JPH0468972A - Image data compression system - Google Patents

Abstract

PURPOSE:To attain high speed processing by providing a data buffer storing a reference line data to each of plural compressors of the modified MR system and applying compression processing to a line at a head of 2nd and succeeding blocks while using a final line of one preceding block as a reference line. CONSTITUTION:A data buffer 22 to store a reference line data for compression processing is provided to each of plural MMR(modified modified read) type compressors 21. When a data of a head line of a 1st block of an image data is subjected to compression processing, a virtual data is stored in the data buffer 22 as a reference line data and when the data of a head line of 2nd and succeeding blocks is subjected to compression processing, a data of a tail end line of one preceding block is stored in the data buffer 22 as a reference line data. Thus, the correlation is not lost at a joint when blocks are linked. Since the compression processing is implemented in parallel by the plural compressors 21, high speed processing is implemented.

Description

[Detailed description of the invention] [Summary of the invention]

Regarding the method of compressing 2D image data using the MMR method, the purpose is to speed up the compression process by dividing 2D image data consisting of multiple lines into multiple blocks in the line arrangement direction, and M of pieces
An MR type compressor is provided, and each compressor performs MMR type compression processing for each block in parallel, and reference line data for compression processing is stored for each compressor. When compressing the data of the first line of the first block of the image data, virtual data is stored in the data buffer as reference line data, and the data of the second line of the image data is stored as reference line data. When compressing data on the first line of a subsequent block, the data on the last line of the block one block before that block is stored in the data buffer as reference line data, and the data on the second and subsequent lines of each block are stored as reference line data. When data is compressed, the data of the line one line before that in the block is stored in the data buffer as reference line data.

[Industrial application field]

The present invention relates to a method for compressing two-dimensional image data using the MMR method. For example, as a method for compressing image data of facsimile signals, the MH (Modified Huffman) method,
The MR (modified read) method and the MMR (modified MR) method are known. MH method and MR method are for G3 facsimile.
The method is a one-dimensional encoding method that compresses data by encoding the length of consecutive white pixels or black pixels in the line direction of image data. The MR method is a two-dimensional encoding method that also uses the correlation between adjacent lines of image data (correlation in the line arrangement direction), but one line out of every k lines is one-dimensionally encoded. be. In this MR method, by setting the number k of inserted lines for one-dimensional encoding to a small value, it is possible to reduce data loss when a transmission error occurs. The MMR method is for G4 facsimile, and the algorithm is the same as the MR method, but in order to maximize encoding efficiency, it is an encoding method that does not perform one-dimensional encoding as described above.

[Prior art] Conventionally, in the case of the MH method and the MR method, in order to perform data compression at high speed, a plurality of LSI compressors for each method are provided in parallel, and image data is processed for each multiple line. The image data is divided into a plurality of blocks, and each of the plurality of compressors compresses each block of image data in parallel. In the case of the MH method, since it is a one-dimensional encoding method, it is possible to divide the image data into a plurality of blocks for each line by dividing it into a suitable number of lines and perform compression processing. be. In addition, in the case of the MR method, even if it is a two-dimensional encoding method, when compressing original image data consisting of n lines as shown in FIG. 4A, as shown in FIG. 4B,
There is a one-dimensional compression process line for every k (-4 in the example) line, and the compression process that uses the correlation between the previous and succeeding lines stops at that one-dimensional process line, so the one-dimensional compression process This is because if the lines are divided into blocks, the MR compressor can perform the division process. However, since the MMR encoding method is a complete two-dimensional encoding method, there is no one-dimensional encoding process line, and the MH
EOL code indicating the beginning of a line such as MR method or MR method
doesn't exist either. That is, FIG. 4C shows compressed data of the MMR method. Since there is no reference line for the first line, the MMR compressor uses virtual white data, that is, all “0” data, as the reference line data. Two-dimensional compression is performed using the data, and subsequent lines are subjected to two-dimensional compression using the previous line as a reference line. In addition,
In FIG. 4C, the EOP code is a code indicating the end of image data processed by the compressor. Therefore, when compressing image data using the MMR method, if the image data is divided into blocks for each line and processed by multiple MMR compressors, each MMR
The MR compressor compresses the first line of the block to be compressed using the virtual white data as a reference line, and when multiple blocks are combined, the correlation between the preceding and succeeding lines becomes incorrect at the block break. , the original image data cannot be accurately restored during decompression. Therefore, in the case of the MMR method, conventionally, as shown in FIG.
The data is compressed by encoding the data, and the compressed data 5 is stored in the compressed image memory 4. [Problems to be Solved by the Invention] As described above, when compressing image data using the MMR encoding method, conventionally the compression processing must be performed using one compressor, so it is necessary to increase the processing speed. I couldn't. In view of the above points, it is an object of the present invention to enable high-speed compression processing even in the MMR method.

[Means to solve the problem]

In order to solve the above problems, in the present invention, two-dimensional image data consisting of a plurality of lines is divided into a plurality of blocks in the line arrangement direction, and a plurality of MMR
A compressor is provided, and each compressor performs MM for each block.
A data compression method that performs R-type compression processing in parallel, and includes a data buffer for storing reference line data for compression processing for each compressor,
When compressing the data of the first line of the first block of the image data, store virtual data in the data buffer as reference line data, and compress the data of the first line of the second and subsequent blocks of the image data. When compressing data, store the data of the last line of the previous block in the data buffer as reference line data, and when compressing the data of the second and subsequent lines of each block, An image data compression method is provided in which data of a line one line before that block is stored in the data buffer as a reference line.

[Effect]

With the above configuration, each compressor performs compression processing on the first line of the second and subsequent blocks using the last line of the previous block as a reference line. Said 1
When looking at the original image data, the last line before a block is the line before the first line of each block and is the original reference line, so even when blocks are combined, there is a correlation at the joint between the blocks. Since it does not collapse, one M
Compression can be performed in exactly the same way as when image data is compressed using an MR compressor. At this time, since compression processing is performed in parallel by multiple compressors, high-speed processing is possible.

【Example】

FIG. 1 is a block diagram of an embodiment of a compression processing apparatus that executes a compression method according to the present invention. In the figure, 10 is an original image memory, which stores two-dimensional image data IM. In this example, as described later, the image data IM is divided into a plurality of blocks B o
−B Divide into m and process. 201 to 204 are compression processing units, and in this example, four blocks of data are simultaneously processed in parallel by these four compression processing units 201 to 204. This compression processing unit 201~
204 each includes an MMR compressor 21, a data buffer 22 for storing reference line data for two-dimensional compression, a DMA controller 23, and a FiFo 24.
It is equipped with A compressed image memory 30 has a storage section 31 for compressed block data b o -b n and a storage section 32 for the final compressed image data im. 40 is a CPU for controlling the compression processing operation.
It is. Further, 41 is a system bus. The compression processing operation of this example will be explained below with reference to the flowchart of FIG. First, set line number j to O in step 1, 01.
), the first compression processing unit 20+ is activated (step 102). After this activation, the process proceeds to step 103, where the line number i is rewritten to i+, and here to line number k. Next, it is determined whether or not there is an unprocessed line in the image data IM (step 104). That is, it is determined whether the line number i is smaller than the last line n of the image data IM, and when it is determined that it is smaller than n, that is, there is an unprocessed line, the process proceeds to step 105, and the unused free space is compressed. Determine whether there is a section. If there is a free compression processing section, the process jumps to step 107; if there is no free compression processing section, the process waits until a free space becomes available (step 106), and then proceeds to step 107. In step 107, an empty compression processing unit, in this case the next compression processing unit 202, is activated. When this startup is finished, return to step 103 and change the line number i to i+k (-
2k). Then, in the same manner as described above, the next compression processing unit 203 is activated in steps 104 to 107. Furthermore, return to step 103 and set the line number i to i+k.
(=3k), and in the same way, the compression processing unit 204
Start. When the four compression processing units 201 to 204 are activated in this manner, in step 106, the process waits for the processing in the compression processing units to be completed. On the other hand, the compression processing unit 20. processes the block Bo of the first line of the image data 31, but for the data of line 1, the compression processing unit 20. Virtual mill ("O") data is written in the data buffer 22 of. Then, MMR compression processing is performed sequentially from line 1, and from the second and subsequent lines, the block B1
The data of the previous line in the middle is rewritten to the data buffer 22 as reference line data, and each M
MR compression processing is performed. The compression processing unit 202 also processes the image data line (
The second block B consisting of lines from line (k+1) to line (2k) is processed, but the data of the first line (k+1) of this block B1 is stored in the data buffer 22 of this compression processing unit 202. is the previous block Bo
Write the data on the last line of . Then, compression processing is performed sequentially from line (k+1), and from the second and subsequent lines, the data of the previous line in block B1 is rewritten to the data buffer 22 as reference line data, and each is subjected to MMR compression. Processing is done. Similarly, the compression processing unit 203 compresses the 3rd and 11th block B2, and the compression processing unit 204 compresses the 4th block B3. In this case, the data of the last line one block before is sent to each compression processing unit 203.20.
4 data buffers 22, respectively MM
R compression processing is performed. After that, when the compression processing section where the compression processing has finished occurs,
The process is started again in step 107, and the next block of data is compressed in the same manner as described above. In this way, the compressed block data bO to bn of each block B o to B n are sequentially written to each corresponding storage section 31 of the compressed image memory 30 . In this case, as shown in FIG.
At the end of n, a code EOP indicating that it is the end of the image data is added. When all the data of the image data IM has been compressed, it is determined in step 103 that there is no unprocessed line, and the process proceeds to step 108, in which all the compression processing units 20. Wait for the processing in steps 204 to 204 to be completed. When the processing is completed, the process proceeds to step 109, and as shown in FIG. Delete and combine. With the above steps, the process ends. In this way, the compressed image data im, which is the final result equivalent to compression processing by one MMR compressor, is obtained and stored in the storage unit 32. Then, if necessary, the data is stored in an external storage device such as a DASD or an optical disk. In the above case, multiple MMR compressors are used to process image data in parallel, so high-speed MMR
MR compression processing can be realized. Note that the number of compression processing units may be two or more and any number, and the number of lines in each block does not need to be the same as in the above example, but may be different.

【Effect of the invention】

As explained above, according to the present invention, image data is divided into a plurality of blocks, and a plurality of MMR compressors are used to divide image data into blocks.
Even if multiple blocks are processed in parallel, the correlation between the lines before and after each block will not be lost at the joint between each block.
MMR compression processing can be performed at high speed.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of a compression processing device that executes an image data compression method according to the present invention, FIG. 2 is a flowchart for explaining an embodiment of the compression method according to the present invention, and FIG. 4 is a diagram for explaining the method of combining compressed block data, FIG. 4 is a diagram for explaining the original image data, MR method compressed data, and MMR method compressed data, and FIG. 5 is a diagram for explaining the conventional method of combining compressed block data. FIG. 2 is a block diagram of an MMR compression processing device. 10; Original image memory 20, ~ 204; Compression processing unit 21, MMR compressor 22; Data buffer 30; Compressed image memory 47>... Patent applicant: Fujitsu Limited K (

Claims (1)

[Claims] Two-dimensional image data consisting of a plurality of lines is divided into a plurality of blocks in the line arrangement direction, and a plurality of M
An MR type compressor (21) is provided, and each compressor (21) performs MMR type compression processing for each block in parallel. A data buffer (22) is provided for storing reference line data for processing, and when compressing data of the first line of the first block of the image data, virtual data is stored in the data buffer (22). When data is stored as reference line data and the data of the first line of the second and subsequent blocks of the image data is compressed, the data of the last line of the block one block before that block is stored as the reference line data. When storing the data in the data buffer (22) and compressing the data of the second and subsequent lines of each block, the data of the line one line before that in the block is used as reference line data and the data is stored in the data buffer (22). A compression method for image data stored in .