Abstract
The aim of this contribution is to present new high order neural network structure for effective recognition of human faces in gray scale irrespective to their position, orientation and scale. The new Pi-Sigma-Pi network is superior to previous high order based approaches. Invariances can be easily incorporated into its structure (there is no need to learn them and the learning phase is performed with only one view of each object). It has small number of adjustable weights, rapid learning convergence and excellent generalization properties. Methods for reduction of equivalence classes are described, which make future hardware realization more feasible (few interconnections) and reduce significantly sensitivity of the recognition system to difficulties arising from the fact, that considered transformations are not realizable on a square grid. Many experimental results are described, in which proposed approach has shown to be very effective giving high accuracy (100% for 5 objects and over 95% for 20 different human faces). This framework is compared with previous high order based approaches which are able to give proper responses only with a small number of simple binary objects.
This work was in part supported by the Polish Science Foundation (FNP)
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
R. Foltyniewicz and S. Skoneczny. An Improved High Order Neural Network for Invariant Recognition of Human Faces in Gray Scale. Proceedings of the WCNN, I:587–592, July 1994.
Kunihiko Fukushima. A neural network model for selective attention in visual pattern recognition. Biological Cybernetics, 55:5–15, 1986.
Kunihiko Fukushima, Sei Miyake, and Takayuki Ito. Neocognitron: A neural network model for a mechanism of visual pattern recognition. IEEE Transactions on Systems, Man, and Cybernetics, SMC-13(5):826–834, September/October 1983.
C.L. Giles and T. Maxwell. Learning, Invariances, and Generalization in High-Order Neural Networks. Appl. Opt., 26:4972–4978, 1987.
T.L. Hemminger and C.A. Raez. Using a Hopfield Network for Rotation and Scale Independent Pattern Recognition. In Proceedings of IEEE ICNN, pages 3003–3008, 1994.
A. Khotanzad and Y.H. Hong. Invariant Image Reprezentation by Zernike Moments. IEEE Trans. on PAMI, 12(5):489–497, May 1990.
A. Khotanzad and J.H. Lu. Distortion Invariant Character Recognition by a Multi-Layer Perceptron and Back-Propagation Learning. IEEE ICNN, I:625–632, July 24–27 1988. San Diego, California.
S. Perantonis and P. Lisboa. Translation, Rotation, and Scale Invariant Pattern Recognition by High-Order Neural Networks and Moment Classifiers. IEEE Trans. on NN, 3(2):241–251, 1992.
C. Pernell and M. Acheroy. Face Recognition with Artificial Neural Networks. IMACS Annals on Computing and Applied Mathematics Proceedings MIM-S 2, 3–7 September 1990. Brussels, Belgium.
D.E. Rumelhart and J.L. McClelland. PDP. MIT Press, 1986.
Y. Shin and J. Ghosh. The pi-sigma network: An efficient higher-order neural network for pattern classification and function approximation. International Conf. on Neural Networks, Seattle, I:13–18, 1991.
L. Spirkovska and B. Reid. Coarse-Coded Higher-Order Neural Networks for PSRI Object Recognition. IEEE Transactions on NN, 4(2):276–283, March 1993.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Foltyniewicz, R. (1995). Efficient high order neural network for rotation, translation and distance invariant recognition of gray scale images. In: Hlaváč, V., Šára, R. (eds) Computer Analysis of Images and Patterns. CAIP 1995. Lecture Notes in Computer Science, vol 970. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-60268-2_325
Download citation
DOI: https://doi.org/10.1007/3-540-60268-2_325
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-60268-2
Online ISBN: 978-3-540-44781-8
eBook Packages: Springer Book Archive