BG61354B1 - Device for document cataloguing - Google Patents

Abstract

The device is designed for reading , processing and cataloguing black and white and multicolour documents of standard sizes such as bank cheques , invoices, pools cards and other based on the preset symptoms. It has improved reliability and fast operation. The device contains document carrier (1) fitted on the focus plane of an optical system (2) to which an image sensor (3) is connected to the output of which a former (4), compression unit (6), data base unit (7) and control unit (8) are connected in series. An image processing unit is also connected to the output of former (4) containing buffer memory (9) to the out put of which a first (10) and second (11) image zoning unit are connected in parallel. To the unit (10) output a formalized image generator (12) unit for bringing it back to the original (13) and a comparison unit (14) are connected in series. The output of the first unit (10) is also connected to the second input of the comparison unit (14). To the output of the second unit (11) a generator of zone symptoms (16), a second unit of the compression unit (6) and a unit for adaptive zone binarization (15) are connected in parallel, to the output of which a third input of the compression unit (6) and the first input of the image recognition unit (17) to the second input of which the output of the generator of zone symptoms (16)is connected. The output of the image recognition unit (17) is connected to a fourth input of the compression unit (6). The output of the control unit (8) is connected to the second inputs of the first and second unit for image zoning (10&11) and a data base unit (7). The output of the comparison unit(14) is connected to the third inputs of the first and second units for the image zoning (10&11). A lighting system (5) is mounted above the documents carrier (1).

Description

The invention relates to a device for document cataloging, which finds application in reading, processing and cataloging of various documents in a standardized form, such as: bank checks, borders, questionnaires, TOT-fiches and other prior art

A device / 1 / is intended for scanning documents produced in two substantially different colors, which comprises an image creation unit made up of a series-coupled optical sensor (lens), an SSD-linear image sensor, an amplifier, an analog- digital converter (ADC) and image processing unit.

The output of the image creation unit is coupled to an image processing unit that contains two identical, parallel connected, binaries, respectively, for the first and second threshold levels. The output of the first binarizing circuit is connected to the input of the first of three consecutively delayed lines, the outputs of all three being connected to the three inputs of a filter whose output is connected to video memory.

The output of the second binarizing circuit is connected to the input of the first of the other three sequentially connected delay lines, the outputs of each of which are respectively connected to the three inputs of an image loss block to which the fourth input is connected to the output of the second delay line from the first binarization scheme.

The output of the loss compensation unit is connected to an image recognition circuit.

If necessary, the known apparatus may also be provided with a lighting system intended to illuminate the document to be processed.

The disadvantages of the known device are the following: low reliability due to the fact that only two levels are analyzed in only three adjacent rows of the image; low speed, since in the case of a vaguely calculated image, it is necessary to re-scan with another threshold level to obtain new additional image data; limited application, as the device is intended for processing only documents executed in two, significantly different colors.

It is desirable to create a document cataloging device with increased reliability and speed, with universal application for reading and processing of monochrome and multicolored documents in a standardized form, providing the cataloging of documents by given or given signs.

SUMMARY OF THE INVENTION

The present invention relates to a document cataloging apparatus comprising a document carrier located in the focal plane of an optical system coupled to a sequentially coupled image sensor and shaper comprising a sequentially coupled video processor, ADC and video memory. An illumination system is mounted above the document carrier.

A compression block, a database block, and a control block are connected in series to the output of the shaper.

The output of the shaper is also connected to the input of an image processing unit which contains a buffer memory, to the output of which the first and second image zoning blocks are connected in parallel. The output of the first image zoning unit are sequentially connected by a formalized image generator, a block for bringing the formalized image to the original, and a comparison block, the output of the first image zoning unit being connected to a second input of the comparison block.

The output of the second image zoning unit are connected in parallel by an adaptive zone binarization unit, a zone feature generator and a second input of the compression unit.

The output of the adaptive zone binarization unit is connected to the third input of the compression unit and to the first input of the image recognition unit to which the output of a zone feature generator is connected to the second input, and the output of the image recognition unit is connected to the fourth input of the compression block.

The output of the control unit is connected to a second input of the first image zoning unit, to a second input of the second image zoning unit, and to a second input of the database block.

The output of the comparison block is connected to the third input of the first image zoning block and to the third input of the second image zoning block.

According to the present invention, the document cataloger is highly reliable since multiple brightness levels are analyzed for each image element (pixel); with increased speed because a formalized image or standard is used; with universal application, as it allows reading and processing of black-and-white and multicolored documents in standardized form, providing cataloging of documents by given or given signs.

DESCRIPTION OF THE FIGURES Attached. EXAMPLE FOR THE IMPLEMENTATION OF THE INVENTION

A document cataloging device intended for cataloging TOTO slips is shown in the figure showing its block diagram.

Document carrier 1 is located in the focal plane of the optical system 2 to which an image sensor 3 and a shaper 4 are connected in series, comprising a video processor, an analog-to-digital converter (ADC) and video memory in series. An illumination system 5 is mounted above the document carrier 1.

The image sensor 3 is a size matrix (X _d , Y _e ), where X and Y _e are determined by the number of sensing elements (pixels).

Compression unit 6, database unit 7 and control unit 8 are connected in series to the output of the shaper 4.

A buffer memory 9 is also connected to the output of the shaper 4, to which the first image zoning unit 10 and the second image zoning unit 11 are connected in parallel.

A formalized image generator 12, a block for bringing the formalized image to the original 13, and a comparison block 14 are connected in series to the output of block 10, with the output of block 10 also connected to a second input of the comparison block 14.

The adaptive zone binarization unit 15, the zone feature generator 16 and the second input of the compression unit 6 are connected in parallel to the output of the second block 11.

The output of the adaptive zone binarization unit 15 is connected to both the third input of the compression unit 6 and to the first input of the image recognition unit 17 to which the output of the zone feature generator 16 is connected to the second input.

The output of the pattern recognition unit 17 is connected to a fourth input of the compression block 6.

The output of control unit 8 is connected to a second input of block 10, to a second input of block 11, and to a second input of block 7.

The output of comparison block 14 is connected to the third input of block 10 and to the third input of block 11.

The compression unit 6 contains a diagram for transforming the three-dimensional (X, Y, Z) matrices obtained at the output of the second image zoning unit, by the coordinate Z, into two-dimensional (Xk, Yk) matrices that generally contain all image coordinates on the X, Y axes at least once and form the following types of filters:

- filters for areas that do not contain information (do not contain any area of interest) and cataloging factors;

- filters for areas with recognized images and cataloging factors;

- filters for areas with binarized (black and white or color) image and cataloging factors;

- Filters for halftone color areas and cataloging factors.

The database unit 7 contains a memory in which the recognized images and images with the corresponding catalog numbers, missed by the filters of the compression block 6, are recorded as a result of the missed catalog coefficients.

The control unit 8 contains the peripheral devices for acting on the unit 10 and on the unit 11 (buttons, keyboard, light pen, etc.).

The first image zoning unit 10 contains four filter groups connected in parallel:

- first group - to define areas with different image classes;

- second group - to improve the image quality in the respective areas;

- third group - to convert the images in the zones from one class to another;

- fourth group - to find characteristic objects, find coordinates and connections between them.

The Formalized Image Generator 12 contains a file cabinet of formalized image documents, arranged as matrices, with coordinates, coordinate ratios, geometric dimensions, and idealized TOTO-fiche images.

These matrices are supplemented, if necessary, by the control unit 8, resulting in the creation of new formalized images in the file cabinet of the formalized image generator 12, by entering coordinates of other rectangular zones and new standards (new games) in the first block for image zoning 10.

The unit for rendering the formalized image to the original 13 comprises linear and nonlinear filters and tables for transforming the coordinate systems so that the formalized (idealized) image is deformed and brought closest to the real one.

Comparison unit 14 includes a matching circuit to which the output of the first image zoning unit 10 is connected to the first input and the output of the formalized image rendering unit to the original 13 is connected to the second input.

The second image zoning unit 11 contains groups of filters constructed as three-dimensional matrices (X, Y, Z), for which the axes X and Y coincide with those of the reduced image, and the Z axis reflects interests: processing method and mode of directories. Some of these matrix filters are pre-built on the basis of the formalized images from which the file cabinet was created. This file cabinet can be supplemented in the process of operation by changing the parameters of some filters using the control unit 8.

The adaptive zone binarization unit 15 contains linear and nonlinear filters for converting homogeneous halftone portions of the image into binary ones, i.e. only those parts of the image (Xd, Ud, Zfl) for which Z from the matrix of zones (X, Y, Z) takes values zj that require binarization as a way of processing the image.

These linear and nonlinear filters for converting homogeneous halftone portions of an image into binary ones work by matching the brightness of each pixel with one threshold value defined for that part of the image in one of the following ways:

- measuring the distribution of brightness in a given area;

- measurement of the brightness gradient (contour finding);

- finding groups of pixels with close brightness values;

- estimation of the maximum difference between the brightness of an individual pixel and the average brightness in the treated area, etc.

The zone feature generator 16 contains a feature file and rules Q (q) for identifying objects that are grouped according to the selected recognition processing.

The pattern recognition unit 17 comprises a comparison circuit, to which the first input receives the binary image signal for a given zone received at the output of the adaptive zone binarization block 15, and to its second input receives a corresponding series of image recognition features .

APPLICATION OF THE INVENTION

The TOTO slip, hand-affixed, is affixed to the document carrier 1 located in the focal plane of the optical system 2.

The image sensor 3, a dimension matrix (Xd, Ud), where Xd, Υλ are determined by the number of sensing elements (pixels) along the two axes respectively, converts the optical signal into electrical.

Through the video processor of the shaper 4, which contains generators and circuits 5 to control the image sensor 3, the analog video information from each pixel enters the input of the analog-to-digital converter and, as a whole, is recorded on the Za axis in the video memory where it is formed 10 the matrix of the original image (Xd, Ud, 2d). Thus, the brightness of the black and white image of the slip is expressed in integers along the Z axis.

The image recorded in the video memory of the shaper 15 image 4 is analyzed and the image defects are eliminated, for example: to compensate for the irregularities introduced by the optical system 2, to eliminate the high-frequency noise from the destroyed or 20 defective pixels from the image sensor 3 by assigns an average of their neighboring pixels.

The halftone digital image signal 25 obtained at the output of the shaper 4 is fed to the input of sequentially connected compression units 6, database unit 7 and control unit 8.

The signal of the halftone digital image 30 received at the output of the shaper 4 is also fed to the input of the buffer memory 9, which is recorded and fed to the first input of the first image zoning unit 10 and the first input of the second zoning block. the image 11.

In the first image zoning unit 10, by means of parallel and sequentially coupled transmission function filters comprising proportional, differentiating and / or integrating units, trigonometric and logarithmic converters, adders, multipliers and limiters, it is determined:

- the distribution of brightness levels;

- maximum, minimum and average brightness levels;

- homogeneous portions of the image;

- intersecting lines;

- coordinates of: centers of contrasting point objects (game code) and points of intersection of lines at right angles (coordinates of the black frame of the fiche).

These coordinates and their relationships form the zones according to the four filter groups.

In accordance with the zones created by the first image zoning unit 10, the formalized image generator 12 recalls from its memory the TOTO-fiche matrix (Xf, Υφ), with the same zones, which, together with the zoned image, enters the input of the the block for bringing the formalized image to the original 13.

In the block for rendering the formalized image to the original 13, a matrix of coefficients for converting the coordinates of the formalized image (Xf, Υφ) to those of the original (Xd, Ya) is created. The result is a reduced formalized image matrix (Xn, Υπ), which, together with the original one (Xd, Ud), enters the two inputs of the comparison block 14 respectively.

The degree of compliance is preset and controlled by the comparison unit 14.

Upon reaching the predetermined degree of correspondence, the comparison unit 14 produces an output at its output that is prohibiting the first image zoning unit 10 and allowing the second image zoning unit 11.

The second block 11 forms a (X, Y, Z) matrix for which the axes X and Y coincide with those of the reduced image Xn and Υπ, obtained at the output of the block for bringing the formalized image to the original 13 and allowed by the comparison block 14, and along the Ζ axis, interests are reflected, with each value zj of Z forming a matrix (x, y, zj) contained in the matrix (X, Y, Z).

In this case, the following matrices are formed:

(x, y, O) no interest;

(x, y, a) binarization;

(x, y, b) printed text;

(x, y, c) handwritten characters;

(x, y, d) recognition of printed text;

(x, y, f) handwriting recognition;

(x, y, h) preservation;

(x, y, t) catalog feature;

. 61354 (x, y, 1) defined by control unit 8;

(x, y, k) others.

In the second image zoning unit 11, together with the image signal 5 received at the output of the reference block 14, there is also a signal from the output of the control unit 8.

The signal of the common matrix (X, Y, Z) formed in the second zoning block 10 of the image 11 is received at the inputs of: a compression block 6; adaptive zone binarization unit 15; zone sign generator 16.

Linear and nonlinear filters for converting homogeneous halftone image portions into binaries in the Adaptive Zone Binarization Block 15 work by comparing the brightness of each pixel with one threshold value defined for that portion 20 of the image by one of the following ways:

- measuring the distribution of brightness in a given area;

- measurement of the brightness gradient (contour finding); 25

- finding groups of pixels with close brightness values;

- estimation of the maximum difference between the brightness of an individual pixel and the average brightness in the treated area, etc. 30

In this way, the adaptive zone binarization unit 15 passes in the binaries only those portions of the image (Xd, Ud, Zfl) for which zj from the zone matrix (X, Υ, Z) takes the values a, b, c, 35 d and e.

The transfer functions of the adaptive binarizers F (A), F (B), F (C), F (D) and F (E) are determined by means of series and parallel coupled filters corresponding to the filters included in the first block for zoning the image 10, acting in accordance with the interest zj on those portions of the image for which zj = a, b, c, d and e.

In the zone sign generator 16, schemas are activated which include groups of signs and rules Q (d), Q (e) for recognizing printed text (zj = d) and handwriting signs (zj = e).

When the correspondences between the binarized images and their corresponding groups of attributes and recognition rules {for (x, y, d) turn on Q (d) and for (x, y, f) - Q (e)} reach a preset value , the code of the recognized image is formed at the output of the pattern recognition unit 17.

The comparison scheme determines the degree of coincidence (difference) between the two signals for each trait or for each set of traits in the series.

When this difference is less than a predetermined value (threshold), a code of the recognized image is generated at the output of the pattern recognition unit 17.

The signal received at the output of the pattern recognition unit 17 (the pattern recognition code) is fed to the compression unit 6 together with the signal received at the output of the adaptive zone binarization unit 15.

The compression unit 6 generates an image compression map using coordinately bound zones 11 in the second image zoning unit; these zones also form the filters by which an image consisting of: sections where no information is stored is stored in the memory of the database unit 7; sections with recognized images; sections containing the original halftone digital image.

All the results of the TOTO-fiche processing are recorded in a file in a block database 7, which is processed and used through a personal computer and responds to requests at different levels of consumption.

Control unit 8 is the unit to which the operator acts when necessary or at will.

Provision is made for operator intervention in defining (supplementing or reducing) interest areas. Through the control unit 8, new formalized images are created in the file format of the documented image generator files of the formalized image generator 12 by entering coordinates of rectangular or other zones in the first image zoning unit 10.

Claims (1)

Claims Document cataloger comprising a document carrier arranged in the focal plane of an optical system (lens) coupled to a sequentially coupled image sensor, an analog-to-digital converter and an image processing unit, with above 5 document carriers being a lighting system, characterized in that a video processor and a video memory are connected in series to the analog-to-digital converter, which together with the ADC form a shaper (4) in series with which ok for compression (6), database block (7) and control unit (8), the image processing unit containing the buffer memory (9) to the output of which is connected to the output of the shaper in parallel, a first image zoning unit (10) and a second image zoning unit (11) are connected in parallel, with a formalized image generator (12), an actuation drive unit, connected in series to the output of the first image zoning unit (10). the formalized image to the original (13) and the comparison block (14), such as the output of n the first image zoning unit (10) is also connected to the second input of the comparison block (14), and to the output of the second image zoning unit (11) are connected in parallel a block for adaptive zone binarization (15), a generator of zone features (16) and a second input of the compression block (6), the output of the adaptive zone binarization block (15) is connected to the first input of the pattern recognition unit (17) and to the third input of the compression block (6) and a second input of the block recognition unit is connected to the output of the zone sign generator (16) times (17) whose output is connected to the fourth input of the compression block (6) and the output of the control block (8) is connected to a second input of the database block (7) to the second input of the first zoning block of the image (10) and to the second input of the second image zoning unit (11) to which the output of the comparison block (14) is connected to the third inputs.