DE2413090A1 - PROCEDURE AND ARRANGEMENT FOR COMPRESSION AND STORAGE OF GRAPHICAL INFORMATION AND ITS SELECTIVE RECOVERY - Google Patents

Description

Boeblingen, March 7, 1974
te-fr

Applicant: International Business Machines

Corporation, Armonk, N.Y. 10504

Official File number: New registration

File number of the applicant: NE 972 001

Method and arrangement for compression and storage
graphic information and its selective restoration

The invention relates to an arrangement for compression and storage graphic information according to the preamble of claim 1.

In bank and postal check transactions, the signature verification is the essential tool for financial transactions
ensure that these transactions are carried out for the authorized persons. Up to now, it has been customary to compare the signatures with stored signature patterns that are recorded in a card index. For example, the verification takes place in a special department of the bank, the employees of which process the documents submitted by the account holders assigned to them and the signature in the process
check using the index cards of these bank customers. The examination of documents, which are handed over to the clerk, can either be done by the clerk himself or by the employees in the said before the transaction is carried out
Department. In the latter case, you must
for this purpose, the documents must be brought to this department or a television system must be available with which the images of the signatures can be transmitted.

40984S / 0708

Requires the storage of larger quantities of signature cards a relatively large amount of space. In addition, cards in a signature file can be lost or incorrectly classified will. Distributing the incoming documents to the employees concerned and sorting them requires additional Time; the same applies to the transport of signature cards within the bank or from the bank to remote branches. Time is still lost when signatures for such account numbers have to be compared with which more as an authorized person has access.

The object of the present invention is accordingly to provide a device which has a large number of can store cpraphic information in compact form in a central memory, which has a small footprint, can be queried quickly and easily from one or more nearby or distant stations and which one is the desired Can restore information to its original form after selection.

2ur solving this problem are those in the characterizing part of the patent claim 1 proposed inventive measures, the advantageous developments of which are characterized in the subclaims are.

The reduction in the graphic information to be stored per signature, in addition to the advantage of the larger number of stored signatures, also results in a reduction in the access time and shorter times for the transmission of the data and thus a shorter processing time per transaction overall. In addition, in the present invention, the graphic data can easily be displayed on the screen with other stored data, which further increases the security of financial transactions. All these benefits can be personalized with the invention without great expense s realized as the proposed arrangement despite their great

4098 45/0708

Efficiency with only relatively little circuit complexity is realized.

An embodiment of the invention will be described below with reference to drawings. Show it:

1 shows a schematic representation of an embodiment of the invention,

Fig. 2 shows a signed document, the sub

writing field occupies a defined position within the document,

Fig. 3 the schematic representation of the compression

Facility,

Fig. 4 shows a field consisting of eight picture elements for

the predictor.

Fig. 1 shows the schematic representation of an embodiment the invention. The data processing device 1 is provided with an input unit 2, a memory unit 3, and a control unit and a modem 5 connected. The input unit 2 comprises an optical reader and a compressor device, which the from Reader converts continuously supplied digital image signals into a compact form. The storage unit 3 is one of the memories with direct access, for example to the disk storage.

The data processing device 1 is directly connected to one or more output units, such as 6 and 7, by means of the control unit 4 ; In addition, there are connections to remote output units 8 and 9 via the modem 5 and a wire or radio transmission link, the modems 10 and 12 and a control unit 11 and 13 being interposed. If necessary, a printer 14 can be connected to one of the remote output units, for example the unit 9. To each of the

NE 972 001 409 845/0 70 8

Output units 6, 7, 8 and 9 include a data decompressor (Expander) which converts the information that has been retrieved from the storage unit 3 back into the uncompressed form convicted; The output units also include a keyboard and a screen display device (e.g. with a cathode ray tube)

The optical reader in the input unit 2 is for the line by line optical scanning of a specific area on the surface of the documents. In addition to that to be scanned Image, which represents a signature, for example, each document contains further information, namely Reference information such as the account number; this information however, it is not scanned by the reader.

Fig. 2 shows schematically an embodiment of a document, namely a card that is in a certain area 16 of their Surface contains a signature 15; in the drawing, said area is indicated by hatching. Of clarity For the sake of this, the additional reference information outside the area 16 is not shown on the document in FIG. 2. the The signature can be handwritten, printed or otherwise visibly applied to the document. Of the optical reader almost the image surface 16 as a matrix field, which in the embodiment discussed from 64 χ 192 = 12 288 square picture elements, which have to be examined for their color content (white or black). the The width of the scanning track (the resolution) of the reader can be set by the operator and then determines the dimensions of the image surface to be scanned, the ratio being Length to width is always 3: 1. For example, if the scanning track has a width of 0.2 mm, the one to be scanned is the one to be scanned

2
Image surface 39 χ 13 mm, while in the case of a 0.4 mm wide

ο scanning track the image area to be scanned is 78 χ 26 mm. As indicated in Fig. 2, the image is always scanned in the same direction during the line-by-line scanning movement, always starting from the same page of the document.

409845/0708

After scanning each line, the reader gives an "end of line" signal (EOS) and after scanning the entire image surface, a signal "end of image" (EOP).

The data processing device 1 controls all operations of the Systems. Used in one of the output units 6,7,8 or 9 requires the display of a specific image, the information content of which is contained in the memory unit 3, the Reference information belonging to this image is entered using the keyboard belonging to the output unit in question will. Then, with the help of the reference information, the associated Searched for compressed digital image information under the control of the data processing device 1 in the storage unit 3. After decompression in the expander of the output unit, the desired image appears together with the reference data as a visible image on the screen of the output unit. If the image is displayed together with alphanumeric information, so the picture itself covers about 1/3 of the screen surface. The starting point of the picture is indicated by a special symbol, the so-called attribute symbol is displayed, which is immediately in front of the compressed digital image information.

The compression technique according to the invention for the digital image information, reading from documents results in a significant reduction in this data (the compression or compression factor is approximately 6: 1); it's over For this reason it is then also possible, in the available storage space of the storage unit 3, to store image information of very high store many more documents than would be possible without compression. The compression of the stream of digital image signals is carried out in a two-stage compression unit (compressor), one embodiment of which is indicated schematically in FIG. 3. The first stage, called the predictor stage, comprises a sample buffer IS, a predictor logic circuit 19 and a Exclusive OR logic circuit 20.

972 001 409845 / 07Q8

The predictor stage comprises one consisting of eight picture elements Predictor field, which is shown in FIG. 4. P represents represents a picture element which is being scanned; for this the color (black or white) or the bit value (1 or 0) is predicted based on the color (black or white) of the earlier sampled eight picture elements in the Predictor field. The scanning of the picture elements in FIG. 4 takes place in the direction of the arrow 17. FIG. 4 clearly shows the positions of the eight substantially square picture elements which lie in consecutive lines. About these eight picture elements belongs to an element 1 which is on the same line one position before the picture element P which is currently being scanned also four consecutive picture elements L-2, L-1, L and L + 1 on the first above and before scanned line and finally three successive picture elements 2L-1, 2L and 2L + 1 in a second overlying one and previously scanned line. The predictor field with eight picture elements has proven to be correct in terms of the prediction of the color or bit value of the picture element P turned out to be very efficient if the image to be scanned is other than horizontal and vertical lines also includes oblique lines such as those with an angle of plus 45 ° or minus 45. When selecting the predictor field, it was taken into account that there is a high probability with normal handwriting and in signatures instead, there are more lines from the bottom left to the top right than there are lines from the left run up to the bottom right. The predictor field with eight picture elements therefore does not represent any restrictive condition regarding the nature of the image to be scanned, which is different from thick lines running in different directions can exist and which differently shaped planes may have.

The predictor table listed below, which is based on refers to the predictor field shown in Fig. 4, with the help of a statistical analysis of the black-and-white distribution in a

^103972001 409845/0708

large number of different signatures, manuscripts and printed symbols. In this table is the predicted Color of the picture element P by "x" or "." marked; this is intended to indicate that the predicted Color is black or white on the assumption that the color of the picture elements is black in various combinations. For the sake of simplicity, a different notation is used in the table for the eight picture elements mentioned, namely A, B, C, D, E, F, G, H instead of the previously used 2L-1, 2L, 2L + 1, L-2, L-1, L, L + 1 and 1.

/ * TAB 1

Predictor table

F. G O • D. C. C. A. A. A. A. B. B. B. B. B. B. B. B. F. G H X • D. C. C. A. A. A. A. F. X • • • D. D. C. C. C. C. F. H X X X • • • • • D. D. D. D. G X X • • X • X B. • • • • • • • . G H X X X X X X • • X • X • X • X • X X X X X X X X • • • • • • • • H X X • • • • • • X X X X X X X X F. G X X X X X X • • X X X X X X X X F. G H . · • X X • • • • X X X X X X X X F. X X • • X X X X X X X X • • ■ φ F. H X X • X • • • • X X X X X X X X E. G • X • • • • • • • • • • • • • • E. G H • • • • X X • • • X • X * • B. E. • X • • X X • • • • • • • • m • E. H X • • • X X • • X X • • X X m ■ E. X • • X X X • • • • • X X • • E. X X • • • • • X • • X X X • E. X X X X • • • • X • X • E. X X X X X X X X

NE 972 001

409845/0708

The capacity of the buffer memory 18, which consists of a shift register exists and continuously receives the digital image information signals from the reader, is chosen so that the memory at least the image information of the line that has just been scanned and the image information of the two previous and already scanned lines Can accommodate lines. The buffer memory 18 contains eight connection points at which the bits of the picture elements 2L-1, 2L, 2L + 1, L-2, L-1, L, L + 1 and 1 appear. These bits are the predictor logic circuit 19, the function of which is based on the predictor table given above. The exclusive-OR logic circuit 20, one input of which is connected to the output of the predictor circuit 19 and the second of which Input directly receives the information of the scanned pixel, compares the predicted color (black or white) of the Picture element P with the actual color (black or white) of the real picture element in the original image. If the prediction is correct, the exclusive-OR combination circuit 20 outputs a bit with ' Value O to the second stage of the compression device, while in the case of an incorrect prediction, the circuit 20 turns on Forwards bit with value 1 to said second stage. If some of the predictor orbits P lie outside the image surface of the Document, these predictor orbits are interpreted as O bits. The current thus generated at the output of the exclusive OR circuit 20 digital information represents a transformed image that deviates from the original as a result of incorrect predictions and contains a considerably smaller number of bits of value 1 (black dots) than the image information from the original image. However, the total number of bits remains the same. So up to this point no compression has yet been achieved. Of the the actual compression process now takes place in the second stage, which is referred to as the coding stage. However, since in the the number of bits with the value 1 was reduced in the first stage, the effect of the second stage of the compressor device increases, i.e. the compression factor increases.

It turns out that an extension of the previous predic

NE 972 001 409 84 5/0708

241

gate field with eight picture elements, i.e. the inclusion of new picture elements the number of correct predictions is allowed to increase and thus the number of bits with value 1 in the transformed Can reduce the image even further. This increases the compression factor and the information is saved requires even less storage space. In this way, however, one finally reaches a state in which the increasing Cost of the additional facilities in the compression device and the expander no longer by the advantage of a Storage space savings are outweighed.

The second stage of the compression device contains a counter 21 which is connected to the output of an AND logic element 23 connected is; the counter determines the number of bits with the value 0 in each scanned line, i.e. the number of correct predictions, which precede each bit with the value 1. The counter 21 thus also determines the length of each series of bits it traversed Value 0. The various pass lengths are converted into code words by the encoder 22. Because short run lengths occur more frequently than long, a code is used that takes this fact into account.

The following table provides the code words for run lengths, which apply to counter readings from 0 to 191 as well as the special codes for the signals EOS and EOP. The first mentioned code words, which always begin with a digit less than 4 and the following digits are always greater than 4, comprise three Bits or a multiple of three bits up to a maximum of 12 bits.

972 001 40984 BV 0708

Counter reading or Code for special • Code for Pass length Pass length symbol Special symbol O 0 04 1 1 14th 2 2 EOS 24 3 3 34 4th 05 044 5 15th 144 6th 25th • ■ • EOP 374 15th 37 0444 1444 16 045 • 17th
■ 145
* 3774 31 375 32 046 33 146 * 'B
m 63 377 64 0445 65 1445 - 191 3776

When each line is scanned, the counter 21 is reset each time by a signal from the AND logic circuit 23, if a bit with the value 1 appears on the relevant line. Counting bits with value 0 on the same line starts over after each reset until the entire line has been scanned. At the end of each line (see Fig. 2) returns the reader an EOS signal, whereby the counter 21 is reset and the AND logic circuit 24 is opened, so that between each series of code words with the compressed digital information of a scanned line an EOS code word can be inserted. Finally, if the entire image area of the document has been scanned, the reader generates a (not drawn) EOP code signal, which is processed in the same way as the EOS signal and thereby the series of

NE 972 001

409845/0708

241

Terminates code words with an EOP code word. The compressed digital information is generated with the aid of the OR logic circuit 25 brought to a buffer memory 26, where it is temporarily stored before it is under the control of the data processing device 1 (Fig. 1) is finally written into the memory unit 3 (Fig. 1).

The decompression of the compressed digital information which is read out from the storage unit 3 takes place in each of the Output units 6,7,8 and 9 (Fig. 1) through an expander, taking the same steps as in the compression unit, however can be carried out in reverse order. First, the code words are separated from one another and the coded ones Pass lengths decoded to make series of the correct length to obtain. The transformed image thus restored is then input line by line into a circuit whose Function is based on the same predictor table on which the predictor circuit 19 of the compression device was based, however, they are used in an inverse manner. The structure of the original image is shown in the upper left corner started using a line above and outside the picture that contains only white picture elements. the Color predicted for the first picture element, which is a white element, then becomes the color of the first picture element of the transformed image is compared so as to reconstruct the first picture element of the original image. This Image element is stored and used for subsequent predictions. By repeating this process you can do the whole Original image can be reconstructed, with the picture element P on - the image surface from left to right and from top to bottom is gradually shifted.

972 001 409 845/0708

Claims (10)

PATENT CLAIMS 1. Process for the compression of graphic data and for their Recovery, with the graphic data, for example, in the form of a binary or black-and-white raster pattern present, characterized by: a) a two-stage data compression process, in the first or predictor stage of which the color value or the bit value of a picture element is predicted and, in addition, the bit values of several selected picture elements that have already been scanned according to a logical function optimized with respect to the type of graphic data to be processed (Predictor function) are linked and if the predicted and the actually sampled match Bit value a binary 0 or, if they do not match, a binary 1 is generated, so that a to build the transformed image with an unchanged number of image elements and in its second or coding stage in each image line the transformed one containing only a few binary ones Image the length of the all zeros bit strings of each line in coded form transferred and buffered together with special codes for end of line and end of image; b) the final storage of the compressed image information after completion of the compression process for an overall picture; c) a two-stage decompression process which, analogous to the compression process, consists of two stages, wherein the first or decoding stage reconstructs the transformed image from the stored code words and wherein the second or inverse predictor stage is reversed that used in the compression logical function the original image point by point and line by line from the transformed image rebuilds. «» »« »409845/0708 2. Procedure for determining the logical predictor function according to claim 1, characterized by the following steps: a) Selection of a predictor field consisting of several image elements that can be moved across the image surface, which has a defined position with respect to the picture element just scanned and exclusively has previously sampled picture elements; b) Selection of a logical link (predictor function) among the binary values of the predictor field by setting up a predictor table (truth table); c) Application of this function to a large number of data to be compressed according to the method according to claim 1 ; d) Determination of the prediction probability text or the Compression factor for the selected ensemble of graphic data; e) Modification of the predictor function and / or the predictor location fairness; f) repetition of steps c) -e) until an optimal prediction probability is reached Predictor function for the underlying ensemble of graphic data. 3. Arrangement for carrying out the method according to claim 1, marked by: - a two-stage data compressor with a predictor stage (18, 19, Fig. 3) and a coding stage (21-26), - a memory with direct access for storing the compressed and encoded image, - a two-stage data expander with decoding stage and inverse predictor stage. 4. Arrangement according to claim 3, characterized in that the picture elements of the grid pattern are square Have shape and are processed line by line. ^{NE 972} ° ⁰¹ 409845/0708 241 5. Arrangement according to claims 3 and 4, in particular for Processing of manuscripts and printed documents, characterized in that the predictor device has one of eight picture elements existing predictor field is based on the color or bit values of the picture elements in the predictor field logically linked to one another in accordance with a predictor table (TAB 1) in a predictor circuit (19, FIG. 3) and that the output of the predictor circuit through an exclusive OR circuit (20) with the value of the even scanned picture element is linked. 6. Arrangement according to claims 3 to 5, characterized in that the predictor field (Fig. 4) above and to the left the pixel being scanned and the first left neighbor of the pixel is on the same line includes, as well as four picture elements in the first overlying picture line and further picture elements in the second line of images above. 7. Arrangement according to claims 3 to 5, characterized in that the predictor device is a shift register (18) contains which the picture elements of the picture line just scanned and several, in particular saves two previous picture lines and which one with terminals according to the selected predictor field is provided for feeding the image values to the predictor circuit (19). 8. Arrangement according to claims 3 to 7, characterized in that the coding stage has a counter (21) downstream encoder (22) comprises, which the run length of series of binary zeros on each Determine the line of the transformed image and reset the counter when a picture element is encountered with value 1 or at the end of a scanned line, whereupon the coded content of the counter is temporarily stored in a buffer memory (26). NE 972 001 409845/0708 9. Device for the registration, storage and retrieval of graphic data using the method and the arrangement according to claims 1 and 3 to 7, characterized by: a) an input device (2) with an optical reading device for line-by-line scanning of certain areas on entered documents, whereby the color values are preferably represented in binary (black & white) and for End of line and end of frame special signals are emitted; b) a compression device in the input device (2) for compressing the graphics generated by the reading device Information consisting of a predictor level and a coding level; c) a memory (3) with direct access for storing the compressed image information; d) several output units (6, 7, 8, 9) each with one Decompression device (expander) for restoring the graphic data, a display device for displaying the generated original image, a keyboard and control or connection units (4, 10, 11, 12, 13); e) a data processing device (1) for controlling the overall system, in particular for storage and for Reading out the compressed image information from the memory unit (3). 10. Device according to claim 9, characterized in that on the screen display device in addition to the restored Original image, additional information, such as alphanumeric data, can be displayed. 001 409845/0 70