water in the wood, a wide range of permeability measure-
ments is possible for small and large anatomic elements (fibers, vessels, etc.)
which are in fact capillary tubes.
Another important feature of NMR imaging is its ability to identify wood
structure and defects. An example of this is given by the pulsed NMR mea-
surements of black oak (Quercus velutina) samples used by Coates et al. (1998).
An example of the identification of major structural areas is shown in Fig. 6.25.
The spatial variations observed are probably due to variations of the moisture
content in the sample and associated variations of density in growth rings and
rays.
In all defective regions, the gray level of some pixels widely deviate above
and below the gray levels of the clear wood region. In clear wood areas, the
variations in the gray levels are also important due to growth rings and rays
also because of random noise. Constant threshold limits are not able to give
the degree of variability of gray levels found in a particular log. A more effec-
tive approach is to introduce threshold limits that follow the contour of the
average moisture content. The largest gray level deviations are associated with
the defect area. Using values obtained for clear wood zones and subtracting
the largest deviations associated with defect areas, it is possible to obtain a
clear map of the defects. Computer algorithms have been developed by Vincent
(1993) and Peters (1995) to enhance the imaging of defect regions (Tsai 1985).
Transversal section
Longitudinal· radial section
Fig. 6.24. Imaging of transverse and longitudinal radial section of a beech. (Photo Escanye, in
Bucur 1990, with permission)
100
80
/
pith
\ knot
scar
40
20
~'I
0.0)0 0.100 0.1)0 0200 0.250
X lIXi$ ( 10e3)
d)
Fig.6.2Sa-d. Comparison between the photographic image and the automatically segmented
image of oak, using the algorithm proposed by Coates et al. (1998). a Photographic image of the
cross section. b Magnetic resonance image of the section. c Deviation of 40 gray levels or more
between original pixel and median axis. d Central profile of c where pith, knot and scar are visible.
(Courtesy of Forest Products J)
6.4 Applications
In this section, the physical and chemical characterization of wood using NMR
imaging techniques are discussed in relation to the inspection of trees, logs
and lumber and the possible industrial control of drying, impregnation of
wood products with preservatives and adhesion. For all such applications, it is
important to be able to quantify the water content, and the corresponding
Applications 245
moisture content in wood, and, on the other hand, the structural features at
macroscopic and microscopic level. Different aspects of wood processing
include the control of the moisture content in lumber during drying, the
control of the moisture content in wood-based composites, the control of
impregnation processes with preservatives and the kinetics of curing during
adhesion.
NMR imaging can also be used to image bulk physical properties of wood,
such as electrical conductivity. For many applications, it would also be useful
to be able to image the elastic properties of wood in situ. A very interesting
field of potential application is the simultaneous measurement of elastic con-
stants of bulk objects. Until now, this approach has not been used for wood
materials. Nevertheless, we present a theoretical background for this applica-
tion, which seems to be very promising for the mechanical quality estimation
of green lumber, trees and logs.
6.4.1 Inspection of Living Trees, Logs and Lumber
Examples of NMR imaging for the in situ inspection of living plants have been
given by Halloin et al. (1994) and Nakanishi and Matsubayashi (1997). Exam-
ples illustrating distribution and movement of water in roots and meristems
during development, in leaves or needles, and seeds have been reported.
Mac Fall et al. (1990) investigated the water depletion region in loblolly pine
roots. During the first stages of growth, the roots were easily visible and the
formation of a distinct water-depletion region around the roots was evident.
Later, this region extended around the lateral roots and clusters of mycorrhizal
short roots. The development of roots of Douglas fir seedlings, grown in con-
tainers at normal temperature, has also been investigated (Brown et al.I991).
The lengths of the roots determined from NMR images was in agreement with
the measured lengths of excavated roots. Southon et al. (1992) studied the influ-
ence of freezing on the roots of Scots pine and Norway spruce seedlings over
a large temperature range from +lto -20°e. The amount of root material
detectable by NMR imaging decreased with decreasing temperature, presum-
ably because of a significant increase in T1 and T2 in the material as the polar
mobility of water molecules, particularly in root material, decreases with the
temperature.
Several aspects of leaf and needle anatomy and morphology have been
discussed by Masuch et al. (1991), Veres et al. (1993) and Millard and Chudek
(1993). In vivo measurements of growth were made as a function of the water
content of the soil. The NMR signals of needles from declining forests showed
a larger NMR signal and a different distribution of free water from that of
healthy needles. NMR imaging has been also used for the discrimination of the
germinability of seeds as noted by Vozzo et al (1996).
6.4.1.1 Water Content in Living Trees
In this section, we will discuss techniques for water content imaging in living
trees, in fresh logs and in green timber. Previously, it was noted that the nature
of NMR signals is strongly dependent on the amount of water present in dif-
ferent wood tissues. The presence of magnetic ions (Cu, Fe, Mn) in the sap and
in the cell walls causes a strong reduction in the nuclear relaxation times T1
and T2' For example at 1% of copper sulfate solution in water reduces T1 and
T2 from about 3 s to 20 ms. NMR imaging and spectroscopy can also provide
information about the anatomic structure of healthy and infected wood of
living trees which will be discussed in this section.
6.4.1.1.1 Water Content in Healthy Wood Tissues
It is generally agreed that increment cores bored into the trunk of living trees
are representative of the quality of the tree. Byrne et al. (1986) pioneered the
measurement of the water content of trees by using an NMR spectroscopic
technique on 5-mm-diameter increment cores bored from healthy young Pinus
radiata and Eucaliptus grandis trees, using NMR at 60 and 90 MHz at two tem-
peratures (20 and 30°C), to investigate their frequency and temperature depen-
dence, as a noninvasive technique for measurement of water flow in young
trees. NMR measurements of T1 and T2 were made using two spectrometers.
T1 was measured by the 180°,90° sequence inversion recovery pulse sequence.
T2 was measured from the decay of spin-echo signals as a function of pulse
spacing in both species. The two measured relaxation times T1 and T2 had two
constituents, one long and one short, and were determined from logarithmic
plots of the NMR decay envelope at ambient pressure and temperature. The
moisture content of the two specimens ranged from 39 to 43%. To identify any
changes between the freshly taken samples and those used for laboratory
experiments, specimens were tested in their fresh state and after 3-day storage
in a sealed container.
The results are given in Table 6.4. The scarcity of experimental data permits
only quantitative conclusions:
- for both species, T1 long decreases with increasing frequency.

- for pine, T1 short also decreases with increasing frequency.
- for both species, T210ng and T2 short at 90 MHz are greater than the values
at 60 MHz.
- for fresh eucalyptus specimens, all measured constants are higher than after
storage for 3 days.
The data reported in this paper suffer from some inconsistencies (reference
to fresh and old eucalyptus samples). The presence of the two-component
relaxation times can be attributed to the different mobility of water molecules
Applications 247
Table 6.4. Some values of the relaxation times T, and T2 (ms) measured for Pinus radiata and
Eucalyptus grandis at 40% moisture content (Byrne et al. 1986,2002, with permission of Elsevier
Science from Agricultural and Forest Meteorology, vol. 38: 307-317, Bynne GF, Fenn MD, Burgar
MI 1986, reprinted from Nuclear magnetic resonance studies of water in tree section)
Frequency Time Pinus Eucalyptus Eucalyptus Eucalyptus

radiata old 3 days fresh
60MHz T,long 140 205 310

T, short 22
T2 long 10 7 7
T2 short 0.8 0.9 0.7
90MHz T,long 115 120 230 330
T, short IS 22 26 80
T2 long 22 21 33 46
T2 short 1.6 3.6 3.0 1.0
in different regions. The differences in T1 measurements at different frequen-

cies can be due to the mobility of the water molecules as well as to different
experimental conditions. More experimental data are needed to clarify this
point. The longer values of T2 on storage could be related to the mobility or
diffusion length of water molecules. Short T2values indicate a highly saturated
wood structure and the presence of small capillary channels (T 2is proportional
to the ratio a2/diffusion coefficient, where a is the size of a capillary). Short T2
values will also be associated with bound water present in the cell walls. T1 and
T2 long values can be associated with bulk water in the xylem vessels, fibers
and tracheids, while T1 short and T2short could arise from water trapped in
the wall material. The water in fresh trees, more exactly in the sap, has a T2 of
about 46 ms and is in a different physical state from free water, for which T2 is
only 3s. Byrne et al. (1986) demonstrated that NMR can be used for measure-
ment of the water flow in trees, and that only the cost of the equipment is a
serious impediment for field applications.
Another interesting imaging technique involves the passing of current to
the object to be imaged. This current produces a local magnetic field, reflect-
ing the current distribution through the sample. In this way, it is possible to
distinguish regions of high electrical conductivity from those of poor con-
ductivity. An interesting example of this technique was shown previously in
Fig. 6.21, which provides additional high quality image information to that
obtained by conventional imaging alone. The aim of Beravs et al. (1998) was
to create an image in very young oak trees (2 years old) of the cambial zone,
the xylem, the phloem, the first annual ring and the pith. The cambial zone is
located between the xylem and the secondary phloem and both originate from
the cambium. The imaging conditions were as follows: frequency 100MHz,
repetition time 600 ms, echo time 25 ms, field view 2 cm, slice thickness 4 mm,
Table 6.5. Average values of electric current density, of electric conductivity as a function of
wood density and wood moisture content in a young oak tree .(Beravs et al. 1998, courtesy of
Holzforschung)
Zone Tissues Electric Electric Density Moisture

current density conductivity (kg/m3) content ('Yo)
[A/m2] (s/m)
A Cambium, xylem, phloem 1582 0.070 610 119

B Second annual ring 1314 0.058 550 63
C First annual ring 842 0.037 430 52
D Pith 1205 0.053 610 74
pixel grid 256 x 256. The imaging time was 15 min to average six scans. The
voltage applied to the sample was 180V, with a current pulse length of lOms.
The 3-D images were obtained from two perpendicular orientations of the
sample. As is evident from the image, current density imaging offers a higher
contrast between the different tissues than conventional NMI imaging.
The current density image shows a conductivity contrast in four different
regions of the oak sample. The corresponding values of electric current den-
sities, electric conductivity, and density and moisture content of the wood are
shown in Table 6.5. The increase in the moisture content from the first annual
ring to the cambial zone results in an increase in all measured parameters.
All reported experimental data in this section were obtained at room tem-
perature. The influence of low temperatures on relaxation times was studied
qualitatively by Johansson (1985) with frozen logs in Sweden. No NMR signals
were obtained in this case, probably because of the solid state of frozen water.
The mobility of water molecules decreases with decreasing temperature so that
there are fewer higher frequency components of the local fluctuating magnetic
fields to restore Mz to its equilibrium value (hence TI increases) with similar
effects on T2'
6.4.1.1.2 Water Content in Infected Wood Tissues
The second part of this section is devoted to the imaging of fungal coloniza-
tion of young living trees. The imaging of fungal attacks with different patho-
logical agents of living sapwood of sycamore has been reported by Pearce et
al. (1994, 1997). The infection agents for sooty bark disease were Ustulina
deusta, Chondrostereum purpurem, Cryptostroma corticale, and Ganoderma
adspersum. The development of fungal lesions in the sapwood of living trees
is restricted by an active defense mechanism. The pathological anatomy of
sooty bark disease lesions can easily be seen by imaging shown in Fig. 6.26.
The inoculated zone is surrounded by a reaction zone located between the
Fig.6.26a,b. Different sections through a stem of young Acer pseudoplatanus tree wound inocu-
lated with Ustulina deusta. Longitudinal-radial section inoculated with Ustulina deusta where
one can see D decayed zone; H healthy zone; R reaction zone, which separates the healthy and
decayed zones; C callus beginning to grow over the wound faces. Scale bar 10 mm. a NMI obtained
with a spin-echo image sequence for a 2-mm slice thickness, with a resolution higher than
0.47 mm. b Comparison of pixel intensities in specific areas of Mo maps and maps of measured
TI and T, have shown that both techniques are similar for the illustration of different lesions (L)
in different tree regions. Reaction zone had a high contrast. Scale bar 10mm. (Pearce et al. 1997,
Blackwell Science LTD, with permission)
decayed and the healthy wood. It was demonstrated that changes in moisture
content are associated with the formation of decay lesions in living trees. Spin-
lattice Tj maps showed little contrast compared with proton density maps.
Measured relaxation times were within the range of 700 to 1000 ms.
The spin-spin relaxation time T2 for healthy wood varied between 35 and
127ms. In reaction wood, T2 was reduced by as much as 40%. Neither Tj nor
T2 was altered in the drier wood adjacent to naturally occurring Cryptostroma
corticale, but both were reduced in the infected wood. The moisture content
was higher in the reaction zone than in the healthy area. Calculated proton
density images showed the delimitation of the reaction zones. The contrast of
the images of lesions was attributed to proton density variations and to the
decrease in spin-spin relaxation time T2 in the vicinity of the lesions. In the
reaction zone, the values of spin-lattice relaxation time T were less sensitive
j
to the pathological modification of the tissue than T2'
6.4.2 Growth Rate and Other Structural Features in Logs and Lumber
6.4.2.1 Structural Features at the Macroscopic Level
The detection of annual rings, knots, reaction wood, wet wood, decay and other
structural features with the NMR technique is a basic approach and is a chal-
lenging task for image reconstruction and processing. Pattern recognition
techniques to identify, classify and quantify structural features are needed for
the on-line internal inspection of logs before sawing at normal production
speeds in saw mills and other wood processing. NMR image processing tech-
niques are similar to those used for X-ray computed tomography, but the infor-
mation content of the NMR images is higher than that obtained from X-ray
computed tomography and ultrasonic imaging. Because the NMR signal inten-
sity is directly proportional to the moisture content, the higher the moisture,
the better the image. However, even for samples with a low moisture content,
a reasonable image can still be obtained by averaging signals over a long time.
Hall et al. (1986a,b), using a relatively low field of 0.14 tesla, obtained clear
images of the macroscopic structure of fresh-cut bolts of aspen and spruce of
25 cm diameter. Annual ring patterns, buried knots and decayed regions were
clearly observed. Similar images have been reported by Wang and Chang
(1986) and Swanson and Hailey (1987).
In this section, we present images from a beech bolt of irregular shape and
elliptical cross section (75cm long and maximum diameter 15cm), cut from a
tree at a height of 20 m from the ground, on which bolt branches were formed.
In this sample, the wood structure is very complex, with normal wood, reac-
tion wood, knots and other features. Beech was chosen for imaging because of
its commercial importance in Europe. The internal structure of the bolt is given
Applications 251
a) b)
c) d)
Fig.6.27a-d. Views of the internal structure of a beech bolt. Images in the transverse plane at
different heights (74, 33,21, 15cm) noted a, b, c, and d, respectively. NB: The images were taken
on a medical scanner for 0.5 tesla, with a spatial resolution of 1.2 x 1.2 mm. The echo time
(30 ms) permitted imaging only of the distribution of mobile, free water. (Photos Escanye in Bucur
in Fig. 6.27 with four transverse sections at different heights (15,21,33, 74cm)
and sagittal views. Well-imaged features are the annual rings, the pith, the
red heart zone, the sound wood, the tension wood, and even the bark. The
image clearly shows branch formation, which corresponds to the physical state
of the bolt. The bark is seen as a very clear zone. Loss of image intensity is due
to the low water content. The oval-shaped bright areas correspond to branch
formation with decreased water content and reveal the existence of buried
knots.
The abnormally bright area around the pith corresponds to red heart. The
development of red heart is different at different heights. The dark protuber-
ances of the red heart contain less water than the surrounding area, probably
because of to the presence of tyloses.
The diameter of the pith is 3 mm, and the first annual ring has a diameter
of 4 mm. In the healthy zone, the earlywood of the annual rings shows a lighter
color than the latewood of the same ring, which implies either that it contains
more water or that this water has a longer relaxation time. The latewood zone
in all rings is always dark, indicating low water content. The natural distribu-
tion of water during the life of the tree can be seen. The influence of climatic
changes, or of physical and biological damage can be determined from such
NMR images. By looking at the image carefully, one ring of the tension wood
with a darker latewood zone can be seen in transverse image at 25 mm from
the periphery of the bolt. In the sagittal plane image, the same dark vertical
projection of the ring can be observed. This ring is therefore likely to be tension
wood with less moisture content than the surrounding rings and a shorter
relaxation time. White irregular small bright stain checks between the annual
rings signify a high level of water content
6.4.2.2 Structural Features at the Microscopic Level
It is not the aim of this section to review the NMR techniques that reveal the
porous nature of solids in general, but rather to select the studies on those
materials of most relevance for the understanding of wood. Cohen and
Mendelson (1982) studied the influence of the size and geometry of pores in
rocks on NMR signals. Assuming the validity of the Bloch equations with mag-
netic relaxation on the surface of the pore voids in a uniform field, T 1 and T2
could be defined as a single relaxation rate that was related to the surface area
and the volume of the pore spaces. The NMR measurements of T 1 and T2 were
made to determine the spin-lattice relaxation and the diffusional attenuation,
using a spin -echo signal. A strong magnetic field was used to polarize the water
molecules in the pores of the sedimentary rocks. When the magnetic field is
reduced, the proton magnetization relaxes to a new equilibrium value. The
decay in the amplitude, at a rate determined by T[) is sensitive to the geome-
try of the pore space of the rock. Similar approaches can be extended to the
specific case of porous woods, and a new method for porosity measurement
could be developed in the future.
In this section, we consider the sensitivity of the nuclear magnetic imaging
technique to the structural features of wood on a microscopic scale, as a func-
tion of cell size and the dimensions of earlywood and latewood in the annual
ring.
Wycoff et al. (2000) estimated the cell size in four softwood species (eastern
red cedar, redwood, eastern white pine and sugar pine) using NMR measure-
ments. The diameter of the cells in these species is between 25 and 65/-lm and
Applications 253
the lengths between 2 and 5 mm. Measurements of the translational diffusion

coefficients in a field gradient spin reveal barriers to diffusion along the length
of the cells and indirectly provide a method to measure boundary dimensions
and geometry. Calculation of the apparent diffusion coefficients was per-
formed with a model proposed by Callaghan (1995), in which the signals from
both free and bound water were considered. Fields along the diameter in a tan-
gential anisotropic direction were chosen to measure the cell dimensions,
because in this direction the cellular wall has the smallest effect on the mobil-
ity of the absorbed water. Comparative measurements were performed also in
the longitudinal direction to establish the expected anisotropy.
For moisture contents above the fiber saturation point, the NMR signal is
primarily due to the "free water" associated with the cell lumina. Only a small
portion of the signal is contributed by water molecules "bound" to the cell walls
by strong hydrogen bonding. The diffusion of the bound water is also very
small and the corresponding diffusion time is very long relative to the total
diffusion time. For convenience, the cells are assumed to have a rectangular
cross section so that they can be described by a parallel plane geometry. For a
short field gradient, the attenuation of a spin in a 90-180° pulse sequence can
be considered in terms of relaxation of spin the diffusion by the cell walls
For example in Fig. 6.28, the variation of the diffusion coefficient of water
is plotted as a function of time diffusion in eastern white pine. The values mea-
longitudinal
2.0
1
'"
0
~ 1.5
.~
IE
§
"
.~
@
'il
0.5
0.0 '---_I....--"I....---l_--1_--L_....:.._-I.._....:.._ _
o 200 400 600 800
!!. (ms) = diffusion time

Fig. 6.28. Relationships between the apparent diffusion coefficient and the diffusion time of
water in longitudinal and tangential directions of a wood sample of eastern white pine. (Wycoff
et al. 2000; courtesy of Wood Fiber Science)
Table 6.6. Diameters of cells in the tangential direction and cor-

responding standard deviation, measured with the NMR tech-
nique and with optical microscopy. (Wycoff et al 2000, courtesy of
Wood Fiber Science)
Species Cell size with Cell size with NMR

microscopy (~m) technique (~m)
Sugar pine 42 ± 3 45 ± 3
Eastern red cedar 17 ± 1 17 ± 2
Eastern white pine 31 ± 1 35 ± 3
Redwood 40 ± 1 43 ± 3
sured for the field gradient in the longitudinal direction are very close to the
values for bulk water (2.2 x 1O-9 m 2/s), demonstrating that there is no signifi-
cant barrier to diffusion in this direction. The deduced diffusion coefficients
were between 2.1 x 10-9 and 2.4 x 1O-9 m 2/s. The diffusion length along the cells
corresponding to a diffusion time of 903 ms corresponded to a diffusion length
of 150 ~m. For pulsed gradient fields in the tangential direction the barrier
walls limit the diffusion length.
Good agreement between the dimensions of cells measured by optical
microscopy and NMR is evident from the data shown in Table 6.6. The values
measured with the NMR diffusion technique, using water molecules as probes,
were only slightly higher than those obtained by microscopy. Wycoff et al.
(2000) demonstrated that the NMR provides accurate measurements for cell
sizes at moisture contents above the fiber saturation point. However, for dry
wood, it would be more difficult to separate the signals for free and bound
water.
One-dimensional NMR microscopy can be used to produce images at the
millimeter and submillimeter level corresponding to regions with different T2
values. Menon et al. (1987) demonstrated that NMR signals from water in
western red cedar (Thuja plicata) can be separated into three components with
different T2 values such as:
- values of a few milliseconds corresponding to water bound within the cell
walls,
- values of about 50 ms corresponding to water located in latewood tracheids
and ray lumina,
- values of over 100 ms corresponding to water in the lumina of the earlywood
tracheids.
It was also found that the relaxation time corresponded to lumen water T2
scales with lumen diameter (D) as D2.
The image in Fig. 6.29 was obtained with a Carr-Purcell-Meiboom-Gill
(CPMG) pulse sequence in a constant field gradient of 19.4 Gtesla/cm. To
isolate an image of each T2 component of water in wood, it is assumed that
Applications 255
Fig. 6.29. Images of free water in the lumina of earlywood tracheids (EW), of the lumina of
latewood tracheids (LW), and from bound water in the cell wall (CW) superimposed by the
scanning electron micrograph of western red cedar. (Menon et al. 1989, Academic Press, with
permission)
S(t) = 0.l1e-t/s.6 + 0.12e- I/46 .8 + 0.77e-t/184.1 (6.9)
where t is the time (ms) from the start of the CPMG sequence and S(I) is the
amplitude of the CPMG echo sequence.
The three terms of the equation correspond to the three T2 components of
water content. When the water content in latewood, earlywood and cell wall
components were superimposed on the corresponding scanning electron
micrograph, near perfect agreement was observed. The authors also demon-
strated that the variation of bound water obtained from one-dimensional
images across the annual rings on a green specimen is perfectly matched by
that of the oven-dry wood density in the radial direction when measured with
an X-ray micro densitometric technique on the same specimen, as can be seen
from Fig. 6.30. The major contribution of Menon et al. (l989) has been to
demonstrate that NMR imaging can be used to separate images on the basis of
spin-spin relaxation time T2, so that images on a submillimeter scale can be
achieved. Such studies will enable the role of the bound water to be better cor-
related with other physical properties, such as dimensional stability, thermal
and electrical conductivity and mechanical properties.
1000
400
800
1600
.€
'"
~ 400
"8
~ 200
o 2 3 4 (rom)
Fig. 6.30. The profile of the bound water in the cell wall compared with the X-ray densitometric
profile of annual rings in western red cedar. (Menon et al.1989,Academic Press, with permission)
, , ,,
I
td 90°:, ts :90°
,
tm i
90 0 I tc I,, ta
,
I
..
I I
I ~ I
+x: +y : ±x +x~ +y :
I , '----
I
I ,
Fig.6.31. Representation of the pulse sequence MOPS to monitor [H and BC spin diffusion.
t, Contact time; t. data acquisition time; td recovery delay time; t, spin locked time of 8 ms; tm
remaining magnetization time; -x and +x pulses phased along x-axis. (Newman 1992; courtesy
of Holzforschung)
6.4.2.3 Spatial Distribution of Chemical Components in Cell Walls
Because the Larmor frequencies are specific to different nuclei, NMR methods
can be used to investigate the variation of the chemical composition of a wood
sample. The spatial variation of different chemical components in the wood
cell wall was first studied by Newman (1992) using NMR. He found that natural
differences in the proton spin relaxation time constants governed different T2
values corresponding to ordered and less ordered cellulose and amorphous
lignin. The cellulose signals with chemical shift 0 = 89 ppm were assigned to
13C in cellulose and 0 = 56 ppm 13C at nuclei in lignin. T1 and T2 for 13C can be
obtained from the same experiment at 50.3 MHz as described in Fig. 6.31 for
Applications 257
0.9
'-'
en
.9
.=:
v
0.8
u
6\
00
~
a.
::I:_
f--; 0.7
.S
c '1
1f ""1------1 Eucalyptus
\, , ,....l
<;:;
'-D
~
, ,
0.6 ,,~~
a.
:r::_
f--;
0.5
0 20 40 60 %
water content (%w/w)
Fig. 6.32. Proton rotating frame spin relaxation time signals vs. the water content for E.
delegatensis and Pinus radiata and for cellulose and lignin. (Newman 1992; courtesy of
Holzforschung)
Eucalyptus delegatensis and Pinus radiata specimens. The spin locked time was
ts = 8 ms. The value of tm, the time needed for the remaining magnetization
to return to an orientation parallel to the static field, was variable. Proton
magnetization was monitored as a function of tm, with a standard cross-
polarization sequence. During the time tm, the phase of the second 90° pulse
in the MOPS sequence was alternated to suppress interference from TI relax-
ation. An important experimental detail to note is that D2 0 was used in mois-
turizing the specimen (with 35% by weight), which results in a longer value of
TI. Figure 6.32 shows the variation of the ratio of lignin to cellulose relaxation
times TIp H in Eucalyptus delegatensis and Pinus radiata versus water content.
Up to 20% of water content, the curves are superimposed. However, above 30%,
corresponding to the fiber saturation point, the values for eucalyptus are
slightly higher than those for pine. The differences between the spin-lattice
relaxation time, TIH , in cellulose and lignin is again evidenced in Fig. 6.33. The
increase in TIp H suggests that the spin-lattice relaxation is dominated by mech-
anisms involving water molecules or hydroxyl groups. The influence of mois-
turizing on Pinus radiata is more important than on Eucalyptus delegatensis
and can be explained by the difference in strength of bonding of water mole-
cules to the different chemical components of each wood species. The average
1.0 legend
E = Eucaliptus
P = Pinus
0.4
0.2
'U;'
0.1
;:2::
f-< 0.08
0.06
0.04
0.02
0 20 40 60 %
water content (%w/w)
Fig. 6.33. Proton spin-lattice relaxation time vs. water content for cellulose and lignin in
eucalyptus and pine. (Newman 1992; courtesy of Holzforschung)
Table 6.7. Proton rotating-frame spin relaxation times TlpH (ms) for Pinus radiata moisturized
with H,O and with D,O to 35% of total weight. (Newman 1992, courtesy of Holzforschung)
Specification Values H,O D,O Differences (%)
Cellulose (0 = 56ppm) Experimental 3.9 6.2 37

Theoretical 2.8 5.0 44
Lignin (0 = 89ppm) Experimental 7.3 9.8 25
Theoretical 12.4 14.2 13
difference between cellulose and lignin is 22%. To enhance this difference, the
increase in TJpH was demonstrated for Pinus radiata at 35% moisture content,
with the H 20 replaced by D20, as can be seen in Table 6.7.
The experimental values were influenced by diffusion between domains.
Theoretical values were deduced by computing simulation, using the model
cited before with which spatial segregation of chemical components was
observed, as can be seen in Fig. 6.34, where the signal height differences
between cellulose and lignin were plotted versus the time tm. The perturba-
tion of initial magnetization disappeared after about 20 ms of mixing and no
difference between cellulose and lignin was observed. The difference between
the theoretical and the experimental values for cellulose are in the range of
experimental uncertainties, but for lignin these differences are affected by a
Applications 259
10-4 10-3 10-2 10- 1

tm (s)
Fig.6.34. Cellulose and lignin signal height ratios from MOPS experiments in Pinus radiata
moistened with D20 at 35% vs. the magnetization time (trn). (Newman 1992; courtesy of
Holzforschung)
factor of 1.7. It can be concluded that the rotating-frame spin relaxation

mechanisms involving water are important in lignin but not in cellulose.
Mixing of cellulose and lignin as wood constituents is illustrate.d by the
similarities in T 1/ .
6.4.3 NMR Imaging in Wood Processing
This section describes the potential use of nuclear magnetic imaging in wood
processing, and particularly its applications to the control of the moisture
content during lumber drying, quality control of wood-based composites and
their moisture content at different phases of processing, control of adhesion
between adhesives and wood, and the control of impregnation processing of
wood with preservatives. The last part of this section is devoted to a study of
archeological wood.
6.4.3.1 Moisture Content in Lumber During Drying
Over the last 30 years, extensive literature has been published relating NMR
spectroscopy to the measurement of the moisture content in wood. For the
forestry industries and for wood science, the most important application of
NMR is undoubtedly the measurement of water flow and diffusion in wood,
both in situ and in vivo specimens. As noted by Nanassy (1973), the physicists
Shaw, Elsken and Rubin in the 1950s were the first to draw attention to the pos-
sibility of using NMR to measure accurately the moisture content of wood.
A serious limitation of this technique, at the time, was the large expense of
the apparatus required. Using a laboratory spectrometer, Nanassy (1973) pro-
posed a graphic method for measurement of the relative intensity distribution
of the narrow component spectral absorption. This approach permitted the
moisture content in yellow birch and white spruce to be derived in the range
of 5 to 100%, on specimens of about 2cm3 volume. The accuracy was around
±3%.A year later, the same author (Nanassy 1974) used a broad-spectrum com-
ponent system to determine the fiber saturation point, called by the author
"critical moisture content" defined in relation to "the phases of the sorbed
water, and the ratio of wood protons participating in chemical exchange
between wood and water protons." The fiber saturation points at 25°C for white
birch and white spruce were both found to be 38% of the protons in specimens.
This value was higher than the generally accepted value of about 28% mois-
ture content for wood species from temperate zones. Nanassy (1974) explained
that the diffusion of water in wood over the cell wall surfaces is dependent on
the bonding energy of the water molecules to the cell walls, and that such water
is "strongly bound to the rigid lattice of wood." The additional free water, called
by the author "mobile water;' is only weakly bonded to the cell walls. Experi-
ments using improved electronics and corresponding NMR systems during the
last 10 years have validated Nanassy's statements, as shown in many articles
published in recent years.
As seen in Section 6.2, the diffusion of water in wood is anisotropic
(MacGregor et al. 1983). Measurements have been performed with two species,
Douglas fir and sugar maple, using a pulsed gradient NMR technique. These
measurements confirmed that diffusion along the fibers is higher than diffu-
sion in the transverse directions (Table 6.8). Water diffusion in maple is more
Table 6.8. Effective diffusion coefficients D' (l0-Scm2/s) in

Douglas fir and sugar maple, determined with the pulsed field gra-
dient NMR technique, as reported by MacGregor et al. (1983, with
permission)
Species D' parallel" D' perpendicularb Ratio
Douglas fir 1.7 ± 0.1 0.88 ± 0.05 1.9

Sugar maple 1.2 ± 0.1 0.36 ± 0.05 3.3
"The coefficient D' parallel was measured in longitudinal

anisotropic direction of wood direction corresponding to the axis
of the cylindrical specimen.
bThe coefficient D' perpendicular was measured in the perpen-
dicular direction (radial or tangential anisotropic direction) not
specified explicitly by the authors.
Applications 261
anisotropic than in Douglas fir, which is in agreement with the anatomical

structure of these species. It was also suggested that further experiments
should be made using three different cylindrical specimens. Each specimen
must have the long axis oriented in one anisotropic direction, L, R or T. In this
way, the influence of the specimen geometry on the diffusion phenomena can
be avoided.
During the 1980s, many ingenious methods of spin imaging were developed,
making possible enhanced one-, two- or three-dimensional NMR imaging. It
is undoubtedly of great interest to compare the information obtained from
NMR imaging methods with those obtained from high resolution NMR spec-
troscopy in order to evaluate the state of different wood tissues within the
sample under inspection. For clarity, in the following pages, the distribution of
water during drying will be studied in small, clear specimens and in boards of
industrial size.
6.4.3.1.1 Water During Drying of Small Clear Specimens
The distribution of water in a small, clear specimen can be characterized by

one-dimensional radial imaging across the growth rings. This may be achieved
with only a single field gradient, which involves only the slice perpendicular to
the field gradient as demonstrated by Quick et al. (1990), for western red cedar,
and by Araujo et al. (1992) for white spruce.
Lindgren (1994) presented one-dimensional images of pine in the radial
direction during drying from 27% moisture content to 11% (Fig. 6.35). One-
dimensional images were obtained from the Fourier transform of the FID.
Amplitude
20 xl 04. - - - - - - - - - - - - - - - - - - - - - ,
o 10
------ u=25%
-----u=20%
-----u=15%
------- u=11 %
-O.5xlO 4i - - - - - . - - - - - - , - - - - r - - - - - - , - - - - - - - i
~~
100 115 130 145 160 175
Fig. 6.35. One-dimensional imaging with proton gradient during drying of small, clear pine
specimens from 25 to 11 % moisture contents. (Lindgren 1994, with permission)
They represent the radial moisture profile with amplitudes proportional to the
spatial variation of the water content. For the small, clear specimens an accu-
rate analysis of water distribution above and below the fiber saturation point
can be made.
Since the aim of this section is to relate the 1-D distribution of water protons
and the structural features of wood, it was therefore natural to superimpose
the SEM optical image of the corresponding anatomic structure on the proton
profile (Fig. 6.36). To illustrate this statement, we refer to the results on white
spruce published by Araujo et al. (1992) and to the definition of moisture
content proton density in Eq. (6.6). The drying of white spruce sapwood
from 100 to 17% moisture content proton density was followed in four steps,
corresponding to 100, 86, 59 and 17%. The radial moisture density profile
for total water content, for free water and for bound water was obtained and
superimposed on the SEM image. One-dimensional images were obtained
from the Fourier transform of the T2 signals. They showed that the total area
under the profile of the 1-D image is proportional to the mass of water in the
sample.
One-dimensional images for the representation of the amount of total water
(Fig. 6.36a) can be analyzed for each moisture density level as:
- for 100%, the moisture density is higher in the earlywood zone of all rings,
because of the larger tracheid lumina in earlywood.
- for 86 and 59%, the rate of drying is more important in earlywood than in
latewood.
- for 17%, the distribution of water is quasi-uniform in earlywood and in
latewood. It was stated that this curve corresponds to the bound water.
From Fig. 6.36b, representing the 1-D images of the free water present in the
lumina, it can be seen that the profiles are very similar to those discussed pre-
viously. For the 17% water content sample, the central component indicates the
presence of a small amount of water inside the lumina.
In Fig. 6.36c, showing I-D images of bound water, we observe a superim-
position of the bound water profile on the growth ring profile. This allows us
to conclude that above the fiber saturation point the bound water distribution
is in agreement with the density of solid wood, which is higher in latewood
than in earlywood. Below the fiber saturation point, the distribution of the rel-
ative moisture density is uniform in the earlywood and latewood. With such
evidence, it was possible to suggest that the bound water profile was unrelated
to the wood density.
Another possible interpretation of data produced from T2 spectra is to relate
histograms of the lumen radius to that of the proton volume and to the cell
number (Fig. 6.37). Both histograms in white spruce sapwood and in com-
pression wood show similar distributions.
This allows us to conclude that the lumen radius is related to the proton
volume and that the peaks in the continuous spectra can be attributed to the
anatomical location of the water. Different tissues, for example juvenile wood,
0.6
0.5
""'
5 0.4
M
~
c-
' ih
0.3
c
<>
"'" 0.2
~
E
'8 0.1
E
o
- 0.11--- - , -- -- . , -- - - , - - - - - - ,
-4 -2 o 2 4
a) position ( mm )
0.6
0.5
M
E
~ 0.4
c- 0.3
.;;;
c
<>
"'~" 0.2
2
'"
'0 0.1
E
o
-0. 1 +-----,-- - - - - . - - - - - - . - - - - - ,
-4 -2 o 2 4
b) po ilion ( mm )
0.6
0.5
r'l
E
~ 0.4
C- 0.3
.;;;
c
<>
~ 0.2
2
.~ 0. 1
E
o
-0.1 + - - - - - r - - - - - - . - - ----.------,
-4 -2 o 2 4
c) position ( mm )
Fig.6.36a-c. One-dimensional radial moisture profile of protons superimposed on the anatomic

structure of white spruce at 100, 86, 59 and 17% moisture contents. The radial moisture density
profile corresponds to a total water distribution; b free water and c bound water distribution.
(Arauja et al. 1992; courtesy of Wood Science Technology)
0.20
..:!) ..:!)
~
u
cells ~
u
cells
'-H 0.15 '-H 0.15
...
0
(!)
...
0
(!)
,D ,D
S S
;:l
;:l
I:i 0.10 I:i 0.10
(!) (!)
.~ :>
ttl '.g
~
... 0.05
~
... 0.05
[ rrr , ~- .'
protons
protons
0.15 0.15
(!) (!)
S S
;:l
..2
"0 0
:> :>
(!)
0.10 (!)
0.10
:> :>
'"§ '.g
~... ~
...
0.05 0.05
o
rfl-1-ri
15
~ I
20
I
25 o 10 15 20 25
5 10 5
lumen radius (Jlm) lumen radius (Jlm)
a) white spruce sapwood b) compression wood in white spruce heartwood
Fig.6.37a,b. Histograms of numbers of cells and of protons as a function of lumen radius in a
white spruce sapwood samples; b white spruce heartwood compression wood. (Arauja et al. 1992;
courtesy of Wood Science Technology)
compression wood or heartwood, are characterized by different cell sizes

and can be distinguished by the shape of their T z spectra, as can be seen in
Fig. 6.38.
Flibotte et al. (1990) used small, clear specimens of 10 X 5 X 5 mm, measur-
ing free induction decay curves (FID) for a series of specimens including
undried sapwood, heartwood, juvenile wood and decayed wood in western red
cedar. The corresponding hydrogen contents were 6.28,6.14,6.13 and 5.27%,
respectively. In these specimens, the moisture content was between 30% in
juvenile wood and 235% in decayed wood. To monitor the drying of the
samples, a pulse sequence was used. The gradient of moisture content during
drying is shown in Fig. 6.39. In all signals three distinct components were
observed; one corresponding to a rapidly decaying signal, induced by the
immobile protons in wood and two other, slowly decaying signals from mobile
water molecules, corresponding to water in the lumina of earlywood and late-
Applications 265
0
W- 4 10- 3 10-2 10- 1
a) T2 (s)
18
15 -
,\
{\
/' \
-
12 - ",
-
Q) :i ", \
:', \
"0
.~ 9- ",
0..
§ :'/ -.,I
6- :' I :, \
3-
1\ .. / ",I
f-
/!'~
'/,\
/,~,\
' I '
"y', I ".\
0
10-4 10- 3 10- 2 10- 1 10°
b) T2 (s)
Fig,6.38a,b, Continuous spectra T2 in white spruce, a Juvenile wood (solid line) and heartwood
(dashed line); b in rehydrated samples of heartwood compression wood (solid line), heartwood
(long dashed line) and heartwood with incipient decay (short dashed line), (Arauja et al. 1992;
courtesy of Wood Science Technology)
wood tracheids and in ray lumina. The moisture content deduced from NMR
signals was obtained by "multiplying the ratio of water to wood signal inten-
sities by the ratio of the hydrogen content of wood to that of water:' Compared
to values for oven-dried wood, the NMR values are overestimated at higher
moisture contents and underestimated at lower moisture contents by several
160.------------------------. 1000,-------------.
~
140 I--'.~\--------------.- .. --
.. ---.s:-.lo,...w----j
1201-__~'------------~-~-=-~··m~e~dwiullim~
......... slow
---··medium
I early wood
'E fast -fast ~
100f--------/-,--,.,.-.-.+;-1
lumena water
~ 1001-----~------------------~
~ 801------·-~·~,,~--------------__j -_.-._..- ~
1
8 "', -....... late wood

~:~ ~/----~..".. "- lumena water
r '.',
20 ......-":> .
.. ::..~-.-.- ~-f- bound water
o , , 1 "----..----..----,.--J
250 220 190 160 130 100 70 40 240 180 120 60
a) oven-dry moisture content % of wood b) moisture content %
Fig.6.39a,b. Moisture content determined with NMR technique and the relaxation time T2 vs. the
gravimetric moisture content in cedar. a Juvenile wood; b decayed cedar wood. N.B. The mois-
ture content deduced from NMR signals was obtained by "multiplying the ratio of water to wood
signal intensities by the ratio of the hydrogen content of wood to that of water." (Flibotte et al.
1990; courtesy of Wood Fiber Science)
percent. Nevertheless, bearing in mind the experimental errors, this method

can be considered valid for the measurement of moisture content of wood
above and below the fiber saturation point.
6.4.3.1.2 Water During Drying of Boards
We will now consider nuclear magnetic imaging in boards of industrial size.

For 3-D imaging, only the presence of the fluid in the pore space is observed
during drying. Olson et al. (1990) noted that NMR imaging is more effective
than NMR spectroscopy in the determination of moisture gradients within a
large wood sample. Three-dimensional imaging was used by Olson et al. (1990)
to monitor the moisture distribution in oak boards during a 14-day period.
Boards of three thicknesses were selected (25, 38 and 50 mm) with a moisture
content ranging between 68% at the beginning and 44% at the end of the exper-
iment. A clinical NMR scanner was used for imaging the internal structure of
the boards, giving an image intensity proportional to the free water present.
The image area was 30 x 30 cm with a spatial resolution of 1.2 x 1.2 mm.
Figure 6.40 shows multislice spin-echo images for the transverse section of
the board. The images exhibit the variation of moisture content during the
drying time. The quantitative estimation of the brightness of the images can
be obtained with a routine optical density technique. The brighter core areas
represent higher moisture content in contrast to the darker shell area around
the surface. The rays and the annual ring are well resolved. The rays have a
Applications 267
a)
7 days I day
c) d)
15 days 4 days
Fig.6.40a-d. Transverse views of oak lumber (SO-mm thickness) during natural drying. a One-
day drying; b 4-day drying; c 7-day drying; d IS-day after cutting. (Wang et al. 1990; courtesy of
higher moisture content than the surrounding annual ring area. It is evident
that the drying process takes place from the outside to the side of the board.
The kinetics of drying can be observed for all specimens and information
about the moisture gradient and distribution can be deduced. A 25-mm-thick
board dried at a faster rate than the 50-mm board because it simply takes
longer for water to diffuse out from the center of the thicker boards. The dark
area is dependent on orientation of the growth ring. The brighter area in the
middle region of the cores indicates that diffusion is faster from the central
core areas than from the outer surfaces of the board.
We have shown that NMR imaging is a valuable nondestructive and non-
contact method for qualitative and quantitative studies of the kinetics of wood
drying. Its only drawback is the cost of such equipment, which is similar to the
medical device used in hospitals. However, important information can also be
obtained by 1-D scanning, with much less expensive spectrometers, using small
core samples taken in vivo at various processing stages.
6.4.3.2 Quality Control of Wood-Based Composites
The application of solid-state NMR techniques to the quality control of wood-

based composites is related to two main aspects: the nondestructive measure-
ment of the moisture content and the analysis of the adhesive bond lines. For
measurement of the moisture content of wood-based composites, the litera-
ture is very scarce. Wolter and Netzlmann (1996) reported the utilization of
single-side access NMR facilities for measurement of moisture content and
density distributions in wood-composite boards. The goal of this preliminary
research was to develop an in situ device
An understanding of the mechanisms of adhesion in wood-based compos-
ites is important for new engineering applications. Physicochemical phenom-
ena that govern adhesion are related to molecular interactions between the
adherent and the adhesive. In wood particles or flakes, as in all solid polymers,
molecular ordering and motion can be investigated by measuring relaxation
times. Changes in molecular structure of wood composites can be observed if
the number of molecular associations between the adhesive molecules and the
wood molecules is large enough. Gil and Neto (1999) and Marcinko et al. (1998,
1999) have reported information on the molecular dynamics in wood and
adhesives like urea-formaldehyde (UF) and polymeric diphenylmethane diiso-
cyanate (PMDI). BC spectra were collected using a standard proton longitudi-
nal relaxation time IHT 1p pulse sequence at 50.77 MHz. Southern pine particles
were mixed with resin and cured in a forced-air oven at 140°C for Ih. Table
6.9 gives the IHT 1p data for lignin, hemicelluloses and cellulose. Data obtained
for southern pine solid wood and UF-coated southern pine particles are
similar. Relaxation times corresponding to PMDI-coated southern pine parti-
cles are about 25% less, and this is probably due to the fact that PMDI changes
the macromolecular motions of wood molecules. UF has a relatively small
effect on the molecular-level dynamics of southern pine wood. In contrast,
PMDI reacts strongly with wood molecules and penetrates deeply into the
adherent wood, inducing rapid stiffening of the composite.
Table 6.9. Proton longitudinal relaxation time 'HT,p (ms) for solid pine wood, for PMDI/pine
and for UF/pine expressed as a function of the chemical shift (ppm) for lignin, hemicelluloses
and cellulose. (Marcinko et al. 1998, courtesy of Forest Products J)
Components Chemical 'HT,p 'HT,p for UF 'HT,p for PMDI

shift for pine coated-pine coated-pine
(ppm) (ms) cured (ms) cured (ms)
Lignin 54 8.0 8.5 6.3

Lignin 150 8.5 8.8 6.2
Hemicellulose 21 9.2 9.5 5.5
Hemicellulose 173 9.5 9.2 6.2
Cellulose 73 9.7 8.8 7.9
Cellulose 105 10.1 8.9 8.0
Applications 269
7.00 r-----------------,
6.30
I0..
......
5.60
<> ?i...
........... '0
:t 4.90 .<5 .................... .
4.20 .~......
C"!
--
. ~ ••.... ::III.IEi·····I •••••••• •••i!!!!.
~
••.
3.50 L---L-----'-_.l...---'-----'_-'----'-_L---L-----'-----'
o 2 3 4 5 6 7 8 9 10 11
cure time (minutes)
Fig.6.41. The relaxation time 'H T,p (ms) for flakes cured with phenol formaldehyde resin at
110°,135° and 175°. (Schmidt and Frazier 2000; courtesy of Wood Fiber Science)
The challenge of NMR laboratory testing is to the analyze the adhesive bond
lines on pilot and industrial levels. Schmidt and Frazier (2000) demonstrated
the viability of cross-polarization magic angle spinning (CP/MAS) NMR mea-
surements for evaluating the degree of adhesive cure (phenol-formaldehyde)
in poplar flakeboards. Figure 6.41 shows the relaxation time IHT 1p (ms) for
flake samples cured with 13C phenol-formaldehyde resin. The curing tempera-
tures investigated were 110, 135 and 175°C. An increase in the relaxation time
1HT Ip with curing time and temperature was observed. These results give infor-
mation on the effects of temperature on the adhesive bond lines. CP/MAS
spectra of 13C flake samples of the skin and core of the boards at 13 and 24%
moisture contents are shown in Fig. 6.42. Skin samples are different from core
samples in the range of 35 and 63 ppm, which correspond to methylene and
hydroxymethyl, respectively. Skin samples have an increased methylene peak
and a diminished hydroxymethyl peak compared with core samples. The mean
ratios of corrected peak integrals of methylene carbon to the hydroxymethyl
carbon at 13% moisture content was for the core 0.8 and for skin, 1.6. At 24%
moisture content, these ratios were 1.0 and 1.8. Data for mean ratios of cor-
rected peak integrals for the core and skin were significantly different at K <
0.001. This difference is explained by the temperature difference between
the high temperature surface regions (200°C) and the cold face and the core
(150°C).
In conclusion, it can be noted that for pilot-scale flakeboards, the relative
degrees of resin cure are measurable by proton longitudinal relaxation time
eHT 1p ) and by the ratios of corrected peak resin integrals on CP/MAS spectra
of 13C nuclei.
110 100 90 80 70 60 50 40 30 20 10
ppm
Fig. 6.42. CP/MAS spectra of BC-labeled flake samples of the core and of the face of the board at
13 and 24% moisture contents. (Schmidt and Frazier 2000; courtesy of Wood Fiber Science)
6.4.3.3 Control of Adhesion
Industry requires reliable methods to evaluate the properties of adhesive-

bonded systems. Niemienen and Koenig (1988) used NMR imaging to test the
joint between polyvinyl acetate and pine wood. The behavior of the adhesive
bond formed between water-based polyvinyl acetate adhesive and pine was fol-
lowed during curing. Cylindrical specimens of furniture grade wood of 35 mm
diameter and 70 mm length were studied. The wood specimens were typical
furniture grade pine wood of 35mm diameter and 70mm length. The speci-
mens were split into two equal parts. In addition, a trough of 50 x 11 x 3 mm
was cut into one of the wood pieces to make it possible to follow the behavior
of the adhesive. The specimens were bonded with 50 g/m2 of adhesive
emulsion, and after 1 min the system was pressed by hand and tightened with
paraffin.
All the images taken during curing were obtained on an NMR imaging spec-
trometer at 300 MHz. The experiment used a standard Carr-Purcell spin-echo
pulse sequence. As can be seen in Fig. 6.43, the kinetics of drying as a function
Applications 271
a) b) c)
Fig.6.43a-c. Kinetics of drying a water-based polyvinyl acetate adhesive emulsion between two
pine specimens as reported by Niemienen and Koenig (1988) after a lOmin; b 4h; c 16h. (VSP
International Science Publisher, with permission)
of time shows that after 3 min (Fig. 6.43a) a very thin glue line between the
wood specimens had been formed. The ll-mm trough in the specimen is also
evident on the right side of the wood sample.
After 4 h of drying (Fig. 6.43b), the adhesive line between the specimens is
dry and no longer visible on the slice. Only the still wet adhesive in the trough
is imaged. After 16h, the adhesive is still not completely dry, and an empty
space in the trough is visible, probably because the external pressure applied
to the adhesive joint was not sufficient. As expected, the adhesive strength of
the structure is decreased because of the void presence. The NMR signals are
due to water in the adhesive and not water that had penetrated from the adhe-
sive to the wood. The authors noted also that the mobility of thermoplastic
polymers increases at elevated temperatures and that NMR images can be
recorded even at 180 ac.
In routine industrial situations, voids in the adhesive layer must be detected
as early as possible to guarantee the quality of the adhesive bonds in mass pro-
duction offurniture, veneer, glue-lam, etc. The limitation of the proposed tech-
nique is related to the difficulty of detecting solid inclusions in the joints, such
as hard resins, because their very short relaxation times are not imaged. The
main advantage of NMR imaging for adhesive bonded structures is related
to the ease with which the images can be interpreted and to the informa-
tion provided about the adhesive itself and about the whole structure under
inspection.
6.4.3.4 Control of the Impregnation Process
Another attractive idea is to use NMR imaging for the visualization of the
impregnation process of wood with preservatives or other liquids, that is, the
process of fluid absorption.
First, we will discuss the state of the art of the subject related to other
materials, such as fiber-reinforced composites and rocks, and then consider
measurements on wood.
Rothwell and Holecek (1984) demonstrated the feasibility of using NMR
imaging to follow the dynamics of water diffusion in inorganic materials. The
mechanism of water absorption depends on the physical and chemical para-
meters of the composite studied (glass fiber-reinforced epoxy-resin compos-
ite). Rapid characterization of the distribution of absorbed fluid in a polymeric
material is possible.
Kleinberg (1999) and many other authors before him, referred to the earlier
contributions that NMR had made to studies of the mesoscopic properties of
rocks, such as pore sizes and their distribution. Relaxation time measurements
can be used to determine a volumetrically weighted distribution of pore sizes
spanning several orders of magnitude. The quantitative measurement of the
amplitude of the NMR signal and the calibration of proton spin density mea-
surements provide a very accurate method to determine the porosity with a
noncontact and noninvasive technique. Similar methods can be used to inves-
tigate the porosity and penetrability of wood during preservation.
Hall and Rajanayagam (1986), using three-dimensional proton NMR
volume imaging, studied the impregnation of cylindrical maple specimens
of 1.2 cm diameter with preservative containing oxides of Cr, Cu and As and
with 1 M manganese chloride solution (MnCl z). The specimens were soaked
overnight with these substances. With a very coarse spatial resolution, Fig. 6.44
shows the relative intensity of the NMR signals from the "spin warp" technique,
measured on a sample compared with the spin lattice relaxation time T, of the
water protons. These images depend on the relationship between the pulse
intervals chosen for the image measurements and the characteristic relaxation
time T, of the water present in the wood. If the time between pulse sequences
is too short, the water protons in wood have not enough time to relax during
the measurement, and the image can have a low intensity. Moreover, a part of
the signal can disappear before it can be measured, if the signal from the water
has a short spin-spin relaxation time Tz The manganese chloride solution has
Tz = 10-5 s, much shorter than that of the water in the wood that is Tz = 10-3 s.
The Mn ions are magnetic and significantly reduce the T, and Tz relaxation
times of the absorbed fluids, which is an important technical aspect of these
measurements. Consequently, with a relatively short time between pulse
sequences, only a signal from the manganese chloride solution is observed.
This enables the diffusion of the MgCl z into the bath to be measured. The influ-
ence of impregnation time is shown in Fig. 6.45, where one can see the effect
after 3.5 and 0.5 h compared with the control measurements on a fresh wood
specimen.
Applications 273
a)
aDla
b)
c) fresh wood
d) impregnated wood
Fig.6.44a-d. Imaging with the "spin warp" technique of maple specimens impregnated with dif-
ferent preservatives, compared with a control specimen of fresh-cut wood, at 80MHz frequency
for protons and 15 min of total experimental time. The contour levels are logarithmically spaced.
The top contour level is at 0.7 of the tallest peak. The following level is at 0.7 of the previous level
height. The relative intensity of the NMR signal has been compared with the spin-lattice relax-
ation time of water protons. a Planar image of freshly cut maple. b Effect of heating one of them
for 3h at 1l0°C. c Planar image of three pieces of freshly cut wood. The piece in the middle has
been used as reference. d Piece soaked in a preservative that shows a substantial change. (Hall
and Rajanayagam 1986; courtesy of Wood Science Technology)
6.4.3.5 Examination of Archaeological Wood
In this section, we describe NMR applications to the study of wooden archae-

ological objects. High resolution techniques using I3C-NMR have been used
successfully for this purpose (Lambert et al. 2000). The CP/MAS spectra of
archeological wood and fresh wood are similar, as can be seen by comparing
a)
b) 3.5 hours 0.5 hours 0.5 hours
Fig.6.45a,b. The influence of impregnation time with a 1-M solution of MnCI 2 on the distribu-
tion of the preservative in wood. a Fresh wood. b Sample after it had been soaked 3.5 and 0.5 h.
(Hall and Rajanayagam 1986; courtesy of Wood Science Technology)
200 150 100 50 o

ppm
Fig. 6.46. The CP/MAS spectrum of wood. (Lambert et a1.1991, with permission)
Fig. 6.46 with Fig. 6.2. Carbohydrate resonances are found with chemical shifts,
0, between 50 and 90 ppm. The lignin peaks are located between 0 = 100 ppm
and 0 = 160ppm. Archeological wooden artifacts, like elements of old ship-
wrecks, are of particular interest for the development of NMR imaging. These
objects survive only because they are water-logged and proton NMR imaging
is the most appropriate technique for studying their internal structure. The
very long-term action of water changes the physical and chemical properties
Applications 275
of wood. Different parts of excavated ships show different degrees of

degradation.
Cole-Hamilton et al. (1990) examined samples from a Spanish Armada ship
"La Trinidad Valencera", which was excavated in 1970 on the northern Irish
coast, after more than 380 years of submersion. The wood species was clearly
identified as ash. The grain structure and the growth rings were visible with
structural details such as rays. The image resolution was 10 11m. Macroscopi-
cally, the structure of the old ash was not very different from that of fresh ash.
Hatcher et al. (1981) compared the l3C spectra of contemporary spruce with
spruce samples buried 450 years ago in Rotterdam and a 1O,000-year-old
spruce from northern Michigan. The spectra were not very different. Small dif-
ferences were observed in the distribution of the carbohydrate peaks. In the
Michigan sample, the () = 75 ppm peak was relatively strong while the () =
105 ppm peak was much weaker. These changes may be due to the beginning
of the coalification of the wood.
Wilson et al. (1993) used solid state l3C-NMR spectroscopy to analyze the
structure of wooden elements of five old ships submersed in the Indian Ocean.
The structural elements submersed in an anaerobic site appeared to have
undergone little or no degradation. However, samples in the proximity of a
marine environment lost carbohydrates, whereas the inner wood remained
relatively intact. Carbohydrates are subject to hydrolysis and leaching. Fengel
(1991) noted that for most waterlogged woods in a moist and strongly salty
environment, part of the hemicelluloses are solubilized.
Studies on the permeability of waterlogged archaeological wood with a
view to conservation were reported by Robertson and Packer (1999). One-
dimensional NMR profile imaging in oak timber from the Mary Rose ship sunk
in 1545 in Portsmouth, UK, were obtained. Increment cores of 4mm diameter
were extracted at different depths, ranging from 3 to 26mm from a 56-mm-
thick, water-saturated wood piece. After cleaning the wood of all soluble para-
magnetic salts, the samples were introduced into a tube filled with D2 0, and the
subsequent uptake of D20 was monitored at specific time intervals in a one-
dimensional lH-NMR profile experiment. The profiles were obtained with a
spectrometer operating at 20 MHz and equipped with a pulsed field gradient
capacity. The images obtained with a spin-echo sequence were produced with
a 90 0 pulse length of 2.4l1s for radiofrequency excitation and with a 1800 pulse
applied after 2.5 ms to form a spin echo in the presence of field gradient of
0.26t!m, which gave a spatial resolution of 80 11m. The record of the complete
echo was obtained with 1024 points in both real and imaginary channels at a
l-l1s dwell time. The one-dimensional profile of the diffusion of D20 in the lon-
gitudinal anisotropic direction of the wood (Fig. 6.47) was obtained from the
spatially encoded frequency domain signal, determined by Fourier transfor-
mation. The lH signal outside the sample does not change with time. Inside the
sample, the concentration of D2 0 increases from the surface until it reaches
equilibrium. The diffusion coefficients for D20 were calculated as a function
of the depth into the sample and the anisotropic axis, as can be seen in
A 622-10 3s
• 437-10 3s
[J 292-10 3s
• 184-10 3s
V 107-10 3 s
... 55-10 3s
o 23-10 3s
• 7-10 3s
-15 -10 -5 o 5 depht (mm)
Fig. 6.47. One-dimensional NMR profile of the ingress of D20 in archeological oak (from the
Mary Rose ship) in longitudinal direction as a function of the depth of the sample_ (Robertson
and Packer 1999, with permission)
Fig. 6.48. The higher diffusion coefficient, as expected, was measured in the lon-
gitudinal direction. In this direction the diffusion pathway is along the fibers
and vessels and is similar to the pathways of fresh wood, but in an archeologi-
cal sample the diffusion can be blocked by deposits of calcareous material,
which can also block the diffusion through the medullary rays. In addition, the
permeability in the radial direction can be enhanced by the deterioration of pit
membranes. The permeability of archeological wood is considerably modified
by the degradation process. The solid core is less degraded and has a high
content of cellulosic components, but it is surrounded by a heavily degraded
surface layer. The variation of the diffusion coefficients in all anisotropic direc-
tions, expressed as a function of the depth of the specimen corresponds to a
typical pattern of deterioration of buried archaeological wood, with more
degradation at the surface than in the internal part of the specimen.
The potential areas of use to NMR imaging and spectroscopic techniques
are related to the identification of the tree species of archeological objects, esti-
mation of the age of samples from the ratio of unsaturated (lignin) to satu-
rated (carbohydrate) resonances, to assessment of the level of decomposition
of archeological wood prior to conservation and to the measurement of the
ingress of preservatives into the wood in order to stabilize it
6.4.4 Further Applications for the Measurements of Elastic Constants
NMR imaging for the measurement of the elastic properties of bulk solids was
first proposed and realized by Lewa (1991, 1992, 1996). This technique uses the
Applications 277
+· .
2.5 I
• longitudinal
. .
2.0
.. • tangential -
·
'·1- ... radial
~
~ 1·.-1
N
.L'.
51.5
.
'''''-
'"0 :, f·.-1
~
ci
1.0
,
.. ..... ~.!..j..... J-!-{ ................ f-•• j
.. f-H.
"~~~
0.5 ~~.~
·····················~·················t!=1
.......................
•
o I I 1
o 5 10 15 20 25 30
Oepht (mm)
Fig. 6.48. Diffusion coefficient of D,O as a function of depth and of anisotropic direction in
archeological oak from the Mary Rose ship. (Robertson and Parker 1999, with permission)
magnetic field gradient located in time and space and relates the spatial coor-
dinates with frequency. Such transformations are possible when the position
of an element or the value of the magnetic field intensity are time-dependent.
For a time-independent magnetic field with a constant gradient, the Larmor
frequency can be modulated by varying the posi,tion of the sample. This can
be accomplished by using a longitudinal mechanical wave. The modulation of
the magnetic field intensity affects the nuclei present in the displacement
element and depends on the argument of Bessel's functions. This argument is
expressed as a function of the angle between the displacement and the field
gradient vectors and the coefficient of the adiabatic compressibility of the
medium. The latter depends on the elastic properties of the solid and on instru-
mental factors such as: the type of paramagnetic nuclei, the adiabatic com-
pressibility and the density of the solid, and wavelength and velocity of the
mechanical wave.
The presence of the mechanical wave in the MRI field enables the display of
the bulk object by modulation of the Larmor frequency with a mechanical
wave applied to the sample in the presence of the magnetic field gradient. The
possibility of generating mechanical waves with wave-lengths comparable to
the MRI resolution and the cellular size, enables observations at a cell level.
Until today, the main application of this elastomagnetic resonance spec-
troscopy has been in the biomedical field, but extensions of such a technique
to characterization of materials, acoustic emission and photo-acoustics appear

to be possible and should be investigated.
6.5 Summary
NMR imaging is one of the most powerful and versatile techniques for the
characterization of materials. The development of high speed computers and
the introduction of high field superconducting magnets have increased the
speed, sensitivity and flexibility of such characterizations. Only the cost of such
equipment limits the utilization of NMR imaging in the fields of wood science
and technology. Standard NMR parameters are the resonance frequency,
the magnitude of the signal proportional to the density of the nuclei, the
spin-lattice relaxation time T 1, the spin-spin relaxation time T2 , the diffusion
coefficient, the flow velocity, the spin-spin coupling and the spectrum of cor-
relation times. The values of these parameters depend on species, moisture
content, physiological parameters of wood and on several instrumental and
measurement factors such as the Larmor precession frequency, temperature,
etc.
The NMR tomographic image depends on the scanning technique used,
pulse sequence, and magnetic field intensity. The NMR imaging technique
relies on the interaction of nuclear magnetic moment (nuclei) in only a small,
controlled zone of the sample under inspection, and is achieved by placing the
measured body in a spatially inhomogeneous field. Its nuclear resonance fre-
quency is matched to the RF signal only in the corresponding zone of the
object.
The NMR imaging technique can be used to investigate the spatial distri-
bution of all parameters that can be determined by NMR, such as densities, T 1>
T2 , diffusion terms, etc. Usually, the nuclear spin density and relaxation time
are mapped as a function of their spatial position. Using basic spatial encod-
ing and slice selection principles, different techniques are available to form
I-D, 2-D or 3-D images, using various spin-echo, stimulated-echo and
gradient-echo pulse sequences. NMR signals inherently depend on nuclear
relaxation time constant, which, in turn reflect the structural environment of
the emitting nuclei. There are several modalities of spatially encoding the
signals. One of these is to apply a linear magnetic field to the original static
field. In this way, nuclei on one side of the sample will feel a weaker total mag-
netic field than those on the other side. From such a set of data, the image of
the sample is reconstructed with an appropriate algorithm. Conventional NMR
spectroscopy can be coupled with the corresponding imaging technique, and
chemical structures of the specimen can be determined.
The main advantages of the NMR imaging technique are: the method is non-
destructive, non contact, relatively rapid, can be used in situ and in vivo, and
does not induce any structural damage. In addition to providing a relative
Summary 279
mapping of solid structural inhomogeneities, fluid (water, preservative solu-

tions, etc.) distributions can be observed. The technique has the potential to
provide an absolute measure of fluid absorption. A T2 relaxation map can dis-
tinguish between free and bound water. The bound water is strongly bonded
to the cellulose and has a much shorter T2 than the free water. Because the NMR
relaxation rates depend on the freedom of molecules to move, they are sensi-
tive to indicators of the chemical and physical characteristics of the sample.
For measurements in the presence of fluid-solid interfaces, the decay curve is
a probe of the length scale of the structure. Applications such as measurement
of the moisture content distribution in wood and in wood-based composites,
continuous monitoring of lumber drying, adhesive curing and impregnation
of wood with preservatives can be implemented in industry. NMR imaging is
one of the most powerful new techniques for wood science and technology,
and can be used to monitor industrial processes on a continuous basis in a
production line. Future availability of low-cost and easy to use devices will
contribute to the more widespread use of this technique.
7 Neutron Imaging
7.1 Introduction
Neutron radiography is a nuclear method for wood structure imaging. Rele-

vant aspects about the physics of neutron scattering and about materials
science-oriented neutron scattering studies have been published in reference
books and articles such as: Guinn and Lukens (1965), Bacon (1975), Garret and
Berger (1977), Kostorz and Lovesey (1979), Nicklow (1979), Domanus (1992),
and Dreele (1994).
Neutrons are subatomic particles with zero charge and are the major con-
stituent of the nuclei. Their characteristics are: mass mn = 1.675 x 10-27 kg, spin
112, magnetic moment /In = -1.913 nuclear magneton. They have a half-life of
624 s. They carry no charge and their properties can be expressed in terms of
momentum, velocity, wave vector, wavelength, and kinetic energy. Neutrons in
thermal equilibrium with a moderator material near room temperature are
called thermal neutrons and are relatively slow. The Maxwellian spectrum of
thermal neutrons, which peaks at 25 me V at room temperature, corresponds to
a neutron wavelength of about 0.18nm. The velocity of neutrons is 2200m/s at
290 K (Price and Sk6ld 1986). In the case of wood, which is a hygroscopic mate-
rial, a specific hydrogen image can be obtained with neutron radiography.
Today, the neutron radiography provides images with the highest resolution
(15/lm) for water in wood, either in small, clear specimens or in samples of
large size.
Neutron imaging is one of the most recently developed testing techniques
in wood science and forestry and, atthe same time, it is one of the most promis-
ing. The main research field of neutron imaging is related to water flow in solid
wood and trees. Because of its high specificity to water, the neutron beam tech-
nique can image water movement in living cells during meristem development
with a high resolution. So far, only few laboratories have been involved in the
development of wood structure imaging by the neutron technique, namely, in
Japan at Tokyo University and in Switzerland at ETH Zurich. We note the pio-
neering work of Nakanishi (Nakanishi et al. 1998a; Furukuwa et al. 2001) on
the development of the specific neutron imaging technique to visualize the
moisture content in plants and in fresh wood specimens cut from living trees.
The applications of neutron scattering to specific problems in wood science
and technology are still very scarce. The interest of wood scientists in neutron
activation analysis was originally related to studies on ion migration in wood
282 Neutron Imaging
(Loos 1965a,b; Langwig and Meyer 1973), to diffusion of the preservatives in

solid wood (Siau and Meyer 1973), to nondestructive moisture content mea-
surements (Chang 1975) and resin penetration (White et al.I977),and,20 years
later, to plants and tree physiology (Nakanishi 1994; Nakanishi and Matsub-
ayashi 1997), forestry ecology (Schultz and Ward Whicker 1982), to timber
(Nakanishi et al. 1998b) and structural elements (Niemz et al. 2000a,b;
Lehmann et al. 2001). This chapter is concerned with a survey of several appli-
cations of neutron radiography to wood science.
7.2 Basic Aspects
The development of neutron imaging techniques (Kostorz and Lovesey 1979)

is based on the following:
- the interaction of a neutron beam with a solid is weak;
- the absorption is very small compared with X-rays, and large samples can
be inspected;
- the wavelength can reveal the atomic arrangement in the sample;
- the nonmagnetic scattering amplitude for neutrons varies in a nonmonoto-
nic way from one element to another and, in contrast to electromagnetic
radiation, it does not depend on the number of electrons in an atom;
- being a nuclear property, the scattering amplitudes may differ considerably
for different isotopes of a given chemical species.
The principles of neutron radiography are largely presented in various
books. Detailed comments on this subject are beyond the purview of this
section. A neutron radiographic image of an object is obtained by the attenu-
ation coefficient of a neutron beam that interacts with the matter. The attenu-
ation coefficient can be calculated from the equation:
(7.1)
where I..l is the attenuation coefficient, N is the number of nuclei per cm3 of
attenuating material, O"t is total cross section, which is the probability that a
nuclear reaction will occur and corresponds to the sum of the absorption 0".
and scattering cross sections 0"•.
This equation is valuable for assessing the relative change in transmitted
radiation intensity that might be obtained through the inspected object. In
a radiographic situation, this phenomenon is recorded. The object is placed
in a neutron beam in front of an image detector. Neutron radiography is an
effective method that can provide a specific high resolution hydrogen image
because the attenuation of His 100 to 1000 times higher than that of the other
elements. Images of moisture content distribution will be presented in this
section. The H amount of different components of wood (cellulose,hemicellu-
Equipment 283
loses, lignin, resins, etc.) can be neglected when this amount is compared with
the H present in the moisture of wood (Nakanishi et aI. 1998a).
Development of neutron imaging with computer tomography, in progress
in different laboratories (Furukawa et al. 2001), will give new opportunities in
the future to studies in wood science and tree physiology.
7.3 Equipment
The neutron imaging of wood reported by Nakanishi and coworkers

(Nakanishi et al. 1998a,b) was obtained with a neutron beam produced by an
atomic reactor OJR-3 M) installed at the Japan Atomic Energy Institute in
Ibaraki prefecture. The wood specimens were sealed in vacuum together with
a gadolinium n/y converter and a film. The specimens were irradiated for 19s
with a neutron flux of 8.7 x lO B n/cm3• The image obtained on the film was
recorded on a computer with a CCD camera. The neutron images obtained of
the wood structure were three-dimensional, with z-axis indicating the amount
of moisture present.
The equipment for neutron radiography used by Niemz et aI. (2000a,b,c) has
the characteristics given in Table 7.1. This equipment is located in the Paul
Scherrer Institute at Villigen, Switzerland, and has been described by Bauer
(1998). The neutrons are produced by spallation. The neutron beam line is
shown in Fig. 7.1. The corresponding detector setup is seen in Fig. 7.2. The
moisture content in structural members was determined by image analysis by
subtracting the images obtained for samples under wet and dry conditions.
Table 7.1. Technical characteristics of the thermal neutron beam

line NEUTRA used by Niemz et al. (2000c) for neutron imaging of
wood structural members (with permission)
Technical parameters Units Values
Proton energy MeV 590

Proton beam intensity rnA 2
Thermal neutron flux at sample position cm-2 s-1 3.4
Mean neutron energy meV 25
Beam parallelism (lid) 550
Fraction of thermal neutrons 100
Field of view at sample position mm 350
Spatial resolution of imaging plate 11m 50
Number of pixels per line 1500
Recording area cm2 400
Dynamic range bit 16
284 Neutron Imaging
Fig.7.1. Thermal neutron beam line (SINQ) at the Paul Scherrer Institute in Villigen, Switzer-
land. (Bauer 1998, with permission)
Fig. 7.2. Sample detector setup for neutron radiography. (Niemz et al. 2000c, with permission)
Applications 285
7.4 Applications
In this section, the imaging of the moisture in wood will be studied first on
disks taken from living trees during drying and then on small, clear specimens
during absorption of water For the study of drying phenomena, two cases will
be analyzed, corresponding to short and relatively long drying times. Imaging
of water distribution in structural elements submitted to weather condition-
ing will also be presented.
7.4.1 Imaging of Water Distribution in Trees
The storage of water in a tree is influenced by different factors, such as: the
capillarity of the wood structure and the elasticity of the cell tissue. The water
content in trees is subjected to seasonal changes and is different in sapwood
and in heartwood. Development of new methods for determination of the
water distribution in trees should be emphasized. Among the nondestructive
methods, neutron imaging seems to be promising. Water distribution was
studied with neutron radiography by Nakanishi and Watanabe (1995) who
used trees of different species and by Nakanishi et al. (1998b) with trees belong-
ing to the same species.
7.4.1.1 Water Distribution in Trees of Different Species
Nakanishi and Watanabe (1995) investigated the distribution of water in young

trees of five species (Pinus thunbergii, Metasequoia glyptostroboides, Chame-
cyparis obtusa, Quercus serata and Robinia pseudoacacia) as can be seen from
Fig. 7.3. Table 7.2 gives the characteristics of the disk specimens selected from
those trees. The specimens were irradiated in the atomic reactor of the Japan
Atomic Energy Research Institute with thermal neutrons during 19s. The total
Table 7.2. Characteristics of the specimens irradiated for 2.3min. ( Data from Nakanishi and
Watanabe 1995, with permission)
Species Age of the Specimen Specimen Moisture

tree (years) diameter (em) thickness (em) content
Pinus thunbergii 18 13.5 1-3 Green

Metasequoia glyptostroboides 10 13.2 1-3 Green
Chamecyparis obtusa 19 12.6 1-3 Green
Quercus serata 18 16.5 1-3 Green
Robinia pseudoacacia 8 12.3 1-3 Green
286 Neutron Imaging
d
Fig.7.3a-e. Photographic images and neutron radiography of different species. a Pinus
thunbergii. b Metasequoia glyptostroboides. c Chamecyparis obtusa. d Quercus serata. e Robinia
pseudoacacia. (Nakanishi and Watanabe 1995, with permission)
flux was obtained with (n, y) reaction and was 8.7 x 107 n/cm 2• The water dis-
tribution image was obtained on a film. The resolution of the image was
15 11m. The degree of brightness in the image was proportional to the water
content of the sample.
Having these images as a background, it can be noted that in Pinus thun-
bergii within the same ring, a heterogeneous water distribution has been
observed. The black zone in the pith indicates the decrease of water content,
suggesting the beginning of the necrosis of the tissue. In the image for Metase-
quoia glyptostroboides, we can see that only a small amount of water is
Applications 287
observed in the central zone. The maximum amount of water seems to be

located in the last annual ring. The knots are not precisely defined on neutron
radiography probably because of their lower moisture content. In Chamecy-
paris obtusa the distribution of moisture is very nonuniform, as can be seen
from the image of white irregular rings.
The exact pattern of moisture distribution in Quercus serata is difficult to
identify. The annual ring with water deficiency is represented by a dark zone
that is easy to recognize. Parallel zones with different flow properties occur in
oaks, which have very large earlywood vessels, resulting in a very high longi-
tudinal permeability. The majority of other tissues in the annual ring consist
of thick-walled fibers with extremely low permeability making it difficult for
liquids to penetrate except at very high pressure. This extreme nonhomo-
geneity of wood structure is one of the principal reasons why it is difficult to
observe the moisture content pattern in oak wood.
The photographic image for Robinia pseudoaccacia clearly shows the pres-
ence of the heartwood. In neutron radiography, the heartwood zone was not
revealed, and we can suppose that there was no difference in moisture con-
tent between heartwood and sapwood. The vessels with tyloses in which no
water is present are observed as dark points. The climatic conditions during
the growth of the second, third and fourth annual rings are well illustrated
by the presence of the false rings, which are represented by different gray
levels.
7.4.1.2 Water Distribution in Trees of the Same Species
Nakanishi et al. (1998b) studied the moisture distribution in the inner part of
the sapwood of different cultivars of sugi (Cryptomeria japonica) with special
emphasis on the identification of the presence of a so called white zone in the
sapwood, characterized by a high moisture content. This zone is probably
caused by genetic factors and environmental conditions. It is present in all
specimens, as shown in Fig. 7.4.
In three-dimensional representation of water distribution, the aspect of the
"white zone" is very well defined for all specimens (Fig. 7.5). In addition, near
the bark, this distribution of moisture is not superimposed on the pattern of
the annual rings. Nakanishi and Matsubayashi (1997) suggested that this bright
zone observed by neutron radiography could also indicate the presence of dif-
ferent chemical components rich in H, which are different from those in water
from the cell wall.
In conclusion, it can be said that with neutron radiography, the visualiza-
tion of a patterns of water distribution across the transverse section of a tree,
mainly near the bark, was possible for all species studied, Neutron imaging
allows studies on water distribution in zones near the bark, which are difficult
to perform with other techniques.
288 Neutron Imaging
Fig. 7.4. Photographic images (A, B, C, D) and corresponding neutron images (a, b, c, d) of dif-
ferent cultivars of sugi (Cryptomeria japonica). A 24-year-old cultivar, 25 Gou. B 25-year-old cul-
tivar Honjiro. C 29-year-old cultivar I-Gou. D 3D-year-old cultivar, Sanbusugi. (Nakanishi et al.
I998b, with permission)
7.4.2 Imaging of Moisture Content in Lumber During Drying
Established wood drying techniques are based on the removal of the large
amount of moisture present in green wood and have the purpose of protect-
ing wood materials from excessive drying stresses that can cause defects and
degradation. For wood quality control during drying, several nondestructive
methods have been developed. In this section, we will discuss some laboratory
measurements undertaken with the neutron technique for imaging the kinet-
ics of drying.
Applications 289
Fig. 7.5. Three-dimensional water distribution in a disk of l-cm thickness of green Cryptomeria
japonica. (Nakanishi and Matsubayashi 1997, with permission)
704.2.1 Short Drying Time
Nakanishi et al. (l998b) studied the kinetics of drying of Cryptomeria japon-

ica wood (Fig. 7.6) during 6h at 60°C and at 90% relative humidity, using the
same specimens described in Section 704.1. At the initial stage, which corre-
sponds to green saturated conditions, all disks give a very bright image. An
important difference between the sapwood and the heartwood is well visible
with disks A, Band C. The moisture distribution seems to be uniform within
the annual rings. Disk D shows a similar and very high moisture content in
both sapwood and heartwood. During the drying process of 2, 4 and 6 h, the
moisture content still remains high in the sapwood. The heartwood becomes
darker with increasing drying time. Figure 7.7 is relevant for the profile of
moisture content in wood. The value of the moisture content in sapwood and
in heartwood was calculated precisely as a function of the corresponding area.
For all disks, the moisture content near the bark was higher than that in the
sapwood and in the heartwood. For disk A, the effect of drying time on the
profile of the moisture content in the heartwood seems to be less important
than on sapwood. After 6 h of drying the moisture content in heartwood was
very near that in the sapwood.
From the profile of moisture content distribution, it can be noted that at the
beginning of the drying process, disks Band C have a similar behavior as disk
N
'"a
~
a
8::l
S
~
&'
a)
Fig.7.6a. Imaging of the kinetics of moisture content distribution. a For 6h drying at 60°C at 90% relative humidity of four disks (A, B,
C, D as in Fig. 7.4) of I-cm thickness of Cryptomeria japonica.
Applications 291
255
-
c:
;:,
0
0
255
E
I'll
GI 0
>
:;: 255
I'll
...
li
0
255
0
b) s S H S S H S
Fig.7.6b. Variation of the moisture content profile of four disks A, B, C, D for which the numbers
0,2,4,6 are the drying hours. (Nakanishi et al. 1998b, with permission)
Fig. 7.7. Neutron radiography of spruce drying as a function of time. (Lehmann et al. 2001, with
permission)
A, but after 6 h of drying, the moisture content is higher in the sapwood than
in the heartwood. For all these three disks the heartwood represents about 33%
of the transverse section. By contrast, for disk D, the zone occupied by the
heartwood seems to amount to about 70% of the transverse section. After 6 h
of drying, the heartwood lost a considerable amount of water compared with
the sapwood. The causes of the different moisture distributions are unknown,
but the effects of such differences on the wood during drying can be disastrous
because of the development of a nonuniform drying stress.
7.4.2.2 Long Drying Time
A longer drying process (45 h) was studied with spruce specimens (Fig. 7.7) by
Lehmann et al. (2001). During the drying process at 20°C and 65% relative
292 Neutron Imaging
2h 6h 22h 30h 70h

Fig.7.S. Kinetics of spruce drying. (Lehmann et al. 2001, with permission)
humidity, the wood moisture content decreased from a saturated state to air
dry at about 18%. After 22 h of drying, some anatomic details at the transverse
section of specimens were observed. After 4S h, the disposition of the annual
rings was clearly visible, together with shrinkage of the specimen. The regular
square cross section of the specimen in the water-saturated state had been
transformed during drying into a rhomboid-shaped section. The minimum
shrinkage was along the radial direction of the wood (at 4S0 in the figure)
which corresponds to the direction of the rays. The maximum shrinkage
occurred in the tangential direction to the annual rings. The drying kinetics
(Fig. 7.8) have been investigated by advanced image processing, by subtract-
ing image characteristics in two different drying stages. The image of the sat-
urated sample has been considered as a reference. During drying, a crack
observed in the upper part of the image (after 8h of drying) increased in size
and propagated in the tangential direction to the annual ring. After 70 h of
drying, the crack is well visible on the external, dried area of the specimen.
7.4.2.3 Imaging of Water Absorption
Wood can be protected from the attack of biological agents by applying chem-
ical preservatives. The uniform distribution and sufficient penetration of
these substances can greatly increase the life of wood structures. For an under-
standing of the complex phenomena during the impregnation and for the
determination of preservative effectiveness, it is necessary to utilize nonde-
structive techniques able to visualize in real time the absorption of liquids by
the materials. This section deals with the determination of water flow in solid
wood and wood-based composites with the neutron imaging technique.
Different wood species (beech, pine, spruce, sweet chestnut), particleboard
and MDF specimens (Lehmann et al. 2001) were partially immersed in water,
as shown in Fig. 7.9. Quantitative water absorption as a function of time is
shown in Fig. 7.10. After Sh of immersion, the beech samples absorbed 199
of water, while all other specimens absorbed about 4 g of water. After 22 h of
immersion, the maximum water absorption was 2S g for beech. During the
same period, water absorption was at a minimum, (about 1 g) for the MDF
Applications 293
I mill walcr
Fig. 7.9. Experimental device for water absorption by capillarity in wood samples 200 mm long
x lOmm wide x 16mm thick. (Lehmann et al. 2001, with permission)
30
_ _ beech
25
~
/
/
5
/ _partie leboard
_ spruce
o
IL ~
pine
MDF
o 5 10 15 20 25
time (hours)
Fig. 7.10. Water absorption as a function of time for beech, spruce, pine, chestnut, particleboard
and MDF samples. (Lehmann et al. 2001, with permission)
sample. The kinetics of water absorption for beech, pine and particleboard
samples, with neutron radiography as a function of time, is shown in Fig. 7.11.
On the image, the darker zone, which corresponds to the maximum amount of
absorbed water, is observed for the beech samples.
7.4.3 Imaging of Moisture Content Distribution in Structural Elements
The widespread use of wood in the construction of buildings, windows, doors,

etc. has both an economic and an esthetic basis. The beauty of wood is diffi-
294 Neutron Imaging
a)
b)
3 min 13 min 28 min 118min 180 min 300 min

c)
Fig.7.11a-c. Neutron imaging of water absorption in different samples as a function of time.

a Beech; b pine; c particleboard. (Lehmann et at. 2001, with permission)
cult to match with other materials, when architectural considerations are taken
into account. The importance of a proper connection design cannot be over-
stated (Natterer et al. 2000).
On the connections, the effect of weathering may be undesirable, depend-
ing on the requirements for a particular wood product. Natural weathering can
produce on wood connections a nonuniform moisture content distribution
that induces shrinkage and swelling. The accompanying stress thus induced by
shrinkage and swelling causes excessive dimensional changes, such as warping,
surface roughening and checking, loss of some surface fibers, etc. The physi-
Applications 295
Table 7.3. Specimen characteristics for neutron radiography. (Data from Niemz et al. 2000c, with
permission)
Sample Type of joint Material Connection Joint opening

gap size (mm)
Obtuse joggler Spruce/spruce Obtuse meeting, doweled 0.1-0.4

2 Joint, plug Spruce/spruce Slot-mortise joint 0.1-1.0
3 Joint, doweled Spruce/spruce Bevel, doweled 0-0.4
4 Obtuse insert, glued Spruce/spruce Obtuse insertion, glued 0.30-5
5 Obtuse insert, glued Spruce/beech Obtuse insertion, glued 1-1.5
cal deterioration of wood during weathering depends on wood species, density,

growth annual rate, ring orientation, proportion of latewood, etc. The weath-
ering process is usually accompanied by fungal attack.
During artificial weathering, qualitative and quantitative nondestructive
estimation of moisture content distribution in real size window connections
is possible by using neutron radiography, as demonstrated by Niemz et al.
(2000c). For this purpose, different types of joints have been used, as can be
seen from Table 7.3. For producing structural elements, a single species
(spruce) or two different species (spruce and beech) were used. The dimen-
sions of the samples were 200 mm length, 50 mm width and 30 mm thickness.
The specimens were exposed to different cyclic conditions of relative humid-
ity and temperature as for the first cycle, 120 h at 25% relative humidity and
30°C followed by the second cycle of 120 h at 99% relative humidity and 30°C.
After the second cycle, the samples were conditioned for 24 h at 50°C and 2%
relative humidity. To induce maximum water absorption in the joints, and to
simulate rain, the specimens were submerged in water for 2 h. Radiographic
images were taken of each sample in dry and moisturized stages, as can be seen
from Fig. 7.12a for the spruce/spruce doweled corner connection and in Fig.
7.12b for an obtuse insertion (spruce/beech, glued). For all samples, a visible
gap was observed, ranging from 0.1 to 1.5mm. The profile of the moisture
content as a function of the geometry of the sample was determined (Fig. 7.13).
Water absorption is higher in the transverse section of the spruce specimen.
Moisture content is increasing continuously in the spruce up to the joint. In
beech, the moisture content seems to be constant and no absorption was
observed in the transverse section.
Performance of window connections can be optimized with neutron
imaging and can be evaluated at three levels - short-term, long-term and
product quality assurance. In the short tertn, the designer must take into con-
sideration the properties of wooden connections (moisture level, temperature,
strength and rigidity of the members). The long-term performance of the con-
nections is equated with the ability of the joint and windows to withstand the
296 Neutron Imaging
Dry
Wet
.)
Dry
Fig.7.12a,b. Moisture content distribution in corner connections for dry and moisturized
samples. a Spruce/spruce doweled. b Obtuse insertion spruce/beech, glued. (Niemz et al. 2000c,
with permission)
Summary 297
0.25
~ joint joint
.r: 0.2
c..
0)
"0 _ dry specimen
"0 0. 15 wet specimen
0)
.~
<a 0.1
E
0
spruce bccch sprLIce
c 0.05
"1 Of' <'! r--: <'! r--: <'! r-:
.",. r'l <'l N r'l
Of,
N N
r-:
'"
N <'l .",. '0 00 0
a) distance (mm)
M 0.16
E
<) 0.14 I
Ob
'-' 0.12 I
c 0.1 I
.B 0.08 1
13
u 0.06
~ 0.04
::l
'iii 0.02 1
'0
E 0
0 20 40 60 80 100 120
b) distance (mm)
Fig. 7.13. Moisture distribution profile as a function of the geometry of the joint for obtuse inser-
tion (spruce/beech). a Normalized depth as a function of distance (length of the specimen).
b The corresponding moisture content as a function of distance. (Niemz et al. 2000c, with
permission)
degradation induced by an increasing rate of moisture level, temperature, the

presence of microorganisms, the mechanical level, etc. The maintenance of the
manufacturing process to assure that the product will perform to a certain
quality level is the major concern of the product quality assurance, in detect-
ing unacceptable connectors and products, and in detecting the cause and
correcting it.
7.5 Summary
Neutron beam imaging of wood structure is one of the most recent nonde-
structive techniques developed for this material. The neutrons are subatomic
particles with zero charge. They are the major constituents of nuclei and are
produced in fission reactors. Neutrons in thermal equilibrium with a moder-
ator material near room temperature are called thermal neutrons. Neutron
radiography of wood provides a specific hydrogen image of high resolution
(IS ~m) and allows the imaging of water distribution in wood and in living
wood cells. The interaction of neutrons with wood is weak, and the absorption
is small and, consequently, large samples can be inspected. Today, the applica-
298 Neutron Imaging
tions of neutron imaging to specific problems of wood science and technology

are still very scarce because of cost limitations.
This chapter is concerned with several applications of neutron radiography
for the imaging of water distribution in wood at microscopic and macroscopic
levels. Water distribution imaging in disk-type specimens recovered from
living trees was obtained by irradiation with a thermal neutron flux of 8.7 x
107 n/cm 2 for 19s. The degree of brightness of the image is proportional to the
water content of the sample. The kinetics of the drying process in lumber can
also be observed. The method was successful for the imaging of moisture
content distribution in corner connections of wood elements of real size.
References
Abragam A (1961) Principles of nuclear magnetism. Clarendon Press, Oxford, 599 pp

American Society for Testing and Materials (1986) Standard test for loss characteristics and
dielectric constant of solid electrical insulating materials. ASTM D, Philadelphia, pp 150-
174
Andersen AH, Kak AC (1984) Simultaneous algebraic reconstruction technique (SART): a supe-
rior implementation of the art algorithm. Ultrasonic Imaging 6:81-94
Anfodillo T, Sigalotti GB, Tomasi M, Semenzato P, Valentini R (1993) Application of a thermal
imaging technique in the study of the ascent of sap in woody species. Plant Cell Environ
16:997-1001
Araujo CG, MacKay AL, Hailey JRT, Whittall KP (1992) Proton magnetic resonance techniques
for characterization of water in wood; application to white spruce. Wood Sci Technol 26:
101-113
Arfvidsson J, Lindgren 0, Wiberg P (1997) The development of a moisture sorption model to
predict moisture distribution within solid wood based on nondestructive moisture gradient
measurements using CT-scanning and digital image processing. Int Conf COST Action E 8
Theme: Wood - Water Relations Copenhagen Denmark, pp 107-120
Arita K, Kuratanik L (1984) Wooden pole tester for determining the strength of decayed wooden
poles. Jpn Telecommun Rev 263:167-173
Asplund T, Johansson LG (1984) Feasibility study of X-ray computerized tomography in research
and development for wood-mechanical industry and forestry. (Forstudie-datortomograf for
trateknisk och skoglig forskning och utveckling). Trateknik Rapport, Sweden, no 53, Svenska
Traforskningsinstitutet, A no 904, Stockholm, 35 pp
Bacon GE (1975) Neutron diffraction. Oxford Univ Press, London, 426 pp
Badel E (1999) Determination of elastic and shrinkage properties of an annual ring of oak in
transversal plan: description of the morphology, measurements of microscopic properties and
homogenisation calculations. (Determination des proprietes elastiques et du retrait d'un
cerne annuel de ch~ne dans Ie plan transverse: description de la morphologie, mesure des
proprietes microscopiques et calcul d'homogeneisation). These de Doctorat it l'ENGREF
Nancy, France, 176 pp
Bahr AJ (1982) Microwave nondestructive testing. Gordon and Breach Science Publishers, New
York, 86 pp
Bahyl V (1998) Tomography of the internal structure of wood. In: Kurjatko S, Kudela J (eds) Wood
structure and properties. Arbora Publishers, Zvolen, pp 55-58
Baker JR, Mitchell PL, Cordey RA, Settle JJ (1994) Relationship between physical characteristics
and polarimetric radar backscatter for Corsican pine stands in Thetford Forest UK. Int J
Remote Sensing 15:2827-2849
Balageas DL, Levesque P, Nacitas M, Krapez JC, Gardette G, Lemistre M (1996) Microwave holog-
raphy revealed by photothermal films and lock-in IR thermography: application to electro-
magnetic materials NDE. Proc SPIE 2944, pp 55-66
Barducci I, Pasqualini G (1948) Misura dell' attrito interno delle costanti elastiche dellegno. (Mea-
surement of the internal friction and of the elastic constants of wood). Nuovo Cimento 5:
416-502
Barnes D, Admirall L, Pike RL, Mathur VNP (1976) Continuous system for the drying of lumber
with microwave energy. For Prod J 26 5:31-42
300 References
Bauer G (1998) Operation and development of the new spallation neutron source SINQ at Paul
Scherrer Institut. Nuclear Instrum MethodsPhys Res B 139:65-71
Bauer G, Dai Y, Wagner W (2001) SINQ, Layout, operation applications and R&D to high power.
Proc Symposium Structural materials for hybrid systems: a challenge in metallurgy, Paris,
29-31 October,6 pp
Bauer M, Guntrum C, Ota M, Rippel W, Busse G (1992) Thermographic characterization of defects
and failure in polymer composites. In: Balageas D, Busse G, Carlomagno GM (eds) Proc Quan-
titative Infrared Thermography (QIRT92). Editions Europeennes Thermiques et Industrie,
Paris, pp 141-144
Beall F (1996) Future of nondestructive evaluation of wood and wood-based materials. Holz-
forsch Holzverwert 5:73-75
Beatty R, Choffel D, Charpentier P (1999) Towards products tracking into wood industries. Proc
IWMS:325-334
Beaudoin JL, Merienne R, Danjoux R, Egee M (1985) Numerical system for infrared scanners and
application to the subsurface control of materials by photothermal radiometry. Infrared Techn
Appl SPIE 590:287
Bennet CA, Patty RR (1982) Thermal wave interferometry: a potential application of the pho-
toacoustic effect. Appl Opt 21:49-54
Benson-Cooper DM, Knowles RL, Thomson FJ, Cown DJ (1982) Computed tomographic scan-
ning for the detection of defects within logs. FRI Bull For Res Inst N Z 8:9
Beravs K, Oven P, Sersa I, Torelli N, Demsar F (1998) Electric current density imaging of pedun-
culate oak (Quercus robur 1.) twigs by magnetic resonance imaging. Holzforschung 52:
541-545
Beres C, Fenyvesi A (1994) The role of computer tomographical methods in investigation of water
transport in oak trees. Proc 1st Eur Symp on Nondestructive Evaluation of Wood. University
of Sopron, Sopron, Hungary, 525
Berglind H, Dillenz A (2000) Detection of glue deficiency in laminated wood with thermography.
12th Symp Nondestructive Testing of Wood, University of Western Hungary, Sopron, pp 413-
420
Bernard P, Chalebois A, Pearson N, Samson M (1996) Application of infrared ellipsometry to
detection oflocal grain organization and spiral grain in wood. 10th Symp NDT of Wood. Press
Univ Romandes, Lausanne, pp 165-174
Bernatowicz G (1994) Determination of moisture distribution and check formation by means of
thermography. (Bestimmung des Feucheverteilung und Rissbildung mittels Thermographie).
Ann Warsaw Agric Univ For Wood Technol 45:65-69
Berndt H, Johnson GC (1994) Examination of wave propagation in wood from a microstructural
perspective. In: Thompson DO, Chimenti DE (eds) Review of progress in quantitative nonde-
structive evaluation, vol 14b. Plenum Press, New York, pp 1661-1668
Berndt H, Schniewind AP, Johnson GC (1999) High resolution ultrasonic imaging of wood. Wood
Sci TechnoI33:185-198
Berndt H, Schniewind AP, Johnson GC (2000) Ultrasonic energy propagation through wood:
where, when how much. Proc 12th Symp Nondestructive Testing of Wood. University of
Western Hungary, Sopron, pp 57-65
Bhardwaj MC (2000) High transduction piezoelectric transducers and introduction of noncon-
tact analysis. In: Harvey JA (ed) Encyclopedia of smart materials. Wiley, New York, 25 pp
Bhardwaj MC, Neeson I, Langron ME, Vandervalk E (2000) Contact free ultrasound: the final fron-
tier in nondestructive materials characterization. Paper at 24th Conference: an International
conf on engineering ceramics and structures. 24-27 Jan 2000, The American Ceramic Society,
Cocoa Beach, FL, USA
Biagi E, Gatteschi G, Masotti L, Zanini A, Cerofolini M, Lorenzi A (1994) Tomografia ad ultrasuoni
per la caratterizzazione difettologica dellegno. (Ultrasonic tomography for defect character-
ization in wood). Alta Frequ Riv Elettronica 6 2:48-57
Biernacki JM, Beall FC (1993) Development of an acousto-ultrasonic scanning system for non-
destructive evaluation of wood and wood laminates. Wood Fiber Sci 25:289-297
References 301
Birkeland R (1985) The status of tomographic scanning as a tool for detecting internal log defects.
Proc 5th Nondestructive Testing of Wood Symposium, Washington State University. Eng Pub-
lications, Pullman, pp 231-237
Birkeland R (1990) NDE facilities and research program in Norway. 7th Symp on Nondestructive
Testing of Wood, Washington State University, Madison, pp 57-62
Birsk AS (1972) Particleboard blow detector. For Prod J 22 6:23-26
Bodig J (1994) NDE of wood in North America. Concept and applications. Proc First Eur
Symp on Nondestructive Evaluation of Wood. University of Sopron, Sopron, Hungary, pp
1-l3
Bodig J, Jayne BA (1982) Mechanics of wood and wood composites. Van Nostrand Reinhold,
New York, 712 pp
Bolomey JC, Pichot C (1990) Microwave tomography: from theory to practical imaging systems.
Int J Imag Syst TechnoI2:144-156
Bourbie T, Coussy 0, Zinszner B (1987) Acoustics of porous media. Gulf Publ Co, Houston, TX,
334 pp
Bowyer JL (2000) Wood science in a changing world. Where are we headed? Wood Sci Technol
34:175-181
Brown DP, Pratum TK, Bledsoe C, Ford ED, Cothern JS, Perry D (1991) Nonivasive studies of
conifer roots: nuclear magnetic resonance imaging of Douglas fir seedlings. Can J For Res 21:
1559-1566
Brown JH, Davidson RW, Skaar C (1963) Mechanism of electrical conduction of wood. For Prod
J Oct:455-459
Brownstein KR (1980) Diffusion as an explanation of observed NMR behavior of water absorbed
on wood. J Magn Reson 40:505-510
Buckley J (2000) Air-coupled ultrasound:... a millennial review. 15th World Conf On NDT, Roma
http://www.ndt.netlarticle/wcndtOO/paper/idn507/idn507.htm
Bucur V (1980) Anatomical structure and some acoustical properties of resonance wood. Catgut
Acoust Soc Newsl, series I, no 33:24-29
Bucur V (1986) Les termes non diagonaux de la matrice des rigidites du bois. (Off-diagonal terms
of stiffness matrix of wood) Holzforschung 40:315-324
Bucur V (1990) NMR imaging of green wood. INRA Document a distribution limite, Institut
National de la Recherche Agronomique, Champenoux, France, 10 pp
Bucur V (1992) Anisotropy characterization of structural fiakeboards with ultrasonic methods.
Wood Fiber Sci 24:337-346
Bucur V (1995) Acoustics of wood. CRe Press, Boca Raton, 284 PP
Bucur V, Archer RR (1984) Elastic constants of wood by an ultrasonic method. Wood Sci Technol
18:255-265
Bucur V, Rasolofosaon PNJ (1998) Dynamic elastic anisotropy and nonlinearity in wood and rock.
Ultrasonics 36:8l3-824
Bucur V, Rocaboy F (1987) Surface wave propagation in wood: prospective method for the deter-
mination of wood off-diagonal terms of stiffness matrix. Ultrasonics 26:344-347
Bucur V, Lanceleur P, Roge B (2001) Acoustic properties of wood in tridimensional representa-
tion of slowness surfaces. Ultrasonic International Symp Delft, p 6
Burov VA, Rumyantseva OD (1996) Linearized inverse problem of scattering in monochromatic
and pulse modes. Acoust Phys 40:34-42
Busse G (1979) Optoacoustic phase angle measurement for probing a metal. Appl Phys Lett 35:
759-760
Busse G (1988) Imaging with optically generated thermal waves. Physical acoustics, vol XXII.
Academic Press, New York, pp 403-478
Busse G (1994) Nondestructive evaluation of polymer materials. NDT Eval Int 27 5:253-
262
Busse G (2001) Lockin thermography. In: Maldague X (ed) Nondestructive testing handbook, vol
3. Infrared and thermal testing. American Society for Nondestructive Testing Inc, Columbus,
OH, pp 318-327
302 References
Busse G, Bauer M, Rippel W, Wu D (1992a) Lockin vibrothermal inspection of polymer

composites. In: Balageas D, Busse G, Carlomagno GM (eds) Proc Quantitative Infrared
Thermography (QIRT92). Editions Europpelennes Thermiques et Industrie, Paris, pp 154-
159
Busse G, Wu D, Karpen W (1992b) Thermal wave imaging with phase sensitive modulated ther-
mography. J Appl Phys 71 8:3962-3965
Byrne GF, Fenn MD, Burgar MI (1986) Nuclear magnetic resonance studies of water in tree sec-
tions. Agric For MeteoroI38:307-317
Callaghan PT (1991) Principles of nuclear magnetic microscopy. Oxford University Press, Oxford,
492 pp
Callaghan PT (1995) Pulsed-gradient spin-echo NMR of planar cylindrical and spherical pores
under conditions of wall relaxation. J Magn Reson A 113:53-59
Carlomagno GM, Berardi PG (1976) Unsteady thermographyy in nondestructive testing. Proc 3rd
Biennal Exchange, St Louis, pp 33-39
Carr HY, Purcell EM (1954) Effects of diffusion on free precession in nuclear magnetic resonance
experiments. Phys Rev 94:630-638
Castagnoli A, Marchetti M, Bottai L, Chirici G, Oradini A (1997) Using a hyperspectral MIVIS
(Multispectral InfraredIVisible imaging Spectrometer) to map forest cover at medium and
large scales. Genio-Rurale 60 (9):54-61
Castera P (1998) How far can we control the heterogeneity of wood and wood based materials
through design? In: Frantziskonis GN (ed) Proc PROBAMAT - 21st century: probability and
materials. Test, models and applications for the 21st century. Kluwer, Dordrecht, pp 65-72
Catalano M, Catena G, Palla L (1986) Infrared technique for some pathological phenomena in
trees. (Impiego dell' infrarosso per l'individuazione di alcuni fenomeni morbosi negli alberi)
Agric Ricerca 59:7-16
Catena G (1989) Thermography for the estimation of trees stability (Impiego della termografia
nel processo di valutazione della stabilita delle piante). Atti de la Fondazione Giorgio Ronchi
LIII:755-766
Catena G (1991) Remote sensing in forestry (Telerilevamento nel settore forestale). Monti Boschi
1:17-24
Catena G (1992) Application of thermography in phytopathology. Diagnosis of sanitary state
of ornamental trees. (Une application de la thermographie en phytopathologie. Diagnostic
de I'eltat sanitaire d'arbrres d'ornement). Phytoma - La delfense des velgeltaux 439:46-
48
Catena G (1993a) Use of the thermal scanner to diagnose cavities in trees. (Utilizzazione di uno
scanner termico per la diagnosi di cavita negli alberi). Riv Ing Agraria 24 2:67-75
Catena G (1993b) La termografia nelle indagini fitosanitarie di insiemi arborei. Economia
Montana. Linea Ecol XXV 3:51-56
Catena G (1997) A tree seems to be very damaged it must be harvested or conserved? The ther-
mography helps for a right decision. (Un albero appare molto dannegiato: 10 si deve abbat-
tere 0 conservare? La tomografia aiuta a prendere to decisione giusta). Economia Montana.
Linea Ecol XXIX 1:9-14
Catena G, Catena A (2000) Termography for the evaluation of cavities and pathological tissues in
trees. (Evidenziazione mediante la termografia di cavita e tessuti degradati negli arberi).Agric
Ricerca 185:47-64
Catena G, Palla L, Catalano M (1990) Thermal infrared detection of cavities in trees. Eur J For
PathoI20:201-210
Catena G, Palla L, Catalano M, Mazzola S (1995) Indagine campione sullo stato del verde urbano
di Roma. Seconda nota. Agric Ricerca 105:11-20
Chang BY (1975) Investigation of the nondestructive determination of wood moisture content by
neutron - capture gamma radiation. Diss Abstr Int B 36 (4):1533
Chang CTP, Watson AT, Edwards CM (1999) NMR imaging of fluids and flow in porous media.
In: Wang P-Z (ed) Methods in the physics of porous media, vol 35. Academic Press, San Diego,
pp 387-423
References 303
Chang SI, Cohen M, Wang PC (1989a) Ultrafast scanning of hardwood logs with a NMR scanner.
Proc 3rd Int Conf On Scanning Tech, Sawmilling
Chang SI, Olson IR, Wang PC (1989b) NMR imaging of internal features in wood. For Prod I 39
6:43-49
Charles lA, Wilson DW (1981) A model of passive thermal nondestructive evaluation of com-
posite laminates. Polym Composites 2 3:105-111
Chazelas IL (1991) Caracteristiques physiques et mecaniques locales du bois dans la zone des
noeuds. These de Doctorat Universite Blaise Pascal, Clermont Ferrand, 156 pp
Chazelas IL, Vergne A, Bucur V (1988) Analyse de la variation des proprietes physiques et
mecaniques locales du bois autour des nreuds. (Wood local properties variation around
knots). Actes du Colloque "Comportement Mecanique du Bois". GS Rheologie du Bois,
Bordeaux, France, pp 344-347
Chelidze T, Gueguen Y, LeRavalec M (1998) From classic fractal mechanics of disordered media:
self-consistency versus self-similarity. In: Frantziskonis GN (ed) Proc PROBAMAT - 21st
century: probability and materials. Test, models and applications for the 21st century. Kluwer,
Dordrecht, pp 197-231
Chiorescu S, Gronlund A (2000) Validation of a CT-based simulator against a sawmill yield. For
Prod I 50 6:69-76
Choffel D (1999) Automation of wood mechanical grading. Coupling of vision and microwave
devices. SPIE 3836:114-121
Choffel D, Martin P (1996) Microwaves and vision device for mechanical grading. Proc 10th Symp
Nondestructive Testing of Wood, Press Polytechniques et Universitaires Romandes,
Lausanne, pp 331-339
Choffel D, Goy B, Gapp D (1992) Interaction between wood and microwaves automatic grading
application. Workshop on scanning technology and image processing on wood. Lulea Uni-
versity of Technology, Lulea, Sweden, pp 1-8
Choffel D, Martin P, Gross P (1995) Nondestructutive testing methods on solid timber:
microwaves, ultrasound and spectrometry. Proc Int Conf Progress in Forest Products
Research, Gottingen, pp 86-89
Choong ET (1963) Movement of water through a softwood in the hygroscopic range. For Prod I
13:489-498
Cielo P, Krapez IC, Lamontagne M (1988) Lumber moisture evaluation by a reflective cavity pho-
tothermal technique. Rev Phys AppI23:1565-1576
Coates ER, Chang SI, Liao TW (1998) A quick defect detection algorithm for magnetic resonance
images of hardwood logs. For Prod J 4810:68-74
Cohen MH, Mendelson KS (1982) Nuclear magnetic resonance and the internal geometry of sed-
imentary rocks. I Appl Phys 53 2:1127-1135
Cole-Hamilton DI, Chudek lA, Hunter G, Martin CJM (1990) NMR imaging of water in wood
including water-logged archaeological artefacts. I Inst Wood Sci 12:111-113
Coles ME (1999) X-ray imaging. In: Wong P-Z (ed) Experimental methods in the physical sci-
ences, vol 35. Academic Press, San Diego, pp 301-336
Comino E, Socco V, Martinis R, Nicolotti G, Sambuelli L (2000) Ultrasonic tomography for wood
decay diagnosis. In: Backhauss GF, Bader H, Idezak E (eds) Int Symp Plant Health in Urban
Horticulture. Mitt. Bundesanst Land-Forstwirtschaft, Braunschweig, p 279
Cormack AM (1963) Representation of a function by its line integrals with some radiological
applications. Part I: I Appl Phys 34:2722-2727; Part II: I Appl Phys 35:2908-2913
Corral CZ, Lindquist CS (1998) On implementing Kasa's circle fit procedure. IEEE Trans Instrum
Meas 47:789-795
Cown DI, Clement BC (1983) A wood densitometer using direct scanning with X-rays. Wood Sci
TechnoI17:91-99
Crawford CR, Kak AC (1982) Multipath artifacts in ultrasonic transmission tomography. Ultra-
sonic Imaging 4:234-266
Davis WC, Moschler Ir (1986) A direct scanning densitometer to measure density profiles in wood
composite products. For Prod J 36, 11/12:82-86
304 References
Davis JR, Weels P, Morgan M, Shadbolt P (1989) A field portable X-ray CT pole scanner and CT
log scanning. Proc of 7th Nondestructive Testing of Wood Symposium, Washington State Uni-
versity. Eng Publications, Pullman, pp 251-262
Davis JR, Lerdin A, Wells P, Ilic J (1991) X-ray microtomography of wood. J Int Wood Sci, 12,4:
259-261
Dean SR (1983) The Radon transform and some of its applications. Wiley, New York, 289 pp
Delsanto PP, Romano A, Scalerani M, Moldoveanu F (1998) Application of genetic algorithms to
ultrasonic tomography. J Acoust Soc Am 104:l374-1381
Dench EC (1973) Advantages of microwave processing. Trans Int Microwave Power Inst 1:l3-16
Denis F, Jossinet J, Gimenez G (1986) Tomographie acoustique: etats, applicatiations et per-
spectives. Rev Acoust 79:21-25
Devaney AJ (1982) A filtered back propagation algorithm for diffraction tomography. Ultrasonic
Imaging 4:336-350
Devaney AJ (1986) Reconstructive tomography with diffractive wave fields. Inv Prob 2:161-
183
Dillenz A, Wu D, Breitriick K, Busse G (2000a) Lock-in thermography for depth
resolved defect characterization. 15th World Congress on NDT, Roma, 6 pp,
http://www.ndt.net/article/wcndtOO/papers/idn3111idn311.htm
Dillenz A, Zweschper T, Busse G (2000b) Elastic wave burts thermography for NDE of subsurface
features. Insight 42 12:815-817
Dines KA, Lytle RJ (1979) Computerized geophysical tomography. Proc IEEE 67 7:1065-1073
Divos F, Szegedi S, Raics P (1994) Local densitometry of wood by gamma back-scattering. Holz
Roh Werkst 54:279-281
Domanus JC (1992) Practical neutron radiography. Kluwer, Boston
Douglas Mast T (1999) Wideband quantitative ultrasonic imaging by time-domain diffraction
tomography. J Acoust Soc Am 1066:3061-3071
Dreele von RB (1994) Neutron diffraction. In: Cahn RW, Hassen P, Kramer EJ (eds) Material sci-
ences and technology. A comprehensive treatment, vol 2B. VCH, Weinheim, pp 562-609
Drobakhin 0 (2000) Homographic approach to multifrequency microwave methods of dielectric
material NDE. 15th World Conf on Nondestructive Testing, Roma, 5 pp
Dubey YM, Deorani SC (1997) Dielectric properties of coniferous timbers at microwave fre-
quencies. J Indian Acad Wood Sci 1:77-82
Dumoulin JP, de Belleval JF (1987) Flaw classification in composites using ultrasonic echography.
In: Alippi A, Mayer W (eds) Ultrasonic methods in evaluation of inhomogeneous materials.
Martinus Nijhoff, Dordrecht, pp 207-217
Elbez G (1990) Medical X-ray scanner applied to wood gluing. Centre Technique du Bois et de
l' Ameublement, Paris, 5 pp
EI-Rayes MA, Ulaby FT (1987) Microwave dielectric spectrum of vegetation. Part I: experimen-
tal observations. IEEE Trans Geosci Remote Sensing GE 25 5:541-549
Falls S, Chow T, Yong R, Hutchins D (1989) Acoustic emission analysis and ultrasonic velocity
imaging in the study of rock failure. J Acoust Emission 8:S166-S169
Faust TD, Tang M, Bhandarkar S, Smith JW, Toliner EW (1996) Evaluation of growth related
features in selected hardwoods logs using X-ray computed tomography and image analysis.
Proc lOth Symp Nondestructive Testing of Wood, Presses Poly techniques et Universitares
Romandes, Lausanne, pp 201-208
Fengel D (1991) Aging and fossilization of wood and its components. Wood Sci Technol 25:
153-177
Feraz ES de B (1976) Wood density determination by attenuation of low-energy gamma radia-
tion. IPEF Piracicaba Bresil 12:61-68
Filippini B, Lambert P, Martin P, Mallick G, Leban JM (1990) Research on wood properties and
defects with microwave probe (Investigation des proprielreIs du bois et des singularitels par
capteur micro-ondes). Actes du 3-e2me Colloque. Sciences et Industries du Bois. ARBORA,
Bordeaux Tome I, pp 1l3-121
Fink M (1983) Ultrasonic imaging (L'imagerie ultrasonore). Rev Phys AppI18:527-556
References 305
Fioravanti M, Ricci R (1991) L'impiego della tomografia computerizzata per misure densitomet-
riche sullegno: indagine sperimentale e risultati metodologici. (Use of computed tomogra-
phy for densitometric measures on wood: experimental researches and methodological
results). Ann Acad Ital Sci For XL:I-35
Flibotte S, Menon RS, MacKay AL, Hailey JRT (1990) Proton magnetic resonance of western red
cedar. Wood Fiber Sci 22:362-376
Forrer JB, Funck JW (1998) Dielectric properties of defects on wood surfaces. Holz Roh Werkst
56:25-29
Fortunko C, Renken M, Murray A (1991) Examination of objects made of wood. Proc IEEE Ultra-
sonic Symp, pp 1099-1103
Frantziskonis GN, Blodgett MP (1998) Multiscale material characterization and applications.
In: Frantziskonis GN (ed) Proc PROBAMAT - 21st century: probability and materials. Test,
models and applications for the 21st century. Kluwer, Dordrecht, pp 367-378
Frick J, Sawade G, Schuh H (1997) Defect localization in pillars of marble by ultrasonic tomog-
raphy. Otto GrafJ 8:174-188
Fridel MJ, Tweeton DR, Jackson Jessop JA, Billington S (1992) Mining application of seismic
tomography. 62nd Annu Int Meet Soc Exploration Geophys BG 1 7, pp 58-62
Fukada E (1965) Piezoelectric effect in wood and other crystalline polymers. 2nd Symp NDT of
wood, Washington State University, Pullmann, pp 143-170
Fukada E, Yasuda S, Kohara J, Okamoto H (1956) Dynamic Young's modulus and piezoelectric
constants of old timber. Bull Kabayasi Inst Phys Res 6:104-107
Funt BV, Bryant EC (1985) A computer vision system that analyses CT-scans of sawlogs. Proc of
IEEE Computer Soc Conf on Computer vision and pattern recognition, San Francisco, CA,
9-13 June,pp 175-177
Funt BV, Bryant EC (1987) Detection of internal log defects by automatic interpretation of com-
puter tomography images. For Prod J 37 1:56-62
Furukawa J, Nakanishi TM, Matsubayashi M (2001) Water uptake activity in soybean root revealed
by neutron beam imaging. Nondestr Test EvaI16:335-334
Gadian DG (1982) Nuclear magnetic resonance and its applications to living systems. Clarendon
Press, Oxford, 197 pp
Garrett DA (1977) The technological development of neutron radiography. Atomic Energy Rev
15:125-142
Garrett DA, Berger H (1977) The technological development of neutron radiography. Atomic
Energy Rev 15,2:125-142
Gierlik E, Dzbenski W (1996) Wood densitometry based on the radiometric methods. Proc 10th
Symp Nondestructive Testing of Wood, Presses Poly techniques et Universitaires Romandes,
Gil AM, Neto CP (1999) Solid state NMR studies of wood and other lignocellulosic materials.
Annu Rep NMR Spectrosc 37:75-117
Gilardi G, Abis L, Cass AEG (1995) Carbon 13 CP/MAS solid-state NMR and FT-IR spectroscopy
of wood cell wall biodegradation. Enzyme Microbial TechnoI17:268-275
Gilbert P (1972) Iterative method for the tridimensional reconstruction of an object from pro-
jections. J Theor BioI 36: 105-117
Gilboy WB (1984) X- and y-ray tomography in NDE applications. Nuclear Instrum Meth Phys Res
221:193-200
Gilboy WB, Foster J (1982) Industrial applications of computerized tomography with X- and
gamma-radiation. In: Sharpe RS (ed) Research techniques in nondestructive testing, vol 6.
Academic Press, London, pp 255-287
Giordano G (1971) Tecnologia dellegno. Vol I. Materia prima. Unione Tipografico - Editrice
Torinese, Milano, 1086 pp
Glos P (1982) Machine grading of sawn timber: current technology and comparison of methods.
Holz ZentralbI108:153-155
Golosovsky M, Lann AF, Davidov D, Frenkel A (2000) Microwave near-field imaging of conduct-
ing objects of a simple geometric shape. Rev Sci Instrum 71 10:3927-3932
306 References
Goy B, Martin P, Leban JM (1992) The measurements of wood density by microwave sensor. Holz
Granier A, Anfodillo T, Sabatti M, Cochard H, Dreyer E, Tomasi M, Valentini R, Breda N (1994)
Axial and radial water flows in trunks of oak trees: a qualitative and quantitative analysis.
Tree Physiol 14: 1383-1396
Greaves H (1998) Trends in wood research and utilization - a view from the institute of wood
science, Australian branch. J Inst Wood Sci 146:293-300
Green RE Jr (1987) Ultrasonic nondestructive materials characterization. In: Vary A (ed)
Materials analysis by ultrasonics. Noyes Data Corp, Park Ridge, USA, pp 1-29
Greenleaf JF (1981) Computerized transmission tomography. In: Marton L, Marton C (eds)
Methods of experimental physics, vol 19. Ultrasonics. Academic Press, New York, pp 563-589
Grimberg R, Savin A, Lupu A, Rotundu C, Iancu L (2000) A method to determine the debonding
zones in multilayer wood materials. 15th World Conference on Nondestructive Testing, Roma,
2000,6 pp
Gronlund A, Grundberg S, Gronlund U (1994) The Swedish stem bank - an unique database for
different silvicultural and wood properties. IUFRO S5.01-04 Workshop Proc, Hook, Sweden,
pp 71-77
Grundberg S, Gronlund A (1997) Simulated grading of logs with an X-ray log scanner. Grading
accuracy compared with manual grading. Scand J For Res 12:70-76
Grundberg S, Gronlung A, Gronlund U (1995) The Swedish Stem Bank - a data base for different
silvicultural and wood properties. Res Rep TULEA [BR3] 1995:31. Lulea Unic Technol,
Skelleftea, Sweeden, 14 pp
Grundberg S, Lindgren 0, Gronlung A (1996) The development of an X-ray based scanner for
log grading at full production speed. Proc 10th Symp Nondestructive Testing of Wood Swiss
Federal Institute of Technology, Press Poly techniques et Universitaires Romandes, Lausanne,
p 407
Guinn VP, Lukens HR (1965) Nuclear methods. In: Morrison GH (ed) Trace analysis: physical
methods. Interscience, New York
Habermehl A( 1982a) A new nondestructive method for determining internal wood condition and
decay in living trees. I. Principles, method and apparatus. Arboric J 6:1-8
Habermehl A (1982b) A new nondestructive method for determining internal wood condition
and decay in living trees. II. Results and further developments. Arboric J 6:121-130
Habermehl A, Ridder HW (1992a) Metodik der Computer - Tomographie zur zerstOrungsfreien
Untersuchung des Holzkorpers von stehenden Biiumen. (Computerized tomography applied
to studies on wood of living trees). Holz Roh Werkst 50:465-474
Habermehl A, Ridder HW {1992b) Computer Tomographie am Baum (Computed tomograhy for
tree). Materialpriifung 34:325-329; 357-360
Habermehl A, Ridder HW (1993) Anwendungen der mobilen Computer-Tomographie zur zer-
stOrungsfreien Untersuchung des Holzkorpers von stehenden Biiumen (Application of mobile
computed tomography for nondestructive investigations on wood of living trees) Holz Roh
Werkst 51:1-6
Habermehl A, Ridder HW (1994) Applications of computerized tomography in forest and tree
sciences. Int Symp on Computerized Tomography for Industrial Applications, 8-10 June,
Berlin, pp 71-81
Habermehl A, Ridder HW (1995) Computerised tomographic investigations of street and park
trees. Arboric J 19:419-437
Habermehl A, Ridder HW (1996) Computer Tomographie in der Forstwirtschaft und Baumpflege.
(Teil1 und 2) DGZfP/DACH - Zeitung no 55:48-55 and no 56:47-55
Habermehl A, Ridder HW (1998) y-ray tomography in forest and tree sciences. Proc SPIE (lnt Soc
Optical Eng) 3149:234-244
Haegglund G, Lindgren 0 (1985) Maetning av densitets-och fuktvariationer i traematerial med
hjaelp av datortomografi. (Measuring of density and moisture variations in wood with the
help of computed tomography). Traeteknikrapport no 67. Stockholm Traeteknik Centrum,
Stockholm, 21 pp
References 307
Hagman POG, Grundberg SA (1995) Classification of scots pine (Pinus sylvestris) knots in density
images from CT scanned logs. Holz Roh Werkst 53:75-81
Hailey JRT, Menon RS, Mackay A, Burgess AE, Swanson JS (1985) Nuclear resonance scanning for
log characterization. 5th Symp Nondestructive Testing of Wood, Washington State University,
Pullman
Halabe UB, Ganga Rao H, Rao Hota V (1995) Nondestructive evaluation of wood using frequency
domain analysis. In: Thompson DO, Chimenti DE (eds) Review of progress in quantitative
nondestructive evaluation, vol 14. Plenum Press, New York, pp 1653-1660
Hall LD, Rajanayagam V (1986) Evaluation of the distribution of water in wood by use of three
dimensional proton NMR volume imaging. Wood Sci Technol 20:329-333
Hall LD, Rajanayagam V, Stewart WA, Steiner PR (1986a) Magnetic resonance imaging of wood.
Can J For Res 16:423-426
Hall LD, Rajanayagam V, Stewart WA, Steiner PR, Chow S (1986b) Detection of hidden morphol-
ogy of wood by magnetic resonance imaging. Can J For Res 16:684-687
Halloin JM, Cooper TG, Potchen EJ, Vozzo JA (1994) Magnetic resonance imaging, a technology
for noninvasive plant analysis. In: Research and applications of chemical sciences in forestry.
USDA, GTR SO-104, New Orleans, Louisiana, 9 pp
Harley J, Morris PI (1988) Application of scanning and imaging techniques to assess decay
and wood quality in logs and standing trees. Forintek Canada Corp no FO 42-91-42-(1988)E,
10 pp
Harpole GB, McDonald KA (1981) Investment opportunity: a scanning - ultrasonics cut stock
manufacturing system. USDA, FPL, Research Paper 390, 7 pp
Hatcher PG, Berger lA, Earl WL (1981) NMR studies of ancient buried wood. I. Observations on
the origin of coal to the brown coal stage. Org Geochem 3:49
Hattori Y, Kanagawa Y (1985) Nondestructive measurement of moisture distribution in wood with
a medical X-ray CT scanner. Part I. Mokuzai Gakkaishi 31:974-982
Hauffe P, Mahler G (2000) Evaluation of internal log quality using X-ray and ultrasound. 12th
Symp Nondestructive Testing of Wood, Sopron, pp 259-263
Hearmon RFS (1948) The elasticity of wood and plywood. Dept Sci Ind Res For Prod Res Spec
Report no 7. HMSO, London, 87 p
Hearmon RFS (1965) The assessment of wood properties by vibrations and high frequency
acoustic waves. 2nd Symp NDT, Washington State University, pp 49-66
Hearmon RFS, Burcham IN (1954) The dielectric properties of wood. For Prod Res Special Report
no 8, Dept of Sci and Industry, London, 25 p
Henneke EG, Jones TS (1979) Detection of damage in composite materials by vibrothermogra-
phy. ASTM Spec Tech Publ 696:83-95
Hennecke EG, Reifsnider KL, Stinchcomb WW (1979) Thermography - an NDE method for
damage detection. J Metals Sept:11-15
Herman GT (1980) Image reconstruction from projections. The fundamentals of computerized
tomography. Academic Press, London, 316 pp
Herman GT, Lent A, Rowland S (1973) Mathematics and applications: a report on the mathemat-
ical foundations and on applicability to real data of the algebraic reconstruction technique.
J Theor BioI 43:1-32
Hippel von A (1961) Dielectric materials and applications. Wiley, New York, 302 pp
Hoag M, McKimmy MD (1988) Direct scanning X-ray densitometry of thin wood sections. For
Prod J 38 1:23-26
Hodges DG, Anderson WC, McMillin CW (1990) The economic potential of CT scanners for hard-
wood sawmills. For Prod J 40 3:65-69
Horig H (1935) Theory of elasticity of anisotropic solids applied to wood (Anwendung der
Elastizitatstheorie anisotroper Korper auf Messungen an Holz). Ing Arch VI:8-14
Hounsfield GN (1972) A method and apparatus for examination of a body by radiation such as
X-ray or gamma radiation. Patent Specification 1283915, The Patent Office, London
Hoyle RJ (1961) A nondestructive test for stiffness of structural lumber. For Prod J 11 6:
251-254
308 References
Hsi E, Hossfeld R, Bryant RG (1977) NMR relaxation study of water absorbed on milled North
white cedar. J Colloid Interface Sci 62:389-395
Hutchins DA, Jansen DP, Edwards P (1993) Lamb wave tomography using noncontact transduc-
ers. Ultrasonics 31 :97 -103
Hutchins MT, Windsor CG (1987) Industrial applications. In: Sk6ld K, Price DL (eds) Neutron
scattering, vol 23, part C. Academic Press, Orlando, pp 405-407
Iancu L, Iancu G, Grimberg R, Lupu A (2000) Quantification of defects in wood by use of
ultrasonics in association with imagistic methods. 15th World Conf On NDT, Roma,
http://www.ndt.net/article/wcndtOO/paper
Ida N (1992) Microwave nondestructive testing. Kluwer, Amsterdam, 300 pp
Ida N, Wang JS (1996) Models for microwave nondestructive testing of materials. Mater Sci Forum
210-213:93-100
Jacquiot C, Trenard Y, Dirol D (1973) Atlas d'anatomie des bois des angiospermes. Tome 1. Centre
Technique du Bois, Paris, 150 pp
Jain VK, Sanyal SN (1996) A preliminary note on the attenuation of microwaves by few wood
species. Indian For 122,5:428-431
James WL (1959) A method for rapid measurement of the rate of decay of free vibrations. USDA,
FPL Bull no 2154, USDA, Madison, WI
James WL (1975) Dielectric properties of wood and hardboard: Variation with frequency, mois-
ture content and grain orientation USDA For Serv Res Pap FPL-245, Madison, WI
James WL (1989) Electrical properties. In: Schniewind AP (ed) Concise encyclopedia of wood and
wood-based materials. Pergamon Press, Oxford, pp 99-102
James WL, Hamill DW (1965) Dielectric properties of Douglas fir measured at microwave fre-
quencies. For Prod J 152:51-56
James WL, Yen YS, King, RJ (1985) A microwave method for measuring moisture content, density
and grain angle of wood. USDA For Serv Res Notes FPL-250, Madison, WI, 9 pp
Jartti P, Luukkala M (1977) Ultrasonic method for web speed measurements. Tappi J 60 11:167
Javadpour Z, Hughes D, Keating JG, Feeney FE, Evertsen JA (1996) Assessment of timber quality
using neural networks on CT images. Proc 10th Symp Nondestructive Testing of Wood Swiss
Federal Institute of Technology, Press Polytechniques et Universitaires Romandes, Lausanne,
pp 303-311
Jayne BA (1955) A nondestructive test of glue bond quality. For Prod J 5 5:294-301
Jayne BA (1959) Vibrational properties of wood as indices of quality. For Prod J 911:413-416
Jayne BA (1972) Theory and design of wood and fiber composite materials. Syracuse University
Press, Syracuse, NY, 418 pp
Johansson LG (1985) Experiences from using X-ray tomography, isotope-based nontomographic
measuring and NMR for testing logs before sawing 5th Symp Non destructive testing of Wood,
Washington State University, Pullman, pp 205-229
Kabir MF, Daud WM, Khalid K, Sidek HHA (1998) Dielectric and ultrasonic properties of rubber
wood. Effect of moisture content, grain direction and frequency. Holz Roh Werkst 56:223-227
Kabir MF, Schmoldt DL, Schafer ME (2000a) Ultrasonic detection of knots, cross grain and bark
pockets in wooden pallet parts. Proc World Conf Timber Eng 2000, Whistler, British Colum-
bia, Canada, pp 7521-7528
Kabir MF, Schmoldt DL, Schafer ME (2000b) Detection of defects in red oak deckboards by ultra-
sonic scanning. Proc 4th Int Conf Image Proccesing and Scanning of Wood, Mountain Lake
Resort, Virginia, pp 89-96
Kabir MF, Schmoldt DL, Schafer ME (2001) Roller transducer scanning of wooden pallet parts
for defect detection. In: Thompson DO, Chimenti DE (eds) Review of progress in quantitative
nondestructive evaluation, vol 20. Plenum Press, New York, pp 1218-1225
Kaestner AP, Baath L (2000) Microwave polarimetry based wood scanning. 12th Symposium on
Nondestructive Testing of Wood, Sopron, pp 349-356
Kiitzel R, Ridder HW, Habermehl A (1997) Investigations of ammonia stressed Scots pine trees
(Pinus sylvestris) with computed tomography in an immission area around a cattle farm.
Phyton 371:141-149
References 309
Kak AC (1979) Computerized tomography with X-ray emission and ultrasound sources. Proc
IEEE 67:1245-1272
Kak, AC Slaney M (1988) Principles of computerized tomographic imaging. IEEE Press, New York,
329 pp
Kanagawa Y, Hattori Y (1985) Nondestructive measurement of moisture distribution in wood with
a medical X-ray CT scanner. Part II. Changes in moisture distribution with drying. Mokuzai
Gakkaishi 31:983-989
Kanagawa Y, Furuyama Y, Hattori Y (1992) Nondestructive measurement of moisture diffusion
coefficient in wood drying. Drying Technoll0:1231-1248
Karsulovic JT, Leon LA, Dinator Mi (1999) The use of linear attenuation coefficients of gamma
radiation for detecting knots in Pinus radiata. For Prod J 49 2:73-76
Kasischke ES, Melack JM (1997) The use of imaging radars for ecological applications. Remote
Sensing Environ 59 2:141-156
Keller E, Thiercelin F (1975) The influence of big medulary rays on some properties of beech.
(Influence de gros rayons ligneux sur quelques proprieltels du bois de he3tre). Ann Sci For
322:113-129
Kessler R (1987) Applicability of imaging radar for classification of forest vegetation. Photo-
grammetria 414:221-232
Khan MA, Rizvi TZ, Naheed A, Zaheer MY, Iqbal MZ, Shoaib S (2000) Dielectric observed con-
sequences of microbial treatment. Holzforschung 54:335-339
Kharadly M (1985) Microwave diagnostic for stress rating of dimension lumber. Proc 5th Non-
destructive Testing of Wood, Washington State University, Pullman, pp 445-464
Kim YS, Singh AP (2000) Michromorphological characteristics of wood biodegradation in wet
environments: a review. IAWA J 21:135-155
King RJ (1978) Microwave electromagnetic nondestructive testing of wood. Proc 4th Nonde-
structive Testing of Wood, Washington State University, Pullman, pp 121-134
King RJ (2000) On-line industrial applications of microwave moisture sensor. In: Baltes H, Gopel
W, Hesse J (eds) Sensor update, vol 7. VCH Verlag, Weinheim, pp 109-170
King RJ, Basuel JC (1993) Measurement of basis weight and moisture content of composite boards
using microwaves. For Prod J 439:15-22
King RY, Yen YH (1981) Probing amplitude, phase and polarization of microwave field distribu-
tions in real time. IEEE Trans Microwave Theor Tech 29:1225-1231
Kinney JH, Johnson QC, Bonse U, Nichols MC, Saroyan RA, Nusshardt R, Pahl R, Brase JM (1988)
Three dimensional X-ray computed tomography in material science. MRS Bull 31 1:13-
17
Kittel C (1968) Introduction to solid state physics. Wiley, New York, 399 pp
Klein P, Vogel H (1993) Computertomography - a help for dendrochronology and wood identi-
fication? Proc 4th Int Conf on Nondestructive Testing of Works of Art, Berlin, pp 85-91
Kleinberg RL (1999) Nuclear magnetic resonance. In: Wang P-Z (ed) Methods in the physics of
porous media, vol 35. Academic Press, San Diego, pp 337-385
Kliger IR, Perstorper M, Johansson G, Pellicane PJ (1995) Quality of timber products from Norway
spruce. Part 3. Influence of spatial position and growth characteristics on bending stiffness
and strength. Wood Sci TechnoI29:397-410
Kline R, Sullivan C, Mignona RB, Delsanto PP (1996) A parallel processing algorithm for acoustic
tomography. In: Tortoli P, Masotti L (eds) Acoustical imaging. Plenum Press, New York, pp
561-566
Kline RA, Wang YQ (1992) A technique for ultrasonic tomography in anisotropic media. J Acoust
Soc Am 91:878-884
Kline RA, Wang YA, Mignogna RB, Delsanto PP (1994) A finite difference approach to acoustic
tomography in anisotropic media. J Nondestr Eval13:75-83
Koch M, Hunsche S, Schuacher P, Nuss MC, Feldmann J, Fromm J (1998) THz-imaging: a new
method for density mapping of wood. Wood Sci TechnoI32:421-427
Kollmann F (1951) Technology of wood and wood based composites (Technologie des Holzes
und der Holzwerkstoffe). Springer, Berlin Heidelberg New York, 1048 pp
310 References
Kollmann FFP, Cote WA (1968) Principles of wood science and technology. Springer, Berlin
Heidelberg New York, 592 pp
Kollmann FFP, Krech M (1960) Dynamic measurements of wood elastic properties and damping.
(Dynamische Messung der elastischen Holzeigenschaften und der Dampfung). Holz Roh
Werkst 18 2:41-45
Kostorz G, Lovesey SW (1979) Neutron scattering. General introduction. Treatise Mater Sci
TechnoI15:1-67
Kucera LJ (1990) Current use of the NMR on wood at the Swiss Federal Institute of Technology:
overview and outlook. 7th Int Symp on Nondestructive Testing of Wood, Washington State
University, Pullman, pp 71-72
Kuo PK, Feng Z1, Ahmed T, Favro LD, Thomas RL, Hartikainen J (1987) Parallel thermal wave
imaging using a vector lock-in video technique. In: Hess P, Pelzl J (Eds) Photo acoustic and
photothermal phenomena. Springer, Berlin Heidelberg New York, pp 415-418
Kuroda N, Suzuki Y (1996) Effects of morphological factors on the electrical behavior of wood
with the presence of free water. Current behavior from voltage reversals and tracheary ele-
ments. Mokuzai Gakkaishi 42:817-824
Lambert JB, Frye JS, Carriveau GW (1991) The structure of oriental lacquer by solid state NMR
spectroscopy. Archaeometry 33 87:93
Lambert JB, Shawl CE, Stearns JA (2000) Nuclear magnetic resonance in archeology. Chern Soc
Rev 29 3:175-182
Lanceleur P, de Belleval JM, Mercier N (1998) Synthetic tridimensional representation of
anisotropic materials. Acta Acust 84:1047-1054
Langwig JE, Meyer JA (1973) Ion migration in wood verified by neutron activation analysis. Wood
Sci 6 1:39-50
Lau SK, Almond DD, Milne JM (1991) A quantitative analysis of pulsed video thermography. NDT
Eval Int 24 4:195-202
Laufengerg TL (1986) Using gamma radiation to measure density gradient in reconstructed wood
products. For Prod J 36 (2):59-62
Laurence GR, Banner A (1980) The application of thermography for locating potential frost
pockets in forest cutovers. In: Proc of the 6th Canadian symposium on remote sensing,
Halifax, Nova Scotia. Canadian Aeronautics and Space Institute, Halifax, Canada, pp 369-374
Lebowitz C, Zoughi R, Spicer JWM, Osinder R (1996) Comparison of ultrasonic, microwave and
photothermal imaging of defective graphite-epoxy composite panels. Proc SPIE 2944:67-74
Lehmann E, Vontobel P, Niemz P (2001) Application of neutron radiography as method in the
analysis of wood. (Anwendung der Methods der Neutronenradiographie zur Analyse von
Holzeigenschaften). Holz Roh Werkst 59:463-471
Lehto A, Jaarinen J, Tiusanen M, Jokinen M, Luukkala M (1981) Amplitude and phase in thermal
wave imaging. Electr Lett 17:364-365
Lewa CJ (1991) Magnetic resonance imaging in the presence of mechanical waves. Spectrosc Lett
241:56-67
Lewa CJ (1992) NMR detection of elastic wave effects. In: Alippi A (ed) Acoustic sensing and
probing. World Science, Singapore, pp 435-438
Lewa CJ (1996) Elasto-magneitc resonance spectroscopy. Europhys Lett 35 1:73-76
Liezers M, Miller RM, Spillane DEM, Ryan TH (1985) Developments in thermal wave imaging.
Proc Int Conf Acoustic Emission and Photo acoustic spectroscopy and applications, London,
pp 65-69
Lin RT (1967) Review of the dielectric properties of wood and cellulose. For Prod J 177:61-66
Lin RT (1973) Wood as an orthotropic dielectric material. Wood Fiber 5 3:226-236
Lindberg H, Grahn 0, Lindgren A, Heligren C (1996) Watter diffusion through acrylate latex paint
films measured by computed tomography. Proc of 10th Nondestructive Testing of Wood Sym-
posium. Swiss Federal Institute of Technology, Press Poly techniques et Universitaires Roman-
des, Lausanne, 408 p
Lindegaard-Andersen A, Notea A, Bushlin Y, Rheinlander J (1990) Image-centred frame of refer-
ence in rotational film-based X-ray tomography. NDT Int 23 6:332-334
References 311
Lindgren LO (1985) On the relationship between density / moisture content in wood and X-ray
attenuation in computer tomography. Proc 5th Symp Nondestructive Testing of Wood;
Washington State University, Eng Publications, Pullman, pp193-203
Lindgren 0 (1991a) Medical CAT-scanning: X-ray absorption coefficients, CT-numbers and their
relation to wood density. Wood Sci TechnoI25:341-349
Lindgren 0 (1991b) The accuracy of medical CT-scan images for nondestructive density mea-
surements in small volume elements within solid wood. Wood Sci Technol 25:425-432
Lindgren 0 (1994) NMR for nondestructive wood moisture content measurement. First European
Symp on NDE of Wood, University of Sopron, Sopron, pp 124-129
Lindgren 0, Davis J, Wellq P, Shadbolt P (1992) Nondestructive wood density distribution mea-
surements using computed tomography. Holz Roh Werkst 50:295-299
Liu CJ, Olson JR, Tian Y, Shen Q (1988) Theoretical wood densitometry: I. Mass attenuation equa-
tions and wood density models. Wood Fiber Sci 20:22-33
Loos WE (1965) Determining moisture content and density of wood by nuclear radiation tech-
niques. For Prod J 153:102-106
Loos WE (1965) Neutron activation of Engelmann spruce. For Prod J 154:178
Luong PM (1995) Infrared scanning of failure process in wood. SPIE 2473:298-309
Luong PM (1996) Infrared thermography of damage in wood. 10th Symp NDT of Wood, Press
Univ Romandes, Lausanne, pp 175-185
Luong PM (1998) Fatigue limit evaluation of metals using an infrared thermographic technique.
Mech Mater 28:155-163
Luukkala M, Heikilla P, Surakka Y (1971) Plate-wave resonance. A contactless test method. Ultra-
sonics 9:201-208
Lyamshev ML, Stanullo 1, Busse G (1995) Thermo acoustic vibrometry. Remote in situ monitor-
ing of ceramic sintering. Materialpriifung 37 112:22-24
Lynnworth LC (1989) Ultrasonic measurements for process control. Theory, techniques, applica-
tions. Academic Press, Boston, 694 pp
Lynnworth LC, Magory V (1999) Industrial process control sensors and systems. In: Papadakis
EP (ed) Ultrasonic instruments and devices. Academic Press, San Diego, pp 276-470
Lynnworth LC, Nguyen TH, Smart CS, Khrakovsky OK (1997) Acoustically isolated paired air
transducers for 50, 100,200 or 500kHz applications. IEEE Trans UFFC 44 5:1087-1100
Lytle RJ, Dines K (1980) Iterative ray tracing between boreholes for underground image recon-
struction IEEE Trans Geosci Remote Sens GE 18:234-240
MacFali JS, Johnson GA, Kremer PJ (1990) Observation of water-depletion region surround-
ing loblolly pine roots by magnetic resonance imaging. Proc Natl Acad Sci USA 87:1203-
1207
MacGregor RP, Peemoeller H, Schneider MH, Sharp AR (1983) Anisotropic diffusion of water in
wood. J Appl Polymer Sci Polymer Symp 37:901-909
Madsen B (1994) Radiological density scanning - a portable gamma camera based on back-
scatter tomography. Proc 9th Symp Nondestructive Testing of Wood, Washington State
University, Madison, pp 131-137
Makoviny I (1988) The anisotropy of wood dielectric constants as a function of anatomical
structure. (Zur Anisotropie der Dielektrizitatskonstanten des Holzes in grundlegenden
anatomischen Richtungen). Holztechnologie 29 4:210-213
Malan FS, Marais PG (1992) Some notes on the direct gamma ray densitometry of wood. Holz-
forschung 46:91-97
Maloney TM (1992) Composites for the future. Invited paper. Proc IUFRO Conference, Division
5, Nancy, 605p p
Maloney TM (1996) The family of wood composite materials. For Prod J 46 2:19-26
Mandelis A (2000) Diffusion waves and their uses. Physics Today 53 No 8 Part 1:29-34
Mansfield P, Morris PG (1982) NMR imaging in bio-medicine. In: Waugh J (ed) Advanced mag-
netic. Resonance Suppl no 2. Academic Press, New York
Marcinko JJ, Devathala S, Rinaldi PL, Bao S (1998) Investigating the molecular and bulk dynam-
ics of PMDIIwood and UF/wood composites. For Prod J 48 6:81-84
312 References
Marcinko JJ, Rinaldi PL, Bao S (1999) Exploring the physicochemical nature of PMDl/wood struc-
tural composite adhesion. For Prod J 49 5:75-78
Marok M, Kundela J, Cunderlik I (1996) Identification of reaction beech wood by X-ray computed
tomography. Holz Roh Werkst 54:97-98
Martin P, Collet R, Barthelemy P, Roussy G (1987) Evaluation of wood characteristics: internal
scanning of the material by microwave. Wood Sci TechnoI21:361-371
Martinis R (2002) Nondestructive techniques for decay diagnosis on standing trees. (Analisi di
tecniche non destruttive per la diagnosi di carie su alberi in piedi). Thesis University of
Florence, Florence, Italy, 300 pp
Masuch G, Franz JT, Marsmann H, Gross D, Kettrup A (1991) 'H NMR micro-imaging and
correlated SEM studies of spruce needles from healthy and declined forest sites. Int J Environ
Anal Chern 45:179-191
Masuda M, Takahashi S (1998) Change of thermal images of lumber including knots, finger joints
and metal connectors. Bull Kyoto Univ For (Kyoto Daigaku Nogaku Bu Enshurin Hokoku)
69:114-128
Masuda M, Takahashi S (1999) Thermographical detection of knots and other defects. Proc
INCEUPT '99 For Prod Assoc ROC Bull 16:181-187
Masuda M, Takahashi S (2000) Thermographical detection of defects of lumber and glued
joints. 12th Symp NDT of Wood, University of Western Hungary, Sopron, pp 421-
430
Masuda M, Fujimoto K, Takino S, Sadoh T (1995) Change of thermal images of timber including
knots under repeated bending. Bull Kyoto Univ For 67:167-173
McDonald KA (1978) Lumber defect detection by ultrasonics. Forest Prod Lab Res Paper FPL 311,
USDA, Madison, WI, 25 pp
McGaughey WJ, Young RP (1990) Comparation of ART, SIRT, Least-Squares, and SVD two-
dimensional tomographic inversions of field data. 60th Annual Meeting of Soc Exploration
Geophysicists SEG, pp 74-77
McKeon JCP, Hinders MK (1999) Parallel projection and crosshole Lamb contact scanning tomog-
raphy. JASA 106 5:2568-2577
McLauchlan TA, Norton JA, Kusec DJ (1973) Slope of grain indicator. For Prod J 23 5:50-55
McMillin CW (1982) Application of automatic image analysis to wood science. Wood Sci 14:
97-105
Meiboom S, Gill D (1958) Modified spin-echo method for measuring nuclear relaxation times.
Rev Sci Instrum 29:688-691
Menon RS, MacKay AL, Flibotte S, Hailey JRT, Blook M, Burgess AC, Swanson JS (1987) A NMR
determination of the physiological water distribution in wood during drying. J Appl Polymer
Sci 33:1146-1155
Menon RS, MacKay AL, Flibotte S, Hailey JRT (1989) Quantitative separation of NMR images of
water in wood on the basis of T(2). J Magn Reson 82 205-210
Millard P, Chudek JA (1993) Imaging the vascular continuity of Prunus avium petioles during leaf
senescence using NMR spectroscopy. J Exp Bot 44:599-603
Miller WH (1988) Design and implementation of a portable computerized axial tomography
system for field. Nucl Instrum Meth Phys Res 270:590-597
Moksin MM (1993) Improvement of an opto-thermal technique for wood. Wood Sci Technol28:
53-58
Moschler WW Jr, Dougal EF (1988) Calibration procedure for a direct scanning densitometer
using gamma radiation. Wood Fiber Sci 20:297-303
Moschler WW Jr, Winistorfer PM (1990) Direct scanning densitometry: an effect of sampling het-
erogeneity and aperture area. Wood Fiber Sci 22:31-38
Mothe F, Duchanois G, Zannier B, Leban JM (1998) Analyse microdensitometrique appliquee au
bois. Ann Sci For 55:301-313
Murata K, Sadoh T (1994) Heat absorption and transfer in softwoods and their knot surfaces.
Mokuzai Gakkaishi 40:1180-1184
Musial MW (1988) A nondestructive method for determining the degree of flake orientation in
OSB. Wood Sci TechnoI22:371-378
References 313
Naito S, Sawada Y, Fujii Y, Okumura S (1998) Thermography of wood specimens in static bending
test. Kyoto Daigaku Nogaku Bu Enshurin Hokoku 69:104-113
Naito S, Fujii Y, Sawada Y, Okumura S (2000) Thermographic measurement of slope of grain using
the thermal anisotropy of wood. Mokuzai Gakkaishi 46:320-325
Nakanishi TM (1994) The water image of plants using neutron radiography. Chern Regul Plants
292:182-190
Nakanishi TM, Matsubayashi M (1997) Nondestructive water imaging by neutron beam analysis
in living plants. J Plant PhysioI151:442-445
Nakanishi TM, Watanabe S (1995) Water distribution inside the woods by neutron radiography.
Bull Tokyo Univ For 93:13-19
Nakanishi TM, Karakama I, Sakura T, Matsubayashi M (1998a) Moisture imaging of a camphor
tree by neutron beam. Radioisotopes 47:387-391
Nakanishi TM, Okano T, Karakama I, Ishihara T, Matsubayashi M (1998b) Three dimensional
imaging of moisture in wood disk by neutron beam during drying process. Holzforschung
52:673-676
Nanassy AJ (1970) Overlapping of dielectric relaxation spectra in oven -dry birch at temperatures
from 20 to 100 degrees C. Wood Sci TechnoI4:104-121
Nanassy AJ (1972) Dielectric measurements of moist wood in a sealed system. Wood Sci Technol
6:67-77
Nanassy AJ (1973) Use of wide line NMR for measurement of moisture content in wood. Wood
Sci 5:187-193
Nanassy AJ (1974) Water sorption in green and remoistured wood studied by the broad-line com-
ponent of the wide line NMR spectra. Wood Sci 7:61-68
Natterer J, Sandoz JL, Rey M (2000) Bois matelriau et struture bois. Press Polytheqnique et
Universitaire Romande, Lausanne, 488 pp
Neuenschwander J, Niemz P, Kucera LJ (1997) Studies for visualizing wood defects using ultra-
sonic in reflection and transmission mode. Holz Roh Werkst 55:339-340
Newmann RH (1992) NMR study of spatial relationship between components in wood cell walls.
Holzforschung 46:207-210
Nicklow RM (1979) Phonons and defects. Treatise Mater Sci TechnoI15:191-225
Nicolotti G, Miglietta P (1998) Using high-technology instruments to assess defects in trees.
J Arbor 24 6:297-302
Niemienen AOK, Koenig JL (1988) NMR imaging - a promising new adhesive evaluation
technique. J Adhesion Sci TechnoI2:407-414
Niemz P (1989) Use of online moisture-measuring instruments in the woodworking industry: lit-
erature review (Einsatz von on-line-Feuchtemessgeraten in der holzverarbeitenden Industrie
- eine Literaturubersicht). Holztechnologie 30 1:9-13
Niemz P, Sander D (1989) Mesuring techniques in wood industrie. Prozessmesstechnick in der
Holzindustrie. VEB Fachbuchverlag, Leipzig, 288 pp
Niemz P, Kucera LJ, Flisch A, Blaser E (1997) Anwendung der Computertomographie an Holz.
(Computerized tomography of wood). Holz Roh Werkst 55:279-280
Niemz P, Bodmer HC, Kucera LJ, Ridder HW, Habermehl A, Wyss P, Zurcher E, Holdenrieder 0
(1998a) Eignung verschiedener Diagnosemethoden zur Erkennung von Stammfliulen bei
Fichte (Suitability of different methods for decay detection in Norway spruce). Schweiz.Z
Forstwes 149:615-630
Niemz P, Kucera LJ, Ridder HW, Habermehl A, Flisch A (1998b) Durchblick mit Computertomo-
graphie (Transversal view with computer tomography). Wald Holz 9:7-10
Niemz P, Kucera LJ, Schob M, Scheffer M (1999) Experiments for defect detection in wood with
ultrasound (Experimentelle Untersuchungen zur Erkennung in Holz mit Ultraschal). Holz
Niemz P, Kucera LJ, Lehmann E, Vontobel P, Hanschke S (2000a) Investigation of wooden corner
connection by neutron radiography methods in order to evaluate the ingress behavior of
water. 12th Symp Nondestructive Testing of Wood, Sopron, p 463
Niemz P, Lehmann E, Vontobe1 P (2000b) Application of neutron radiography for investigation of
wood. 12 th Symp Nondestructive Testing of Wood, Sopron, p 462
314 References
Niemz P, Lehmann E, Vontobel P, Haller P, Hanschke S (2000c) Investigation using neutron radi-
ography for evaluations of moisture ingress into corner connections of wood. 12th Symp on
Wood, University of Western Hungary, Sopron, pp 489-497
Nishino Y, Norimoto M (1990) Structure and anisotropy of dielectric constant in hardwood. Wood
Resh Tech Notes Kyoto Univ 26:78-90
Norimoto M (1976) Dielectric properties of wood. Wood Res Kyoto 59/60:106-152
Norimoto M, Yamada (1972) The dielectric properties of wood. VI. On the dielectric properties
of chemical constituents of wood and the dielectric anisotropy of wood. Wood Res Bull Wood
Res Inst Kyoto Univ:31-43
Norimoto M, Zhao G (1993) Dielectric relaxation of water adsorbed on wood II. Mokuzai
Norimoto M, Hayashi S, Yamada T (1978) Anisotropy of dielectric constant in coniferous wood.
Holzforschung 32:167-172
Norton SJ, Linzer M (1980) Ultrasonic reflectivity imaging in three dimensions: exact inverse
scattering solutions for plane, cylindrical and circular apertures. IEEE Trans Biomed Eng
28:202-220
Nystrom J, Earl Kline D (2000) Automatic classification of compression wood in green Southern
yellow pine. Wood Fiber Sci 32:301-310
Obataya E, Yokoyama M, Norimoto M (1996) Mechanical and dielectric relaxations of wood in a
low temperature range. I. Relaxations due to methylol groups and adsorged water. Mokuzai
Oja J, Temnerud E (1999) The appearance of resin pockets in CT-images of Norway spruce (Picea
Abies (L) Karst). Holz Roh Werkst 57:400-406
Oja J, Grundberg S, Gronlund A (2000) Predicting the strength of sawn products by X-ray scan-
ning of logs: a preliminary study. Wood Fiber Sci 32:203-208
Okumura S, Suzuki R, Fujii Y (1996) Preliminary study on thermography of wood under com-
pression load. Kyoto Daigaku Nogaku Bu Enshurin Hokoku 68:161-169
Olek W, Guzenda R, Baranowska HM, Olszewski J (1994) On possible application of nuclear mag-
netic relaxation measurements to investigation of water penetration into wood. First Euro-
pean Symp on NDE of Wood, University of Sopron, Sopron, pp 130-137
Olek W, Guzenda R, Weres J (2000) Computed-aided estimation of thermal conductivity of Scots
pine wood. 12th Symp NDT of Wood, University of Western Hungary, Sopron, pp 431-438
Olson JR,Arganbright DG (1981) Prediction of mass attenuation coefficients of wood. Wood Sci
14:86-90
Olson JR, Liu CJ, Tian Y, Shen Q (1988) Theoretical wood densitometry: II. optimal X-ray energy
for wood density measurement. Wood Fiber Sci 20:187-196
Olson JR, Chang SJ, Wang PC (1990) Nuclear magnetic resonance imaging: a noninvasive analy-
sis of moisture distributions in white oak lumber. Can J For Res 20:586-591
Onoe M, Tsao JW, Tamada H, Nakamura H, Kogure J, Kawamura H, Yoshimatsu M (1984) Com-
puted tomography for measuring the annual rings on a live tree. Nuclear Instruments and
Methods in Physics Research 221:213-220
Pang S (1996) Moisture content gradient in a softwood board during drying: simulation from
2-D model and measurement. Wood Sci TechnoI30:165-178
Pang S, Wiberg P (1998) Model predicted and CT scanned moisture distributed in a Pinus radiata
board during drying. Holz Roh Werkst 56:9-14
Pang S, Keey RB, Laugrish TAG, Walker JCF (1994) Air flow reversals in high temperature kiln
drying of Pinus radiata boards:l: drying of a single board. N Z J For Sci 24:83-103
Papadakis EP (1999) Ultrasonic instruments and devices. Reference for modern instrumentation,
techniques and technology. Academic Press, San Diego, 809 pp
Passialis C (1997) Physico-chemical characteristics of waterlogged archeological wood. Holz-
forschung 51:11-113
Pearce RB, Sumer S, Doran SJ, Carpenter TA, Hall LD (1994) Noninvasive imaging of fungal col-
onization and host response in the living sapwood of sycamore (Acer pseudoplaranus) using
nuclear magnetic resonance. Physiol Plant Pathol 45:359-384
References 315
Pearce RB, Fisher BJ, Carpanter TA, Hall LD (1997) Water distribution in fungal lesions in the
wood of sycamore (Acer pseudoplatanus), determined gravimetrically and using nuclear mag-
netic resonance imaging. New Phytol 135:675-688
Pei J, Yousuf MI, Degertrkin FL, Honein BV, Kuri-Yakub BT (1995) Lamb wave tomography and
its application in pipe erosion/corrosion monitoring. Proc IEEE Ultrasonics Symp 1995,
Seattle, Washington, pp 795-798
Pellerin RF (1965) A vibrational approach to nondestructive testing of structurallumber. For Prod
J 153:93-101
Peters RA (1995) A new algorithm for image noise reduction using mathematical morphology.
IEEE Trans Med Imag 4:554-568
Peterson KR (1993) The role of nondestructive evaluation in assuring the wise use of our
timber resource. 9th Symp NDT of Wood, Washington State University Press, Madison,
pp 7-9
Peyskens E, Pourq M de, Stevens M, Schalck J (1984) Dielectric properties of softwood species at
microwave frequencies. Wood Sci TechnoI18:267-280
Pfeffer PE, Gerasimowicz WV (1989) Nuclear magnetic resonance in agriculture. CRC Press, Boca
Raton, 441 pp
Polge H (1966) L'analyse densitomeltrique de cliches radiographiques: Une nouvelle methods de
determination de la structure du bois. Ann Sci For 20 4:530-581
Polge H (1978) Fifteen years of wood radiation densitometry. Wood Sci TechnoI12:187-196
Poliszko S, Hoffmann G (1985) Dielectric behavior of wood-polystyrene composite. J Appl
Polymer Sci 30:799-804
Price DL, Sk6ld K (1986) Introduction to neutron scattering. In: Sk6ld K, Price DL (eds) Neutron
scattering, vol 23, part A. Academic Press, Orlando, pp 1-98
Purslow DF (1971) The use of microwave moisture meter for studying moisture changes in
timber. J Inst Wood Sci 5 4:40-46
Puttick KE (1987) Thermal NDT methods. In: Summers cales J (ed) Nondestructive testing of
fibre-reinforced plastics composites. Elsevier, London, pp 65-103
Quick J], Hailey JRT, Mac Kay AL (1990) Radial moisture profiles of cedar sapwood during drying:
a proton magnetic resonance study. Wood Fiber Sci. 22:404-412
Quin F Jr, Steele PH, Shmulsky R (1998) Locating knots in wood with an infrared detector system.
For Prod J 48 10:80-84
Radon J (1917) Uber die Bestimmung von Funktionen durch ihre Integralwerte langs gewisser
Manningfaltigkeiten (On the determination of functions from their integrals along certain
mainfolds). Ber Sach Akad Wiss 29:262-277
Rafalski J (1966) On the dielectric properties of compress red beech (Uber die dielekkyrischen
Eigenschaften unterschiedich verdiichteten Rotbuchen vollholzes). Holztechnologie 7 2:118-
122
Ranta L, May HA (1978) Measuring the density profile of particleboard by gamma radiation. Holz
Rantala J, Wu D, Busse G (1996) Vibrothermography applied for nondestructive evaluation of
polymer materials. Mater Sci Forum 210-213:433-438
Raschi A, Tognetti R, Ridder HW, Beres C (1995) Water in the stems of sesilis oak (Quercus
petraea) by computer tomography with concurrent measurements of sap velocity and ultra-
sound emission. Plant Cell Environ 18:545-554
Reimers P, GilboyWB, Goebbels J (1984) Recent development in the industrial application of com-
puterized tomography with ionizing radiation. NDT Int 174:197-207
Reynolds WN, Wells GM (1984) Video compatible thermography. Br J NDT 26 1:40-44
Rice RW, Steele PH, Kumar L (1992): Detecting knots and voids in lumber with dielectric sensors.
Ind Metrol2 3/4:309-315
Richards JA (1990) Radar backscatter modeling of forests: a review of current trends. Int J Remote
Sens 11:1299-1312
Rinn F (1991) Three dimensional impulse tomograph for examination of trees and timber. Engi-
neering and Distribution, Heidelberg, 4 pp
316 References
Robertson MB, Packer KJ (1999) Diffusion of D20 in archaeological wood measured by I-D NMR
profiles. Appl Magn Reson 17:49-64
Rosencwaig A, Busse G (1980) High resolution photoacoustic thermal wave microscopy
Appl.Phys.Lett 36 : 725-727
Rosencwaig A (1983) Photo acoustics in material science. In: Herman H (ed) Treatise on
material science and technology, vol 198. Academic Press, NewYork, pp 67-117
Ross RJ, Pellerin RF (1991) Nondestructive evaluation of wood past, present and future. In:
Ruud CO, Green RE (eds) Nondestructive characterization of material, vol IV, Plenum Press,
New York, pp 59-64
Ross RJ, Pellerin RF (1994) Nondestructive testing for assessing wood members in structures. A
review. Gen Tech Rep FPL-GTR-70, USDA, Forest Product Laboratory, Madison, WI, 40 pp
Rothwell WP, Holecek DR (1984) NMR imaging: study of fluid absorption by polymer compos-
ites. J Polymer Sci Polymer Lett Ed 22:241-247
Roux S (1998) Scales. In: Frantziskonis GN (ed) Proc PROBAMAT - 21st Century: probability and
materials. Test, models and applications for the 21th century. Kluwer, Dordrecht, pp 573-
577
Rust S (2000) A new tomographic device for the nondestructive testing of trees. 12th Symp on
Nondestructive Testing of Wood, University of Western Hungary, Sopron, pp 233-237
Sadoh T, Murata K (1993) Detection of knots in hinoki and karamatsu lumber by thermography.
Mokuzai Gakkaishi 39:13-18
Sahimi M, Arbabi S (1993) Mechanics of disordered solids II. Phys Rev B 47:703-712
Salerno A, Wu D, Busse G, Rantala J (1997) Thermographic inspection with ultrasonic excitation
In: Thompson DO, Chimenti DE (eds) Review of progress in quantitative nondestructive eval-
uation, vol 16. Plenum Press, New York, pp 345-352
Salerno A, Dillenz A, Wu D, Rantala J, Busse G (1998) Progress in ultrasonic lockin thermogra-
phy. In: Balageas D, Busse G, Carlomagno GM (eds) Quantitative infrared thermography.
QIRT 98, Lodz, pp 164-160
Samson M (1988) Traverse scanning for automatic detection of general slope of grain in lumber.
For Prod J 38 7/8:33-38
Sasov A, van Dyck D (1998) Desktop X-ray microscopy and microtomography. J Microsc 191 pt
2:151-158
Schau D (1984) Fourier transform NMR. Elsevier, Amsterdam, 344 pp
Schechter RS, Chaskelis HH, Mignogna RB, Delsanto PP (1994) Real time parallel computation
and visualization of ultrasonic pulses in solids. Science 265:1188-1196
Schechter RS, Mignona RB, Delsanto PP (1996) Ultrasonic tomography using curved ray paths
obtained by wave propagation simulations on a massively parallel computer. J Acoust Soc Am
104:2103-2111
Schienwind AP (1990) Physical and mechanical properties of archeological wood. In: Rowell RM,
Barbour RJ (eds) Archeological wood properties, chemistry and preservation. American
Chemical Soc, Washington, DC, pp 87-109
Schirone A, Lo Monaco A (1988) L'uso dei raggi X in xililogia con particolare riferimento alle
indagini dendroclimatologiche (The utilization of X-ray for dendroclimatology). Economia
Montana (Linea Ecologica) 5:47-54
Schmidt RG, Frazier CE (2000) Solid state NMR analysis of adhesive bondlines in pilot scale flake-
boards. Wood Fiber Sci 32:419-425
Schmidt T (1978) Scanning computing methods for measuring knots and other defects in lumber
and veneer. Proc 4th Nondestructive Testing of Wood, Washington State University, Pullman,
pp 23-25
Schmoldt DL (1996) CT imaging, data reduction, and visualization of hardwood logs. In: Meyer
DA (ed) Proc 24th Annual Hardwood Symposium, National Lumber Assoc Cashiers, National
Hardwood Lumber Assoc, Cashiers, North Carolina, pp 69-80
Schmoldt DL, Zhu D, Conners RW (1993) Nondestructive evaluation of hardwood logs using auto-
matic interpretation of CT images. In: Thompson DO, Chimenti DE (eds) Review of progress
in quantitative nondestructive evaluation, vol 12. Plenum Press, New York, pp 2257-2264
References 317
Schmoldt DL, Li P, Araman PA (1996a) Iterative simulation of hardwood log veneer slicing using
CT images. For Prod J 46 4:41-47
Schmoldt DL, Nelson RM, Ross RJ (1996b) Ultrasonic defect detection in wooden pallet parts for
quality sorting. Proc SPIE 2944:285-295
Schmoldt DL, Pei Li A, Abbot Lyn A (1997) Machine vision using artificial neural networks with
local 3-D neighborhoods. Comput Electr Agric 16:255-27l
Schmoldt DL, Occena LG, Lynn Abbott A, Gupta NK (1999) Nondestructive evaluation of hard-
wood logs: CT scanning, machine vision and data utilization. Nondestr Test Eval 15:279-309
Schmoldt DL, He J, Lynn Abbott A (2000) Automated labeling of log features in CT imagery of
multiple hardwood species. Wood Fiber Sci 32:287-300
Schneider R, Netzelmann U, Kroning M (1996) Characterization of green ceramics by microwaves
and ultrasound. Mater Sci Forum 210-213:77-84
Schniewind AP (1981) Concise encyclopedia of wood and wood-based materials. Pergamon Press,
Oxford, 354 pp
Schomberg H (1982) Nonlinear image reconstruction from ultrasonic time to flight projections.
Acoust Imaging 10:381-396
Schulte M, FrUhwald A, Welling H, Kruse K, Broker FW (1995). Nondestructive in-line quality
control in board production. Tappi Proc of 1995 European plastic laminates forum "The
leading edge", Heidelberg, pp 7l-79
Schultz V, Ward Whicker F (1982) Radioecological techniques. Plenum Press, New York, 298 pp
Senft JF, Suddarth SK, Angleton RD (1962) A new approach to stress grading of lumber. For Prod
J 124:183-186
Sepulveda P, Gonlund A (2000) Measurement of spiral grain with computed tomography com-
pared with pattern on boards. Proc 12th Symp Nondestructive Testing of Wood, Sopron, pp
239-244
Sharp AR, Riggin MT, Kaiser R (1978) Determination of moisture content of wood by pulsed
nuclear magnetic resonance. Wood Fiber 10:74-81
Siau JF (1984) Transport process in wood. Springer, Berlin Heidelberg New York, 245 pp
Siau JF (1995) Wood: influence of moisture on physical properties. Department of Wood Science
and Forest Products, Virginia Polytechnic Institute and State University, Blacksburg, Virginia,
227 pp
Siau JF, Meyer JA (1973) Neutron activation analysis of pentachlorophenol in treated wood. Wood
Sci 6 1:19-21
Sieber AJ (1985) Forest signatures in imaging and nonimaging microwave scatterometer data.
ESA J 9 4:431-448
Skaar C (1948) The dielectric properties of wood at several radio frequencies. Tech Pub no 69.
NY State College of Forestry, Syracuse
Skaar C (1988) Wood in water relations. Springer, Berlin Heidelberg New York, 283 pp
Skatter S (1998) Determination of cross-sectional shape of softwood logs from three X-ray pro-
jections using an elliptical model. Holz Roh Werkst 56:179-186
Slichter CP (1978) Principles of magnetic resonance. Springer, Berlin Heidelberg New York, 397 pp
Schniewind AP (1990) Physical and mechanical properties of archaeological wood. In: Rowell RM,
Barbour RJ (eds) Archaeological wood: properties, chemistry and preservation. Advances in
Chemistry Series no 225, pp 87-109
Sobue N (1993) Nondestructive characterization of wood. Mokuzai Gakkaishi 39:973-979
Socco V, Martinis R, Sambuelli L, Comino E, Nicolotti G (2000) Open problems concerning ultra-
sonic tomography for wood decays diagnosis. 12th Symp NDT of Wood, University of Western
Hungary, Sopron, p 468
Socco V, Sambuelli L, Martinis R, Nicolotti G, Comino E (2002) Feasability of ultrasonic tomog-
raphy for NDT of decay on living trees. Res Nondestr Eval (in press)
Southon TE, Mattsson A, Jones RA (1992) NMR imaging of roots: effects after root freezing of
containerized conifer seedlings. Physiol Plant 86:329-334
Steele PH, Cooper JE (2000) Estimating lumber strength with radio frequency scanning. 12th
Symp Nondestructive Testing of Wood, Sopron, pp 343-348
318 References
Steele PH, Harless TEG, Wagner FG, Taylor FW (1989) Potential dollar increase from internal log
information. Proc 7th Symp on Nondestructive Testing of Wood, Washington State University,
Pulman, pp 241-250
Steele PH, Neal SC, McDonald KA, Cramer SM (1992) The slope of grain indicator for defect detec-
tion in unplaned hardwood lumber. For Prod J 41 1:15-20
Steele PH, Quin F Jr, Cooper E (1998) Infrared detection of knots in southern yellow pine lumber
heated with radiant heat. 11 th Symp NDT of Wood, Washington State University, Pullman, pp
57-62
Steele PH, Patton MD, Cooper E (2000a) Factor influencing differential thermal response of knots
and clear wood. 12th Symp NDT of Wood, University of Western Hungary, Sopron, pp 403-412
Steele PH, Patton MD, Cooper E (2000b) A thermal recognition system to identify knots in wood.
Proc wood technology. Clinic and Shows, Portland, Oregon, pp 255-266
Steele PH, Lalit Kumar, Shmulsky R (2000c) Differentiation of knots, distorted grain and clear
wood by radio frequency scanning. For Prod J 50 3:59-62
Sugimori M, Lam F (1999) Macro-void distribution analysis in strand-based wood composites
using an X-ray computer tomography technique. J Wood Sci Jpn Wood Res Soc 45:245-257
Sulivan C, Kline R, Mignogna RB, Delsanto PP (1996) A parallel processing approach to acoustic
tomograph. J Acoust Soc Am 99:2142-2147
Suryoatmono B, Cramer S, Shi Y, McDonald KA (1994) Within-board lumber density variations
from digital X-ray images. 9th Symp Nondestructive Testing of Wood, Washington State
University. Eng Publications, Madison, pp 168-175
Swanson JS, Hailey JRT (1987) Scanning and imaging techniques for assessing decay and wood
quality in logs and standing trees. Proc 6th Symp Nondestructive Testing of Wood, Washing-
ton State University. Eng Publications, Pullman, pp 83-94
Szendrodi L, Habermehl A, Ridder HW (1994) Computer tomographic investigation of stand-
ing trees in Hungary. 1st European Symp on Nondestructive Evaluation of Wood, Sopron, pp
503-511
Szymani R, McDonald KA (1981) Defect detection in lumber: state of the art. For Prod J 31
11:34-44
Talpe J (1971) Theory of experiments in paramagnetic resonance. Pergamon Press, Oxford
Tan HS (1981) Microwave measurements and modeling of the permittivity of tropical vegetation
samples. Appl Phys 25:351-355
Tanaka T (1994) Thermographic detection of deteriorated locations and nondestructive evalua-
tion of strength of biodeteriorated wood. 9th Symp NDT of Wood, Washington State Univer-
sity, Pullman, pp 78-83
Tanaka T, Divos F (2000) Thermographic inspection of wood. 12th Symp NDT of Wood, Univer-
sity of Western Hungary, Sopron, pp 439-447
Tanaka T, Norimoto M, Yamada T (1975a) Anisotropy of dielectric constant of wood. J Soc Mater
Sci 24 264:867-872
Tanaka T, Norimoto M, Yamada T (1975b) Dielectric properties and structure of wood I. Mokuzai
Taylor FW, Wagner FG Jr, McMillin CW, Lon Morgan I, Hopkins FF (1984) Locating knots by
industrial tomography - a feasibility study. For Prod J 34 5:42-46
Thomas RL, Favro LD, Kuo PK,Ahmed T, Han X, Wang L, Wang X, Shepard SM (1995) Pulse echo
thermal wave imaging for nondestructive evaluation. Proc 15 th Int Confrerence Acoustics,
Trondheim, pp 433-436
Thompson F (1989) Moisture measurements using microwaves. Meas Control 22:210-215
Thompson F (1996) Non destructive moisture measurement using microwaves. Mater Sci Forum
210-213:85-92
Tice D, Euskirchen J (1976) Faster and more effective forest fire surveillance with live IR imaging
superimposed on visible terrain view. Proc XVlth IUFRO World Congress, Oslo, pp 53-59
Tiitta M, Olkkonen H, Lappalainen T, Kanko T (1996) Density measurements of particleboard
veneer and wood specimens by narrow beam gamma absorption technique. 10th Symp Non-
References 319
destructive Testing of Wood Swiss Federal Institute of Technology, Press Univ Romande,
Tinelli A, Catena G (1991) Injuries on ornamental trees in Castelporziano Indagine sulle piante
monumentali della tenuta di Castelporziano. Monti Boschi 5:9-13
Tiuri M, Jokela K, Heikkela S (1980) Microwave instrument for accurate moisture and density
measurement of timber. J Microwave Power 15:251-254
Tognetti R, Raschi A, Beres C, Fenyvesi A, Ridder HW (1996) Comparison of sap flow, cavitation
and water status of Quercus petraea and Quercus cerris trees with special reference to com-
puter tomography. Plant Cell Environ 19:928-938
Tomikawa Y, Iwase Y, Arita K, Yamada H (1985) Nondestructive inspection of rotted or termite
damaged wooden poles by ultrasound. Proc 5th Symposium on Ultrasonic Electronics, Tokyo
1984. Jpn J Appl Phys 24 1985 Suppl24 1:187-189
Tomikawa Y, Iwase Y, Arita K, Yamada H (1986) Nondestructive inspection of wooden pole using
ultrasonic computed tomography. IEEE Trans UFFC 33 4:354-358
Torgovnikov GI (1990) Dielektrische Eigenschaften von absolut trockenem Holz und Holzstoff
(Dielectric properties of oven dried wood). Holztechnologie 30 1:9-11
Torgovnikov GI (1993) Dielectric properties of wood and wood based materials. Springer, Berlin
Heidelberg New York, 196 pp
Tremblay C (1995) Longitudinal and radial variation of slope of grain in black spruce lumber.
For Prod J 45 1:79-83
Troughton GE, Clarke MR (1987) Development of a new method to measure moisture content in
unseasoned veneer and lumber. For Prod J 37 1:13-19
Tsai WH (1985) Moment-preserving thresholding: a new approach. Comput Vision Graph 29:
377-393
Tsuchikawa S, Takahashi T, Tsutsumi S (2000) Nondestructive measurement of wood properties
by using near-infrared laser radiation. For Prod J 50 1:81-86
Ulaby FT, Jedlicka RP (1984) Microwave dielectric properties of plant materials. IEEE Trans
Geosci Remote Sensing GE 22:406-414
Ulaby FT, Razni M, Dobson MC (1983) Effect of vegetation cover on microwave radiometric sen-
sitivity to soil moisture. IEEE Trans Geosci Remote Sensing GE 21:51-61
Ulaby FT, Sarabandi K, McDonald K, Whitt M, Dobson MC (1990) Michigan microwave canopy
scattering model. Int J Remote Sensing 11:1223-1253
Unger A, Planitzer J, Morgos A (1988) X-ray computer tomography and magnetic resonance
tomography for characterizing wet archaeological wood. Holztechnologie 29:249-250
USDA, Forest Product Laboratory, Forest Service (1972) Thermal properties. In: Wood handbook:
wood as an engineering material. USDA, Madison, WI, pp 324-326
Uyemura T (1960) Dielectric properties of wood as the indicator of the moisture. Bull Gov For
Exp Stn Tokyo 119:95-172
Vansteenkiste D (2000) Mise au point d'une methode rapide d'analyse quantitative de l'anatomie
du bois de cMne: analyse automatique des cliches radiographiques. In: Bobillier E, Jacob D
(eds) 6eme journee de la mesure - Electronique, Informatique, Automatique. INRA, Rennes,
8 pp
Varga A, Matusik A, Kiss E, Divos F (2000) Use of medical CT to qualify defects in wood. 12th
Symp Nondestructive Testing of Wood, Sopron, p 466
Vavilov V (1992) Thermal non destructive testing: short history and state of art. In: Balageas D,
Busse G, Carlomagno GM (eds) Proc Quantitative Infrared Thermograpfy (QIRT92). Editions
Europeennes Thermiques et Industrie, Paris, pp 179-194
Veres JS, Cofer GP, Johnson GA (1993) Magnetic resonance imaging of leaves. New Phytol 132:
769-774
Vermass HF (1973) Regression equations for determining the dielectric properties of wood. Holz-
forschung 27:132-136
Vincent L (1993) Morphological grayscale reconstruction in image analysis. Applications and effi-
cient algorithms. IEEE Trans Image Process 2:176-201
320 References
Vinegar HJ (1986) X-ray computed tomography and NMR imaging of rocks. J Petrol Technol
March:257-259
Volgyi F (2000) Microwave NDT of particleboards. 12th Symp Nondestructive Testing of Wood,
Sopron, pp 357-365
Vozzo JA, Halloin JM, Cooper TG, Potchen EJ (1996) Use of NMR spectroscopy and MRI for dis-
criminating Juglans nigra L. seeds. Sci TechnoI24:457-463
Wagner FG, Taylor FW (1985) Economic returns from internal log scanning. Fifth Nondestruc-
tive Testing of Wood Symp, Washington State University, Pullman, pp 267-280
Wagner FG, Taylor FW, Ladd DS, Mcmillin CW, Roder FL (1989a) Ultrafast CT scanning of an oak
log for internal defects. For Prod J 3911112:62-64
Wagner FG, Taylor F, Ladd D,McMillin C, Roder F (1989b) Ultrafast CT scanning oflogs for inter-
nal defects. 7th Symp Nondestructive Testing of Wood Symposium, Washington State Uni-
versity. Eng Publications, Pullman, pp 221-229
Wang PC, Chang SJ (1986) Nuclear magnetic resonance imaging of wood. Wood Fiber Sci 18:
308-314
Wang PC, Chang JS, Olson JR (1990). Scanning logs with an NMR scanner. 7th Symp Nonde-
structive Testing of Wood, Washington State University, Pullman, pp 209-219
Wang YQ, Kline RA (1994) Ray tracing in isotropic and anisotropic materials: application to tomo-
graphic image reconstruction. J Acoust Soc Am 95 2525-2532
Weatherwax RC, Stamm AJ (1947) The coefficients of thermal expansion of wood and wood prod-
ucts. Trans ASME May:421-432
Weibe S (1992) Untersuchungen zur Wundenwicklung und Wundbehandlung an Biiumen
unter besonderer Beruchsichtigung der Holzfeuchte. PhD thesis, Univ Miinchen, Miinchen,
Germany
Wert CA, Weller M, Caulfield D (1984) Dynamic loss properties of wood. J Appl Phys 56 9:
2453-2458
White MS, Ifju G, Johnson JA (1977) Methods for measuring resin penetration into wood. For
Prod J 27 7:52-54
Whittall KP, MacKay AL (1989) Quantitative interpretation ofNMR relaxation data. J Magn Reson
84:134-152
Wiberg P, Moren TJ (1999) Moisture flux determination in wood during drying above fiber
saturation point using CT scanning and digital image processing. Holz Roh Werkst 57:
137-144
Wiberg P (1996) CT-scanning during wood drying for nondestructive moisture content gradient
evaluation. Proc 10th Nondestructive Testing of Wood Symposium. Swiss Federal Institute of
Technology, Press Poly techniques et Universitaires Romandes, Lausanne, p 409
Wickman BE (1985) Comparison of the degree-day computer and a recording thermograph in a
forest environment. Reserch Note PNW-427. USDA, Forest Service, Madison, WI, 6 pp
Willis RL, Stone TS, Berthelot YH, Madigorski WM (1997) An experimental numerical technique
for evaluating the bulk and shear dynamic moduli of viscoelastic materials. J Acoust Soc Am
102:3549-3555
Willis RL, Wu L, Berthelot YH (2001) Determination of the complex Young and shear moduli of
viscoelastic materials. J Acoust Soc Am 109:611-621
Wilson DW, Charles JA (1981) Thermographic detection of adhesive-bond and interlaminar flaws
in composites. Exp Mech 21:276-280
Wilson MA, Godfrey 1M, Hanna JV, Quezada RA, Finnie KS (1993) The degradation of wood in
old Indian Ocean shipwrecks. Org Geochem 20 5:599-610
Winistorfer PM, Moschler WW Jr (1986) A direct scanning densitometer to measure density pro-
files in wood composite products. For Prod J 36 11112:82-86
Wisniewski M, Lindow SE, Ashworth EN (1997) Observations of ice nucleation and propagation
in plants using infrared video thermography. Plant Physiol113 2:327
Wissing ST, Wellinng J (1995) Nondestructive measurement of the density profile in wood based
panels by means of radio frequency. Proc Int Conf Progress in Forest Products Research,
Gottingen, p 165
References 321
Wolter B, Netzelmann U (1996) Nondestructive testing of wood using a single sided nuclear mag-
netic resonance. 10th Int Symp on Nondestructive Testing of Wood. Press Poly techniques et
Universitaires Romandes, Lausanne, p 409
Wu D (1992) Lockin thermography for multiplex photothermal nondestructive evaluation. In:
Balageas D, Busse G, Carlomagno GM (eds) Proc Quantitative Infrared Thermograpfy
(QIRT92). Editions Europeennes Thermiques et Industrie, Paris, France, pp 371-376
Wu D (1994) Lock in thermography for defect characterization in veneers. In: Balageas D, Busse
G, Carlomagno GM (eds) Proc Quantitative Infrared Thermograpfy (QIRT94). Editions
Europeennes Thermiques et Industrie, Paris, France, pp 298-301
Wu D, Busse G (1995) Remote inspection of wood with lock in thermography. Proc of 1995 Euro-
pean Plastic Laminates Forum "The leading Edge", Heidelberg, pp 27-29
Wu D, Busse G (1996) Remote inspection of wood with lock-in thermography. Tappi J 79
8:119-123
Wu D, Karpen W, Busse G (1992) Lockin thermography for multiplex photothermal nondestruc-
tive evaluation Proc Quantitative Infrared Thermography (QIRT 1992). Eurotherm Ser
27:37l-376
Wu D, Salerno A, Sembach, J, Maldague X, Rantala J, Busse G (1996) Wood-based products in-
spection by lockin thermography. Proc 3rd Eurowood Symp and 4th FESYP Technical Conf
Braunschweig, Germany, 5 pp
Wu D, Salerno A, Sembach J, Maldague X, Rantala J, Busse G (1997a) Lockin thermographic
inspection of wood particleboard. SPIE 3056:230-234
Wu D, Sembach J, Salerno A, Hora G, Busse G (1997b) Ensuring the quality of coated wood-based
products by means of lock-in thermography. (Qualitatssicherungg beschichteter Holzwerk-
stoffe mittels lockin-Thermographie). Holz -ZentralbI123:50-775
Wycoff W, Pickup S, Cutter B, Miller W, Wong TC (2000) The determination of the cell size in
wood by nuclear magnetic resonance diffusion techniques. Wood Fiber Sci 32:72-80
Xu Y, Okumura S, Noguchi M (1993) Thermographic detection of starved joints of wood. Mokuzai
Xu Y, Okumura S, Noguchi M (1994) Thermographic detection of starved joints of wood. 9th
Symp NDT of wood. Forest Product Soc, Madison, pp 209-217
Yamanaka K, Nagata Y, Koda T (1991) Enhancement of acoustic imaging of polymers by cooling.
Ultrasonics 29 March:159-165
Yokoyama M, Norimoto M (1996) Contour diagrams of dielectric loss for absolutely dried spruce
wood. Wood Res Kyoto Univ 83:37-39
Yokoyama M, Norimoto M (1997) Cole-Cole plots for dielectric properties of absolutely dried
wood. Tech Notes 33:71-82
Yokoyama M, Obataya E, Norimoto M (1999) Mechanical and dielectric relaxations of wood in
a low temperature range. II. Relaxations due to adsorbed water. Mokuzai Gakkaishi 45:95-
102
Yokoyama M, Kanayama Y, Furuta Y, Norimoto M (2000a) Mechanical and dielectric relaxations
of wood in a low temperature range. III. Application of search law to dielectric properties due
to adsorbed water. Mokuzai Gakkaishi 46: 173-180
Yokoyama M, Ohmae K, Kanayama K, Furuta Y, Norimoto M (2000b) Mechanical and dielectric
relaxations of wood in a low temperature range. IV. Dielectric properties of adsorbed water
at high moisture contents. Mokuzai Gakkaishi 46:523-530
Youngquist JA, Hamilton TE (1999) The next century of wood products utilization: a call for
reflection and innovation. Proc Int Conf on effective utilization of plantation timber, Chi -Tou,
Taiwan. For Prod Assoc ROC Bull 16:1-9
Zhu D, Beex AA, Conners R (1991) Stochastic field-based object recognition in computer vision.
SPIE 1569:174-181
Zielonka P, Dolowy K (1998) Microwave drying of spruce: moisture content, temperature and heat
energy distribution. For Prod J 48 6:77-80
Zimmermann MH (1983) Xylem structure and the ascent of sap. Springer, Berlin Heidelberg
New York, 143 pp
322 References
Zoughi R (1990) Microwave nondestructive testing: theories and applications. lnt Adv Nondestr
Test 15:225-288
Zoughi R (1996) Principles of and applications of microwave and millimeter wave NDE
methodologies: an overview. SPIE Proc 2944:46-74
Zoughi R, Ganchev S, Carriveau GW (1996) Overview of microwave NDE applied to thick com-
posites. Mater Sci Forum 210-213:69-76
Subject Index
A Amplitude (see also magnitude)

Abiotic deterioration 69 Microwave 128,132-137,148,149,
Absorption coefficient of X-rays 15 156-159,167,169,170,174,175,177-179
Abundance 220,232-234 Ultrasound 185,191,198,200,213
Accuracy of measurements NMR 228,229,241,262
Density 20, 28, 40, 55, 60, 68 Neutron 282
X-rays attenuation coefficient 7,20,23, Anatomical structure of wood 14, 33, 36, 46,
24,28,47 49,50,71,147,292
Ray path 35 Cell lumen 232, 254, 255, 262, 264, 265 see
Thickness 118 lumina
Stress grading 91,93 Fiber direction 55,82,84,99, 115, 129,
Ultrasonic velocity 198, 199,213 130,133,138,139,151
Acoustic Fiber-element 242,247,276
Emission 7,51,278 Medullary rays 82,83,98,99,144,181,
Energy 185 242,254,265,266,275,276,292
Field 188 Parallel or perpendicular to fibers
166,
Ray path 182 171,22,260
Impedance contrast 1965-198 Tracheids 46,51,254,265
Microscopy 200 Vessels 49,51,99,115,242,247,276,287
Wavelength 79 Angular frequency 79,127,131,132
Acoustical properties of wood 2,7, 182 Anisotropic
Acryl 10, 40, see polyacryl Axes 27,55,108,127,129,169,198,221
Adhesives 113-116, 192, 268-27l Medium 159,169,182,277
Curing 69, 279 Directions of wood 33, 35, 38, 39, 55, 57,
Degree of condensation 69 64,65,71,79,81,82,102,108-111,128,
Adiabatic compressibility 277 129,133,134,138,139,141-145,147,151,
Aerosol freezer spray 122 152,159,161,170,181-183,189,193-196,
Aesthetic 2 202,209,292
Aging process of the tree 52 Anisotropy 8,38,39,55,80,81,96, 102, 108,
Air 17,20,40,43,47,65,82,84,95,102,133, 110,127,128-139,141-147,159,169,
154,156,159,192-194,196-198,268, 176-179,181-183,196,200,208,210
292 Annual ring
Velocity 65,194 Age 25
Coupled ultrasonic transducers 189,191, Cracks 208
196,198,199,214 Densitometric analysis 36, 99
Algebraic reconstruction technique (ART) Density mapping 98
181,185-187,213 Disposition 92, 11,287,289
Alignment 39,217 Distribution 31
Algorithm Earlywood and latewood 47,99
X-ray 24-27 Growth 287
Microwave 148 Iso-density 38
Ultrasound 181,185-187,202,204,213 Limit 36-39, 49, 64
NMR 237,241,242,244,278 Moisture distribution 289
Aluminum 84,192,197 Pattern 45,56,92,208,210,287
324 Subject Index
Shape 4,31,37 C
Shrinkage 282 C-scan imaging 189
Structure 26, width 4,22,25,45-47,44, Calibration 51,169,174
45,100 Cambial zone 240,247,248
Water deficiency 287 Canopy 161
Antenna 163 Camera:
Archeological 69,259,273-276 CCD 156,157
Architecture of the scanner 28, 29 Video camera in visible light 75
First generation 30, 70 Infrared video camera 76,77,83,85,
Second generation 13,14,30 86,87,90,95,98,113,115,120,
Third generation 30, 70 122
Fourth generation 28,31,70 Capillary
Array of detector 13,19,29,32,36,55,70, Channels 247
109,186,196,208,238 Forces 234
Artifacts 149 Size 247
Beam hardening 22, 39, 40, 71, 72 Tube 242
Polychromaticity 40 Water 235
Artificial neural networks 25, 59 Carr-Purcell/Meiboom Gill sequence 226,
Aperture 35, 39 227,254,255,270,271
Approximation Born, Rytov 187 Cavity resonators 155, 178
Ash ions 33 Classification of NDT methods 4-9
A-scan imaging 189 Climatology 40, 69, 70
Atomic Cellulose 22,23, 34, 35, 125, 126, 142,
Number 14,70 144,220,232,256-259,268,279,
Arrangement 282 282
Reactor 283-285 Cellulose crystallinity 1,146,147,220
Attenuation of Ceramics 28,113,132,198
Microwave 128,133-140,144,150,161, Chemical
170,175-178 Cell wall 256-259
Neutron 282 Characterization 2,244, 279
Ultrasound 119,182,200,214 Constituents of wood 14,22,69,70,77,
X rays 19,20,23,24,28,38 137,146,147,175,282,287
Axes of elastic symmetry of wood, see Environment 217
longitudinal or radial or tangential Exchange 260
Imaging with liquid crystals 77, 122
B Shift 256, 268
B-scan imaging 189 Treatment 3,272,274
Bark 159,160,191,248,251,288,289 Christoffel equation 182, 183
Beam path 16, see ray Chip board 113
Bending loading 92, 112 Coalification of wood 275
Biologic Coding of tomograms
Degradation 49, 54, 69, 252 Color 14,43,44,90,99,106,108,119,120,
Specimen 49 121,182,183,185
Blackbody radiation 75 Gray scale 44
Bloch equations 218,252 Cole-Cole equation 145
Blows 192 Colimated beam 28, 44
Blister 192 Colimators 15
Bolzmann 217 Compensatory materials (paraffin, wax,
Bound water 12,135,172,173,224 granular sugar, acryl) 40
Branches 161 Compression loading 93,94,109-111
Bridge 28, 149 Compression wood 264, 265
Building Compton effect 71
Materials 294 Computation time for image reconstruction
Inspection 70 46
Subject Index 325
Computed tomography Simulated 57

Applications 40-69,89-120,158-177, Visualization 60
200-212,244-276,285-293 Debye effect 127
NMR 215-276 Decay identification see decay of the signal
Ultrasonic 181-213 Decay of the signal
Concrete specimen 28, 84 Time 99,101
Cone beam reconstruction 25, 57 Thermal 79
Control of adhesion 270, 271 NMR 222,227-232,234,235,241,246,252,
Conventional 255,279
Contact mode 199 Decay of wood 90,95,193,199,204,213,
Image processing 67 214,249,250,255,265,266
Kiln drying 64 Defects in trees 14-51,90,202-208
Radiographic images 13,14,37,71 Cavities 91
Conversion of attenuation coefficients in Decay of the trunk 248-250,251,253
density 17 Metallic inclusions 42
Convolution 24, 46 Frost cracks 90
Corner connections 296-298 Wet core 41,43
Cost 21,28-30,35,60,63,71,148,247,267, Defects detection in wood 52-59,92-97,
278,279,298 169,208
Counting Compression wood 264
Equipment 14,68,71 Cracks 25-28,43,54-56,120,170,208,
Rate 43 214,194,199,292
Ring 43 Resin pockets 54, 240
Time 68 Slope of grain 96-98,210
CT number 40, 59, 72 Spiral grain 54, 125
CT see computerized tomography Tension wood 252
Coupling medium 195 Defects in lumber joints 120,121
Cycling loading 76,77,92,93,113,120,122 Defects in poles 212-213
Defects in wood-based composites 208-
D 212
Data with microwaves Delaminations in plywood 35,87,115,214
Collection 156,157,161 Dendroarcheology 40,69,273-276
Experimental 133,147,153,161,166,167 Dendrochronology 40, 69, 70
Real time 169, 173, 179 Densitometric profile of
Data with neutrons Top plate of a violin 70
Real time 229 Wood 38,39,63,67
Data with NMR Density of wood 33,37-73
Collection 238,239, 256 Cell wall 14
Experimental 22, 224, 228, 230, 232, 236, CT numbers 40, 59, 72
241,246,247,248,254,268,269 Fiberboard 67
Data with thermal waves Isodensity 38,37
Collection 98 Oven-dry wood 20-23,64
Experimental 95, 102 Particleboard 67,68
Interpretation 77 Depolarization of microwaves129, 133, 139,
Data with ultrasound 149,166
Collection 181,188,192,200,202,213,214 Detector 13-16,29-32,55,60,63,70,71
Experimental 185-189 Aperture 35
Interpretation 241 Deterioration by
Data with X rays Microbial and chemical agents 69
Collection 24,25,29,30,31,33,43,51,54, Cell integrity 69
55,60,63,70 Weathering 295
Experimental 37,61,67 Dewar chamber 86, 87
Real time 54,71 Dielectric constants 127-159,170-176
Interpretation 36 Real 128,131,132,175
326 Subject Index
Imaginary 128,131,132 Economic ground 28,41

Complex 130131,158,175 Elastic
Dielectric Characterization of wood 10, 129
Capacitance 131, 132, 170 Constants of wood 200,208,212,245,276,
Conductivity 131 277
Permittivity 132 Electric
Susceptibility 131,132,155 Current density imaging 240
Tensor 137,128,129,158,159,166,167 Field 126,127,129,131,132,152,166,
Diffracting sources 46,79, 149, 158, 169, 169
185-188,213,214 Model of annual ring 144
Diffusion 69,79,216,221-226,237,247, Electronic
252-261,265,272,275-278,282 Hammer 119
Digital 20,169,190,200,214 Signal 13,77,85,87,119,122,126
Discrete 36, 159, 166,241 Detection of infrared radiation emitted
Distance between from the object 12
Source/detector 19,24,35,51,55,79,86, Ellipse 55, 133, 134
100,149,156,164 Elliptical
Source/sample 81, 188,202,203 Pattern 25,96,155,167
Distortions Shape 182,250
Energy 187,213 Polarized wave 133,151,169
Fluid 28 Emerging technique 169
Moisture content 13,41,64-66,70, 158, Embolism 49, 90
177 End use capabilities
Power 212 Energy flux 45, 182, 183, 185, 189, 191, 193,
Ray 203,204 196,197,213
Relaxation time 146 Engines 28
Ultrasonic velocity 202 Environmental conditions 287
Domain Equivalent medium 3
Frequency 24,151,185,187,275 Evanescent wave 189
Time 24,185
Space 149, 187 F
Drying Failure 92, 113, 120, 121 see rupture
Board 266 Fan-beam 13,24,28,29,31,33,42,43,52,55,
Conditions 65 70,188,213
Control 63 Fourier
Conventional 142 (FFT) 10,24,84,98,149,166,181,186,
Defect 65 187,189,227,237,261,262,275
Dynamic process 20,66,118,145,173, Diffraction theorem 186, 187
244,267 Heat flow 78
Early stage 65 Slice theorem 181
Kiln 64 Fatigue see cycling loading
Kinetics 267,270, 27l, 292, 298 Fermat principle 182
Lumber 63,72,125,169,245,259-261,267 Fertilization 43
Moisture content 64,65,240,241 Fiberboard 157,170,173,175-177
Pattern 142 Fiber saturation point 82,83,125,137,141,
Rate curve 65 142,175,240
Small clear specimens 261-263 Fibril 5,94
Stress 63 Field
Time 63,266 Distribution 125
Gradient 221
E Field of view (FOV) 86
Earlywood and latewood 25,27,36,49,65, Filtering operation 189
99, 189, 191,208,210,241,252,254,255, Fine art objects 70
262,263,265,287 Finite elements 98
Subject Index 327
Fixed equipment 29-32,83-88,153-158, Heat

189-200,282 Absorption 95
Flakeboard 157,170 Camera 115
Forest: Dissipation 99, 102
Environment 9,90,159 Flux 82,87, 110, 115, 121
Fire surveillance 89 Propagation 122
Frost pocket 90 System 94-97,104,122
Mapping 90 Source 91,93,96-98,106,113,115,122
Fractal mechanics 4 Specific 10 1
Free induction decay (FID) 227-229 Transfer 6
Free water 82,126,135,172,173,175,221, Procedure 76-121
222,266 Heartwood 52
Frequency Heating-up thermography 113, 121, 122
Angular 79,131,132 Hemicelloloses 22, 23, 34, 220, 268, 275, 282
Domain 7,8,10,24,151,185-157,218 Heterogeneous 25,33,71,110,120,179,181
Larmor 217,218,221,231,277 Hydrolysis of polysaccharides 69
Magnetization 225 Hierarchical structure of wood 5
Microwave 131,132,136,140,144,145, High atomic number 13
149,153,155-161,166,170,172,173 High frequency thermal waves 79
Modulated 89, 92 High power heat gun 75
Nuclei resonance 217 Hilbert transforms 203
Precession 217,233,278 Historic musical instruments 70
Radial 79 Holography with microwaves 148
Radio 94,95,170,218,225,237-239, Homogeneous electromagnetic field
270 Hot liquids 75, 122
Resonance frequency method 119 Hounsfield number 19,28
NMR 217,220,228,246-248,237,273,277, Human head 28
278 Hysteresis 93, 120
Spectrum 166,185,213
Temperature 145, 146
Thermal waves 79 Ice nucleation 90
Ultrasonic 189, 191; 193, 196, 198-200, Illumination with radar 161
203,212,214 Image reconstruction with
Fresnel zone 202-204 Microwaves 125-178
Neutron 281-297
G NMR 281-297
Gamma Photographic 90,91,190,286,205,207
Radiation 18,22,28,29,41,55,67,70,71 Thermal waves 75-123
Scanning densitometry 67 Ultrasonic 181-214
General concept of NDT testing 2-4 X ray 13-70
Geometry of the specimen 16,20,21,29,55, Image display software 28
59,79,80,115,118,119,159,161,224, Immersion technique 189,191,192,195,208,
252,253,261 210,212
Geomorphology 41 Impregnation 272
Germinability of seeds 245 Improvement 6,54,57,71,91,93,117,222,
Grain orientation see slope of grain 237
Gray scale 14,25, 116 Incandescent lamps 95,113
Grids 49,61 Inclusions 159
Growth rings see annual ring Ions
Charged 126
H Degraded 276
Halogen lamps 87,115 Disolved 172, 173
Heartwood 43,52-54,215,232,241,264,287, Magnetic 225
289,291 Migration 127
328 Subject Index
Infrared Lignin 22,23,34,35,69,125,146,147,220,

Birefrigence 96 232256-259,268,276
Camera 85,113,115,120 Limitation 30,31,152,179,200
Detector 102, 120, 122 Experimental 30,31,152,179,187,188,
Imaging of specimens 108,110,111,112, 200,2l3
116,120 Mathematical 187
Local infrared emission coefficient 78 Liquid crystals 77, 122
Measuring device 106 Lock-in thermography 85-88,96,117,119,
Quartz lamps 94 121,122
Radiation 115 Lock-in vibro-thermal method 88,89
Radiation 123 Log 54-59,164-168,245,250
Reflectance spectroscopy 118, 119 Longitudinal
Sensor 106, 113 Attenuation coefficient X rays 38, 39
Wave length 122 Axis 27,32,38,39
Interfaces 80,126,153,156,170,196,197 Direction 39,141,144-147,168,210,211,
Interfacial 126, l31 249
Inverse problem 9 Position 63
Inversion algorithm 186 Profile 25
Ionizing radiation l3-73 Relaxation time 268
Wave 181,182,197,213
Loss tangent 141-143 see tan 8
Joints for timber connection 120 LSQ 186
Lumber 60-67,120,169,170,208,245,250,
K 259-267
Kasa circle 177 Mechanical grading 169,170
Kinetics of drying of wood with neutron
radiography M
Klystron microwave generator 149 Machine vision 156
Knots with: Macrovoid distribution 67,68
X ray 26,56,57,62 Magnetic see NMR
Thermal waves 92-96 Magnetic ions 225 see ash ions
Microwaves 166-168,171,172 Magnitude 76, 78, see also amplitude
Ultrasound 192,193,209,210 Major elements (H, C, N, 0) 33
NMR 243,251 Map 70
Mapping 9-11,99,208-211
L Mass
Lamb waves 197,200,201,212 Attenuation coefficient 23
Laminated Fraction of C, 0, H, N 23
Timber, lumber 2, 1l3, 114, 120 Transfer 65
Veneer 1l3, 117 Materials
Lamps Characterization l3, 125
Infrared 95 Conpensatory 40
Halogen 87,115 CT number 40
Quartz 94 Inhomogeneous 70
Larmor frequency 217,218,221,231,256, Inorganic 272
277,278 Liquids 292
Laser 78-80,96-99,102,104,119,122 Plant 159
Latewood see earlywood Polar l31, 132 dielectric l31
Leaves 178 Selective heating 125
Light Standard attenuators 40
Absorption 122 Thermal conductivity 84
Irradiation 80,81 Vegetative 125
Noise 101 Maxwell theory 7
Source 76-80,115 Maxwellian spectrum 281
Subject Index 329
Mechanical MRI see magnetic resonance imaging 216,

Behavior 87 277
Characterization of wood 2, 3, 10
Device 41,42,55 N
Excitation 88 Near-field diffraction theory 79
Level 296 Needle 245
Loading 77,122 Neutron
Motion 30, 31 Attenuation coefficient 2, 282
Operations 29, 61 Imaging 281-297
Parameters 69,156 Radiography 293-295
Performance 113,114 Water distribution 294
Properties 3,4,7,61,69,76,77, 119, 122, New products using wood as a major raw
170,177,210,255 material 3
Quantum 217 Nitrogen 51,87
Shaker 89 NMR
Treatment 3 Imaging technique 237-244
Wave 277 Spin relaxation mechanism 259
Medical scanner 28,55,71,215,251,267, Spectroscopy 215,260
277 Noise of signals with
Medium density fiberboard 113 Microwave 149,169
Metabolism 51 NMR 237,242
Microdensitometric analysis 39,51 Ultrasound 186,198,213
Microtomography 46,47,48 X ray 20-22,35
Mobile components of the protons 228 Non-contact 173,191,267
Mode conversion 181 Non-contact ultrasonic scanning 194-200
Model Nondestructive technique in industry 63,
Microwaves 125,132,144-147 71,113,169,189,199,221,270,279
NMR 222, 223, 225, 253, 258 Nondestructive testing of wood 1,2,4,6,41,
Thermal waves 82, 97 52,54,63,68,70,189,196,212,282,285,
Ultrasound 182,183,187,202 288,295
X ray 25,26,28,55,61,63,65-67 Nuclei, nuclear magnetic moment 216
Modulated energy 78 Numerical processing of wave front 148
Modulus of elasticity see Young's modulus Nyquist theorem 189,200
119
Modulus of elasticity with static methods o
61 Off-diagonal terms of stiffness matrix 208
Moisture content Operation 118,155, 189, 196
Microwaves 125-130,140-147,151-177 Optical densitometry 14,71
Neutron 281-298 Oriented fiakeboards 170
NMR 215,216,220-266 Orthotropic 128,200, 207
Thermal waves 84, 101-106 Overheating of the surface of the sample 76
X ray 18,20-25,41,49,50,52-54,61-69,
72 P
Molecular: Parallel see perpendicular
Macromolecular 135 Particle boards 2,84,117-119,158,170,199,
Motion 220,221 200,208,210,211,212,214 see wood
Structure 145 based composites
Weight 221 Pathological attacks 44,69,86,90,91,
Mono 205-207,243,249,251
Chromatic source 39 Peak amplitude 128, 185
Energetic photons 17 Penetrating radiation 27
Morphologic deterioration of cellular tissue Periodical 24,76,79,84,88,158,176,215
69 Permeability of wood 65, 145, 287
Morphology of defects 3, 59 Permittivity 132
330 Subject Index
Perpendicular 15,30-33,82,84, 130, 133, Polymerization 69

135,138,139,141,166,171,176,184,183, Polysaccharides 69
212,218,222,225,248,260,261 Polyvinyl 271
Phantom 28 Porosity 33, 39, 272
Phase Portable equipment X rays 32
Acquisition 76,87-90,102,118-122, Precession 217,218,221,226,227,278
132-137,148,149,156-158,166-171, Probes see also transducers
174-179,213 Projection data 19,28,35,44,47,55,60,70,
Angle 77,78,84,87,133 129,148,149,208,213
Coherence 219,221,226 in 3D 46,49,59,70,287,289
Image 77,84,117,119 in 2D 29,49, 59, 60, 63, 70
Shift 78, 79, 102, 133, 135, 138, 139, 150, in lD 262
151,185-187 Proton 224,262,267,278,283
Velocity 79, 169, 182 Pulse - ultrasonic
Photo acoustic effect 78, 278 Length 181, 193, 194
Photographic Energy 181,194
Film 14 Generator 198
Image 90,91 Pulsed thermography 99
Recording system 77
Photon 51-53,72 Q
Energy 19,22,34,38 Quality assessment with X rays 52-69
Phytopathology 41 Quality control 4-9,52-68,268-273
Physical Quantum energy 70
Limitation 188
Model 224, characterization 244 R
Parameters 69,82,94,127,156,168,175, Radar 10,148,155,160-166,178
225,239,255,272,279 Radial
Properties 28,69,71, 132, 137,245,274 Attenuation coefficient X rays 38, 39
Resolution 203 Direction 33, 38, 39, 55, 64, 65, 102, 103,
State 250, 251 108,109,141,145,147,152,168,193-195,
Physics of 208,210
Neutron scattering 281 Frequency 79
Diffusion waves 78 Loading 109, III
Magnetic resonance 215 Position 166
Wood 9 Variation of dielectric constant 159, 160
Phytopathologic diagnosis of ornamental Radon 9,24
trees 77 Random 54,67,126,217,222,225,227,242
Piezoelectric transducers see transducers Rate of heat application 77, 122
Pilodyne 7 Ray
Pith 26,27,45,159,160,240,244,247,251, Theory 182
252,286 Projection 213
Pixel 19-27,36,37,47,49,57,59,60,67 Path 16,33-35,39
Planck constant 217 Rayleigh waves 181
Polarization of Reaction wood 240
Microwaves 127,130-134,148,151,158, Receivers see also transducers
159,161,162,163,164,166,167,169,171, Reconstruction:
178, 179 Algorithm 24,25,185-189,241
Ultrasound 182-184 Redheart 251
Poisson's ratios 200, 209 Reference values for X-ray CT in 72,73
Poles 41-49,60,212 Remote sensing 148
Pollution 51-52 Representative
Polychromatic X ray source 40 Elementary volume 4
Polymers 40,76,88,89,189,221,268,270, Size of anatomic structure 36
271,272 Resin 68
Subject Index 331
Resolution Signal to noise ratio, see noise

Spatial 19,33,35,36,44,48,60,72,77,148, Characteristic signature 13, 14
156,200,202,283 Electronic 13
Contrast 35,36,43,59, 7l, 72 Full width at half-maximum (FVVHM) 43
Resonance frequency method see frequency Hilbert transform 203
Ring 27,39,45,46,54,55,61,69,99,215, Sinusoidal thermal wave excitation 87
261,262,267,275 see annual rings, Slices 16, 19,20,24,25,30,32,34,37,43,55,
growth ring 60,67,167,181
Rock 28,252 Slope of grain 125,134,137,150,151,155,
Rontgen densitometry see also 168, 169, 170
microdensiotometry 13 Slowness surface 182, 183, 184, 194,202
Roots 91,245 Small, clear specimens 40,206,214,261,262,
Rotation speed 52 264,281,285
Rupture Source of radiation 13, 15, 16, 18,20,22,24,
Modulus 61 28,29,30,31,35,40-43,55,57,61,63,70,
Phenomena 90,106,111,122 7l
137Cesium 18,39,43,241
S Half life time 43
Sampling of data with ultrasound 181,189, Spatial arrangement of basic constituents of
191 wood 3
SART 186 Distribution 129, 166
Sap flow 90,225,246, 247 Dependence of temperarture 96
Sapwood 20,43,49,51-53,215,229,232, Coherence 80
241,248,262,264,285,287,289,291 Diffusion gradient of diffusion waves 78,
Saw mill 54-59,155,170 80
Sawing pattern 52,54,57,59,61 Resolution of the image 19,3377,148,
Scanning with X rays 156,200,202,283
Beam 25 Spectral
Boards 63 Analysis 149,203
Dendroclimatological data 70 Lines 7,8
Device 55 Selectivity 80
Element 88 Response 87
Fan system 55 Peak 185
Frequency 25 Spin 215-279
Geometry 71 Alignment 221, 222
Growth rings in trees and poles 45 Density 232,278
Logs 61 Echo 232,233,278
Lumber drying 66 Population 221
Parameters 25,58,61,63,66 Quantum number 216
Quality assessment 52 sawmill 15,54 Relaxation time 220,222, 232, 278
Rate 87 Spinning motion 216,220
Tangential system 32, 33, 36, 38 System 218
Time 52 Square array 19
Scattering 187 Standard
Seismic borehole data 202 Adjustment 43
SIRT 186 Attenuators 40
Shear Carr Purcell spin echo pulse sequence
Loading 120, 121 270
VVaves 120,121,181-183,197 Cross-polarization sequence 257
Ship: Deviation 24,59, 135, 158,254
"La Trinidad Valencera", 275 Error 64
Simulation 58,61,98 Methods 68,237
Singular value decomposition (SVD) Multivariate models 61
186 NMR parameters 278
332 Subject Index
Partial least squares regression 61 Effusivity 101, 102, 104

Proton relaxation time 268 Gradient 77,78, 113
Radiation pulse shaping 14 Infrared emission 78
Scattering technique 79 Neutrons 281-289
Spectrometers 215 Radiation 120, 122
Static Shock 77,122
Bending 93 Wave length 79
Compression 93 Waves 75-80,83,87-89,106,116,121,122
Statistical approach 14,25,59,60,71 Temperature
Regression 23,24, 133, 135, 137, 145, Air 95,102
230-236,255 Dielectric phenomena 130,131,140,146,
Partial least squares 91, 174 155,159,170,174-178
Correlation 51,61,135138,140,154,167, Distribution 76,78,87,92,95,98,103,109,
234-236 112-115,121-123,295
Stiffnesses 63 see elastic constants Field 75,91,97,122
Stress Frequency 127,145
Bending 7 Modulation 78, 79, 84, 96
Concentration 120, strain curve 120 Pattern 120
Cyclic loading 122 Resolution 115
Distribution 1l0, ll, 112 Rise 93-96,102,104,109,113,120
Drying 65,288,291 Room 281,298
Failure 113, 114 Surface 75,76,85,86,90,97,101,106, 1l0,
Grading 7,91,93 115
Mechanical 106,108 Thermoelastic coupling effect 120
Periodical 88 Thermographic methods 75-121
Pollution 52 Equipment 83
Rating 6 Inspection of trees 90
Shear stress 121 Thermo-vision system 122
Thermal 122 Timber 91
Wave 185 Time
Structural Computation 46,213
Elements 293-297 Domain 24,185
Lumber 1 Flight 213
Sub-surface defect 77, 115-117 Of exposure 25,61
Surface of the sample 74, 122 Real 169,173,179
Roughness 196 Relaxation 146,242-273
Survival prognosis of tree 52 Tomographic reconstruction with X ray 13,
Synthetic aperture radar (SAR) 161 29,44, 51, 53,65
Tomography:, ionozing computed 13-73
T Toys 40
Tan 8 129, 131-133 see loss tangent Transducers
Tangential Source and detector for X-ray tomography
Direction 31,32,38,39,55,64,65,102, 10,28-32
103,108,141,145,147,152,168,170,182, For ultrasonic measurements 181,
292 189-199,214
Attenuation coefficients of X rays 38, 39 Treatment:
Temperature modulation 84 Chemical 3,272,274
Tension test 89,92, 113, 120 Mechanical 3
Theory of elasticity Trees 41-49,52,90,202-207,245,246,
Thermal: 285-287
Conductivity 76,78,84, 102
Conductivity of wood 84 U
Contact 77,86,106,110,118,122 Ultrasonic
Diffusion length 79 Beam propagation 213
Diffusivity 79,84, 102 Coupling medium 192
Subject Index 333
Elastic characterization 1 VVave

Energy 182-185,189,192,194,197,213 Forme 89, 185
Equipment 189-199 Length 8,9,11,39,75,76,79, 187,277,
Images 10,11,208,210,212,213 281,282
Method 7,9,181-214 Number 188
Polarization 182, 183, 184 Vector 281
Slowness 182-184,194,202 VVelded steel bridge 28
Stiffness 208, 209 VVet core 41,43
Transducers 208,214 VVidth of rings' see annual ring, growth
Velocity 189, 208 rings
VVave 10,183,213 VVindow connections 295-297
VVindow 10 VVood
Ultraviolet region 8 Based composites 67-69,170-177,208,
Uncertainty of measurement 2, 185,202 see 268-270
resolution Impregnation with preservative substances
272,274,279
V VVood properties
Varnish 79 Acoustical 182, 184
Veneer 54,80,102-107,113,117-119, Aesthetical 2
209-211,214,271 Chemical 2, 244, 279
Virtual image 24 Mechanical 3,4,7,61,69,76,77,119,122,
Viscoelastic 119 170,177,210,255
Visible 46,56,166,208,209,244,245,271, Physical 137-140
275,289,292,295 Thermal conductivity 84
Visual inspection 14,32,37,52,57,59,272 VVood based composites 67-69,170-177,
Visualization of water distribution with 208,268-270
neutron radiography 289-298 VVood species 27,33,39,52,72,94,131,136,
Voids 25,63,67,77,208,213,214,252,262, 225,254,260,285
271
Voxel 17,19,24,32,36,37
X
X-ray tube 10,22,25,29,32,44,47,60
VV
X-rays
VVater
Beam hardening 22,39,40, 71, 72
Bound and Free 126,135,172-175,
Beam path 33-35
215-279
Energy 33,61,70-72
Drought 49,51
Intensity 15,17,19,22,29,32,36,37,57,
Embolized zone 49
71,72
Human bone 20
Monochromatic 39
Lumina 14 see cell lumen
Parallel beams 24, 28, 38, 49
Microwaves 125, 126, 135, 142, 159,
Projection 19,21,24,25,28,35,44,46,55,
170-175,178
60,70
Monomolecular layer 279
Propagation 17
Neutron 11,281,285-288,291-294
Sources 18-20,22,24,28-31,35,39,
NMR 222,232,234,246,248,261,262,266
40-43,49,51,55,57,61,63,67,70,71
Soil 51
Tomography 13-70
Thermal waves 82,84,101,104
Transit in xylem 49
Transport Y
Ultrasound 189, 191, 197,208,212 Young's moduli 114,200 see elastic
Xray 28,43,49,52,69 constants
List of Notations
A thermal wavelength
e* complex dielectric constant
ro angular frequency of the intensity modulated laser heat source
o chemical shift
p density
't recovery time
(J screening constant
a thermal diffusivity
~<I> variation of phase of the radio frequency wave
e' and e" components of the complex dielectric constant (real and
imaginary)
~A variation of the amplitude of the radio-frequency wave
canopy equivalent dielectric constant
complex permittivity of dielectric material
free space perimittivity
Larmor frequency
phase constant of the air
wavelength of the electromagnetic wave in air
complex permittivity of water
attenuation coefficient of the ionizing radiation beam
thermal diffusion length
linear attenuation coefficient of X-rays for the x voxel in the cross-
sectional slice
!lw linear attenuation coefficient of X-rays in water
J.L' mass attenuation coefficient
J.Lc attenuation coefficient due to Compton effect
J.LP attenuation coefficient due to photo-electric effect
a cell lumen size (J.Lm) with thermal imaging
Bo static magnetic field
C specific heat of the solid wood
C image contrast
c speed of propagation of the radiation
C*(ro) complex capacitance
Cll> C22 , C33 ultrasonic stiffnesses on axes 1,2 and 3
d sample thickness (mm) with thermal imaging
DandL defect depth and length with thermal imaging
e thermal effusivity
336 List of Notations
EPV energy/pulse length of ultrasonic wave

EV pulse length energy value
f frequency
FID free induction decay
G wood-specific gravity
G conductance of the sample
H Hounsfield, the unit used for X-ray imaging
h Plank constant
1 intensity of transmitted ionizing radiation beam through the
sample
10 intensity of transmitted ionizing radiation beam through the air
k thermal conductivity
kl> k 2 , k3 coefficients of dielectric anisotropy
ka thermal conductivity of the dead air
kL wood thermal conductivity in L direction
k Lw thermal conductivity of cellular wall in longitudinal direction
kT wood thermal conductivity in transverse direction
L,R,T anisotropic axes of wood related to longitudinal, radial and
tangential directions versus the annual ring
M total number of projections
MC wood moisture content (%)
MC~w absolute moisture content deduced from microwave technique
md weight of wood on the basis of absolute dry fraction
Mo magnetization momentum
mw weight of water
Mz and Mxy components of magnetization momentum
N pixel size
p lIv = slowness
P power of the electric field dissipated in the specimen as heat
P(T) temperature-dependent heat losses
Q power flux absorbed at the front surface
Rp Fresnel zone
T absolute temperature (K)
t time
TJ spin-lattice relaxation time or longitudinal relaxation time
T2 transverse relaxation time
TOF time of flight of ultrasonic wave
V phase velocity of thermal wave with which the temperature mod-
ulation moves along z direction
v ultrasonic velocity
va porosity of wood
Vrn fraction volume of water
W width of the pixel
Color Plates 337
b)
24
o
36--- -- ----
42
48
54
60
66
72
78
84
90
96 24
30
6
-----------
42
48
>466
60 72
790
84 96
Fig.2.17a,b. Tomogram on a color scale of a transverse section of a healthy (a) and of a decayed
(b) Norway spruce tree. (Habermehl and Ridder 1996, with permission)
a)
M..
... • .• ..W
...
--
• .W ~ ..
.. .M.. • .•
Fig.2.23a,b. Tomograms of Quercus petraea and Quercus cerris (Tognetti et al. 1996). a Quercus
--
......
cerris; b Quercus petraea. N, E, S, Ware the cardinal points; 1,2,3,4 are the position of measured
points The density scale in CT numbers is represented by different colors (i.e., CT = 30 corre-
sponds to dry condition and CT = 96 corresponds to high moisture content). (with permission)
338 Color Plates
24.00
30.00
36.00
-
42.00
48.00
54.00
60.00
66.00
72.00
-
78.00
84.00
90.00
96.00
24.00
30.00
36.00
42.00
48.00 --
54.00
60.00
66.00
72.00
78.00
84.00
Fig.2.25a-d. The development of sapwood in Scots pine induced by ammonia pollution, com-
90.00
pared with a normal tree at 100cm stem height. Trees are located in a forest from the district
96 . 00
Torgelow, Germany (Katzel et al. 1997, with permission). A Tomogram for tree no. 1, with 624-
cm' cross section located at site 1,200 m from the farm (maximum pollution zone). B Tree no. 8,
with 745-cm' cross section located at site 2, 280m from the farm. C Tree no. 2 at site 3, with 513-
cm' cross section located 2900m from the farm (minimum pollution zone). D Tree no. 8 at site 3
located 2900m from the farm (minimum pollution zone). The density scale in CT numbers is
represented by different colors, CT = 30 corresponds to dry conditions and CT = 96 corresponds
to a high moisture content. The tomograms were taken in September 1993 and 1994
a b c
36~
36 36_ - 85I)oil5O
1 -490-S40
30 _~ .• 540-590
30 _~ 7~
.. &50-?50
30_ . 1~"00 .300-440
24 i·~~
24 .,QOO-'QM 24i ~= 0_
i • 950-1000
~ 0350-450
18 18 18 5
5
12=
:2! ~2!
6 !
0 0 0
0 10 20 30 40 50 60 70 eo 90 100 0 10 20 30 40 50 60 70 eo 90 100 0 10 20 30 40 50 60 70 80 90 100
d Board width (nvn) e Board width (nvn) r Board width (mm)
: 36~ 36~
36 8().ll1O ~ . '4».1 20
I ·, 30 ~ . 15-18
30_ :~:~: 30 - •. 8().100
eo-eo - . '2·\$
: 24i! ~::: 24 ~ •• ~12
6·9
241 ~ 0 0·20 o 3 ·s
18 . 18 . 18 .
S :;
:2! :2! ~21

0 '0 0
0 10 20 30 40 50 60 70 60 90 100 0 10 20 30 40 50 60 70 eo 90 100 0 10 20 30 40 50 60 70 80 90 100
g Board width (nvn) b Board width (mm) i Board width (mm)
Fig.2.32a-i. Dynamics of wood drying (Pang and Wiberg 1998). Experimental tomograms (with permission). a Scanning of the board before drying -
corresponding to wet saturated wood. b Scanning of the board before drying - corresponding to wet wood after 9.6 hours of drying, and to the early stage
of the drying process. c Scanning of the board before drying - corresponding to wet wood after 3004 h of drying and to the late stage of the drying process.
Predicted images for density distribution. d Wet wood density before drying. e Wood density after 9.6 h of drying. f Wood density after 30.4 h of drying.
<J.>
Predicted images for moisture content distribution. g Model of predicted wet wood density before drying. h Model of predicted wet wood density after <J.>
-c
9.6h of drying. i Model of predicted wood density after 30Ah of drying
340 Color Plates
Fig. 3.5. Infrared device for the detection of

cavities in trees. (Catena and Catena 2000,
with permission)
Fig.3.11a, b. Cavities in standing trees. a photographic image. b thermographic image of cavity

(in blue) in trunk and in branches. (Catena and Catena 2000, with permission)
Color Plates 341
(unit: mm)
Fig. 3.17. Thermographic measurements of the slope of grain. Finite element model. The heat
flux is represented by the vertical arrows in the central region. (Naito et al. 2000; courtesy of the
Japan Wood Research Society).
Fig. 3.18. Temperature distribution as a function of heating and slope of grain deduced with the
finite element method time obtained with the finite element method. (Naito et al. 2000; courtesy
342 Color Plates
0
Density (g/cm 3)
2
1.3000 - 1.4000
1.2000 - 1.3000
4 1.1 000 - 1.2000
-
1.0000 - 1.1000
6 0.9000 - 1.0000
0.8000 - 0.9000
8 0.7000 - 0.8000
0.6000 - 0.7000
10 0.5000 - 0.6000
0.4000 - 0.5000
12 0.3000 - 0.4000
0.2000 - 0.3000
14 0.1 000 - 0.2000
0-0.1000
0 2 4 6 8 10 12 14
a) x (mm)
3.0
Xylem vessels
2.5
Transmission
2.0 _ 567 - 600
535 - 567
1.5 502 - 535
470 - 502
1.0 437 - 470
405 - 437
0.5
372 - 405
_ 340-372
0 _ 308 - 340
0 2 3 4 5
b) _ 275 -308
x (mm)
Fig.3.20a,b. Density mapping of a transverse section of two species: a beech and b balsa.
The size of the specimen was 14 x 14 x 1.7mm. (Koch et a1.l998; courtesy of Wood Science
Technology)
Color Plates 343
longitudinal radial tangential

Fig. 3.28a, b. Infrared imaging of pine specimens under compression. a Specimens with major
axis in longitudinal, radial and tangential directions. b Localization of intrinsic dissipation in
wood specimens as a function of anisotropy (each color hue corresponds to 0.2°C) in longitudi-
nal, radial and tangential directions. (Luong 1996, with permission)
344 Color Plates
Fig.3.30a-c. Infrared imaging of specimens under compression

loading related to stress distribution in the longitudinal direction. a
specimen; b load strain relationship; c thermographic images corre-
sponding to the points a, b, c, d, e, f from b. (Okumura et al. 1996;
courtesy of Wood Research Institute, Kyoto University)
Spec,"*, I*IaId
a)
b) • Amount of comprMllon (mm)

,
Compression
load
t
• • • • •
• • •
· ..... ~.-
CH1 . ....
. •••
· . .. ..
Fig.3.31. Infrared imaging
of specimens under
compression loading
related to the stress
distribution in the
transverse plane. Images of
specimens at rupture
under different
compression loads applied
in radial (CH4) tangential (j
( CH1) and inclined o
directions vs. the annual . - ~ ...
...0-
-- ." '1:1
rings (CH3 inclined vs. T
~-• .!'::~..!-
and CH2 inclined vs. R). .. .......... '"
(Okumura et al. 1996;
CH3 ,..
.... ..- . .
courtesy of Wood Research
Institute, Kyoto University)
... *
VJ
""-
VI
Hole of 5 mm <.;J
diameter ~
(')
o
0"
....
:s
280 ~
'"
a) 300
150
g100
~ 50
5 • 'ir¥ .
4~ ....,...
~;j~: .'.'
~<""'''''.' ;.'.~
'..,'.!~,; ~
o
o 5 10 15 20 ;~~:~?? ~~ ~.,.>:. )'
b) OefIecUon(rrm) c)
Fig.3.32a-c. Infrared imaging of specimen under static bending loading. a Specimen with a central hole under static bending. b Stress strain
curve. c Temperature distribution as a function of loading and strain, corresponding to points 1, 2, 3, and 4. (Naito et a1.1998; courtesy of Wood
Research Institute Kyoto University)
Color Plates 347
Fig. 3.34. The thermographic image

for a specimen with two 16-cm
delamination zones. (Masuda and
Takahashi 2000, with permission)
Dome cr
I
\
~,
~
.'
,,"",
.'
I .
, ,
,
~, (mm)
10 .' .,
I
.,
.'
,
~
.'
, ., ,
.'
I
,
.'
I
t, I. ,
• \ .,
,- ,. , ,.,
'.' .' '.' .' '\, \' \ j,
1:' ,:1
:
':1
:
I:'
::
I .. ,
::
4 .'
~ \ \,;
~. .
::
a)
2
b) '- ... .\ _.6." II.
,
0 mm 2 mm
Fig.3.38a, b. Detection of artificially induced
defects in chipboard through the surface veneer
sheets of various thicknesses. a Geometry of
the specimen. b Thermal image. (Wu and Busse
a) b)
Fig.3.39a, b. Detection of knots present in solid wood covered with veneer sheets. a Geometry
of the specimen. b Thermal image. (Wu and Busse 1995, with permission)
348 Color Plates
Fig.3.41a-c. Infrared images of splice joints under shear loading. a Before loading. b Shear stress
before failure. Color hue corresponds to 0.2°C. (Luong 1996, with permission)
Color Plates 349
Fig.4.30. Distribution of knots on the transverse section

of the bolt. (Kaestner and Baath 2000, with permission)
Fig.4.31a-d. Iso-surface on a topographic slice. a Ipi The amplitude; b ythe argument (y, arg p).
c e The ellipticity angle which gives the phase difference between two wave components and
describes the degree of elliptical polarization. d r The angle between the two components of
the wave, corresponding to the tilt of the linear polarization. (Kaestner and Baath 2000, with
permission)
350 Color Plates
Fig.4.33a,b. Images of knots reconstructed with a the polarization ratio; b the phase difference
between the two components of the electric field, parallel and perpendicular to the fibers.
(Kaestner and Baath 2000, with permission)
Color Plates 351
Polan ation
along e
(transverse wave)
Polarisation
along r
(longitudinal wave)
Polari ation
along '"
(transverse wave)
b)
Oak
0.5
-0.5
-1
Fig. 5.2a-c. Three-dimensional

representation of acoustic
properties of oak (c). a Local
basis and color code. b
Variation of polarization angle
on slowness surface for oak.
(Bucur et al. 2001, with
c) permission)
2.50
2.00
1.50
1.00
0.50
[mm]
Fig. 5.5. High resolution images of the transverse section of a tree (Platanus acerifolia)
(Martinis 2002, with permission)
Fig. 5.16. The complex shape of

the transverse section of the trunk
of hackberry (Celtis australis)
(Comino et al. 2000, with permis-
sion)
km/s
600 2.00
1.80
500
1.60
400 1.40
E 1.20
§. 300
1.00
200 0.80
0.60 Fig. 5.17. Tomographic
100 images of the transverse
0.40 section of the trunk in Fig.
0.20 5.16 obtained with 120
100 200 300 400 500 600 independent velocity
measurements. (Comino et
[mm] al. 2000, with permission)
Color Plates 353
Fig.5.1S. Reconstructed image

with velocity values measured
on cubic specimens and posi-
tions of the specimens in the
trunk section. Above Recon-
structed tomographic image.
Below Cubic specimens.
(Comino et al. 2000, with per-
mission)
a)
600
1.80
SOD
1.60
400 1.40
E 1.20
~ 300
1.00
200
0.80
100 0.60
0.40
100 200 300 400 500 600
b) [mm]
2
1.8
I
1.S
1.4
1.2
~
~
0
J 0 .8
0.8
0.4
sample labels
Fig. 5.19. Measured ultrasonic velocities and calculated velocities for three orthotropic direc-
tions, L, Rand T on selected specimens shown in Fig. 5.1S. (Socco et al. 2000, with
permission)
354 Color Plates
a b c
Fig.7.3a-e. Photographic images

and neutron radiography of dif-
ferent species. a Pinus thunbergii.
b Metasequoia glyptostroboides. c
Chamecyparis obtusa. d Quercus
serata. e Robinia pseudoacacia.
(Nakanishi and Watanabe 1995,
d with permission)
Fig. 7.4. Photographic images (A, B,

C, D) and corresponding neutron
images (a, b, c, d) of different cultivars
of sugi (Cryptomeria japonica). A 24-
year-old cultivar, 25 Gou. B 25-year-
old cultivar Honjiro. C 29-year-old
cultivar I-Gou. D 30-year-old cultivar,
Sanbusugi. (Nakanishi et aJ. 1998b,
with permission)

Nondestructive Characterization and Imaging of Wood

Uploaded by

Copyright:

Available Formats

Nondestructive Characterization and Imaging of Wood

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nondestructive Characterization and Imaging of Wood

Uploaded by

Copyright:

Available Formats

Springer Series in Wood Science

Springer-Verlag Berlin Heidelberg GmbH

With 201 Figures, 24 of them also in color and 49 Tables

Library of Congress Cataloging-in-Publication Data

1. Wood-Testing. 2. Non-destructive testing. 1. Title. II. Series.

Typesetting: SNP Best-set Typesetter Ltd., Hong Kong

This book on the Nondestructive Characterization and Imaging of Wood by

Baltimore, December 2002

In writing this book, I have attempted to provide a comprehensive account of

Niemz, Zurich, Switzerland, Dr. Patrick Rasolofosaon, Paris-Malmaison, France,

2 Ionizing Radiation Computed Tomography .............. . 13

2.4.3.2 Control of Wood -Based Composites .................... . 67

3 Thermal Imaging ................................... . 75

4 Microwave Imaging ................................. . 125

4.4.2 Microwaves for Internal Inspection of Logs ............... . 164

5 Ultrasonic Imaging .................................. . 181

6 Nuclear Magnetic Resonance .......................... . 215

7 Neutron Imaging 281

References ................................. . . . . . . . . . . . . . . . 299

Subject Index ............................................. 323

List of Notations ..................... . . . . . . . . . . . . . . . . . . . . . . 335

Color Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337

1.1 Brief Historical Review of Nondestructive Evaluation

Nondestructive evaluation of the physical properties of wood has its origin in

1.2 General Concepts of Nondestructive Testing of Wood

nondestructive inspection of wood-based composites and their components

mechanics of heterogeneous media require the definition of the representative

1.3 Classification of Nondestructive Techniques for Wood

The characterization of wood properties is critical for the understanding of

material parameters at nearly any point, line, surface or volume element of

Material properties Preprocess Process Product Field

X indicates the need for nondestructive characterization

Wood parameter Properties measured Techniques

Moisture content Dielectric properties Impedance DC and high frequency,

This book is an attempt to comprehensively review numerous aspects of

~l Fig. 1.2. Characteristics of the electromagnetic

X-rays, gamma rays

where c is the speed of propagation of the radiation.

Imaging method with Wavelength (m) Scale of wood structure

X-ray 10- 12 ••• 10-9 Submicroscopic, microscopic and macroscopic

1.4 Imaging of the Internal Structure of Wood

The nondestructive imaging of wood structure is based on different methods

different measured parameters or can be obtained by a tomographic tech-

1.5 Summary and Outline of the Book

Computed tomography scanning with ionizing radiation provides three

volume of material can be inspected quickly and implementation of scanning

2.2 Basic Phenomena

The basic phenomena involved in X-ray reconstitutive tomography is illus-

volume elements "voxels"

transmission equipment". The term "absorption coefficient", often used in the

2.2.1 X-Ray Propagation in Solids

The attenuation coefficient /l depends on the mass density of the sample

1/10= e 41Pt (2.4)

and the corresponding logarithm is:

Coefficient /1' Energy source Moisture content Reference Differences

0.191 Am24! 8 Moschler and Dougal (l988)

The attenuation coefficient in inhomogeneous solids is not constant and

CTnumber = (/-Ls - /-Lw) x 1000 (2.10)

2.2.2 Attenuation and Profile of the Inspected Solids

-600 .,:::;_...L.._...L..._-L-_--l_---I_---'L--_L-_-'--_...1-_...L..._-L-_--l_ _L..-