JP2004343459A - Data compression system and data decompression system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To efficiently compress/decompress correlated data for every fixed data unit. <P>SOLUTION: A compression algorithm executing unit 2 compares an input data string with a data string at a position in correlation with the input data string, and encodes the matched position into a code having a fixed bitwidth, and combines data strings at the positions which do not match the code and compresses it into compressed data. A decompression algorithm executing unit 12 decodes the codes in the compressed data string, and combines the data strings at positions which do not match each other. Thus, the input data string is compared with the data string at the position in correlation with the input data, and the matched position is encoded into the code having the fixed bitwidth, and the data strings at the positions which do not match the code are combined and compressed into the compressed data, so that the correlated data can be efficiently compressed/decompressed for every fixed data unit. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a data compression system for efficiently encoding and compressing data having a correlation for each predetermined data unit and a data restoration system for restoring data compressed by the data compression system.
[0002]
[Prior art]
As a technique for storing or transferring a large amount of information with as small a capacity as possible, a high-efficiency coding technique is known. FV (Fixed-to-Variable) coding, LZ (Lempel-Ziv) coding (see Patent Literature 1 below), LZW (Lempel-ZivWilch), and LZW (Lempel-ZivWilch) 2. Description of the Related Art There is known a method of performing compression by using the matching between a data string that has appeared in the past and a data string to be compressed, such as an encoding method (see Patent Document 2 below).
[0003]
[Patent Document 1]
US Patent No. 4,464,650 [Patent Document 2]
US Patent No. 4,558,302
[Problems to be solved by the invention]
However, conventional data compression techniques require that all data to be compressed be once scanned during data compression, and thus cannot perform compression processing while inputting data. There is a problem that it takes a lot of time to complete the compression process.
[0005]
In order to solve such a problem, a so-called one-pass compression / decompression method such as a zero compression method or a run length compression method for performing compression processing while inputting data has been proposed. However, these one-pass compression / decompression methods can achieve a high compression ratio when the data to be compressed has a specific word repetition structure. As described above, it is not an efficient compression method to compress data having a correlation for each fixed data unit.
[0006]
Here, the data having a correlation for each fixed data unit is, for example, as shown in FIG. 7, data of one word (32 bits: bit) unit is n columns in one row (in the example shown in FIG. 7, (8 columns) means data having a data structure that is arranged and has a correlation (similarity) between adjacent rows for each column. To give a specific example, the word in the b-th row and the second column in FIG. 7 has a correlation with the word in the a-th row and the second column, and also has a correlation with the word in the c-th row and the second column.
[0007]
The present invention has been made in view of the circumstances described above, and provides a data compression system and a data decompression system that can efficiently compress and decompress data having a correlation for each fixed data unit. With the goal.
[0008]
[Means for Solving the Problems]
In order to achieve the above object, the present invention compares an input data sequence with a data sequence at a position correlated with the input data sequence, encodes the matched position into a code having a fixed bit width, and converts the position to a code that does not match the code. Are combined and compressed.
[0009]
That is, according to the present invention, means for dividing data to be compressed among input data into input data strings having a predetermined number of compression unit bits,
A storage unit that stores an input data sequence that has been input once to be compared with the input data sequence as a registered data sequence,
Each divided data string obtained by further dividing the input data string divided into the predetermined compression unit bit number into an arbitrary bit number width, and a divided data string at a position correlated with the divided data string stored in the storage means Are compared, the divided data string at the matched position is encoded into a code having a fixed bit width, and the divided data string at a position that does not match the code is synthesized and compressed as compressed data. Compression means for registering in the storage means as a registration data sequence,
A data compression system is provided.
[0010]
Further, according to the present invention, there is provided a data restoration system for restoring data compressed by the data compression system according to claim 1, wherein a storage means for storing a previous restored data string as a registered data string;
Extracting a divided restored data string stored in the storage unit based on the encoded data in the compressed data string compressed by the data compression system, and extracting the divided restored data string at a position that does not match the extracted divided restored data string And generating a restored data string by synthesizing the data, and restoring means for registering the generated restored data string as the next registered data string in the storage means.
A data restoration system is provided.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a data compression system according to the present invention, FIG. 2 is a flowchart for explaining a compression algorithm of a compression algorithm execution unit in FIG. 1, and FIG. 3 is a compression algorithm execution unit in FIG. And FIG. 4 is an explanatory diagram showing an example of encoding.
[0012]
<Data compression system>
As shown in FIG. 1, the data compression system according to the present embodiment includes a data input unit 1 that divides input data to be compressed into a column having a predetermined compression bit width and inputs the divided data, A registered data string memory 3 for dividing the registered data into columns having the predetermined bit width and storing the data, a compression algorithm executing unit 2 for performing compression processing on the input data, and a data output for outputting the compressed data It has a part 4.
[0013]
The compression algorithm execution unit 2 compares the input data sequence input from the data input unit 1 with a registered data sequence stored in a column arrangement correlated with the input data sequence. Then, the input data sequence is further divided arbitrarily and a data sequence that matches the input data sequence and the registered data sequence in a plurality of data widths is determined, and a fixed-width code corresponding to the determined result is determined from the fixed-width code table 5. The acquired fixed width code and the bit portion of the input data string other than the matched data width are output to the data output unit 4 as compressed data of the input data. Further, the compression algorithm execution unit 2 registers the input data sequence before compression as registration data to be compared next at a corresponding column position in the registration data sequence memory 3.
[0014]
Next, a data compression process executed by the compression algorithm execution unit 2 will be described with reference to FIG. In the process of step S0, the compression algorithm execution unit 2 performs an initialization process. More specifically, the registered data counter N, which is an internal counter, is reset (N is set to 1), and the registered data to be compared is divided into columns having a predetermined bit width and stored in the registered data column memory 3. I do. Here, the “registered data counter N” is an index indicating the column number of the input data column input to the data input unit 1, and when processing the input data in the first column, N = 1. When processing the input data in the second column, N = 2,. . . , N = n when inputting the input data in the n-th column.
[0015]
In the process of step S0, as the registration data to be registered in the registration data string memory 3, the data of the first row of the input data is divided into columns having a predetermined bit width and registered, and the second row of the input data is registered. The compression process is started from the eye data, and the data of the first line of the input data is output to the data output unit 4 without being compressed. Alternatively, predetermined dummy data may be registered in the registered data string memory 3 so that processing can be started from the first row of input data to be compressed.
[0016]
The compression algorithm execution unit 2 obtains the Nth input data sequence to be compressed, which is input by the data input unit 1 in the next process of step S1, and sets the value of the registered data counter N in the subsequent process of step S2. The corresponding registered data string stored in the Nth column to be compared in the registered data string memory 3 is acquired. In the processing of the subsequent step S3, the compression algorithm execution unit 2 divides the acquired input data string and the registered data string of the N-th column into a plurality of data having a predetermined arbitrary width, respectively, and then proceeds to the processing of step S4. In step S3, it is determined whether or not the corresponding data of the divided input data and the divided registered data divided by the process of step S3 match.
[0017]
In the processing of the subsequent step S5, the compression algorithm execution unit 2 divides the registration data string a and the input data string b into arbitrary widths in step S3, as shown in FIG. e) A fixed-width code corresponding to the matching / mismatch determination result of the corresponding data in the divided input data (f, g, h) is obtained from the fixed-width code table 5 as shown in FIG.
[0018]
The examples shown in FIG. 3 and FIG. 4 are divided registration data c, d, e obtained by dividing a 32-bit (bit) registration data string a and a 32-bit input data string b into 4 bits, 8 bits, and 20 bits, respectively, It shows an example of determination of a 4-bit fixed-width code (bit value is expressed in hexadecimal) according to the determination of coincidence between c and f, d and g, and e and h at the same position in data f, g and h.
If the post-division registration data c and the post-division input data f match, and the post-division data c and f do not match, a 4-bit fixed width code “4” is set,
When the post-division registration data c and the post-division input data f match, the post-division registration data d and the post-division input data g match, and the post-division data (c, f) and (d, g) do not match Is set to a 4-bit fixed width code "7",
If the post-division registration data c, d, and e do not all match the post-division input data f, g, and h, a 4-bit fixed-width code “1” is set.
[0019]
In the process of step S6, the compression algorithm execution unit 2 combines the acquired fixed-width code and the divided input data that did not match in the process of step S4, and generates data as compressed data of the input data in the Nth column. Output to the output unit 4.
[0020]
For example, in the example shown in FIG.
If the post-division registration data c, d, and e do not all match the post-division input data f, g, and h, the 4-bit fixed-width code “1” and the post-division input data f, g, and h that do not match are used. Output,
If the post-division registration data c and the post-division input data f match but the post-division data c and f do not match, the post-division input data g and h of the part that does not match the 4-bit fixed width code “4” Output,
When the post-division registration data c, d, and e all match the post-division input data f, g, and h, only the 4-bit fixed width code “8” is output. That is, the compression efficiency increases as the number of matching post-division data increases (the correlation degree increases).
[0021]
The compression algorithm execution unit 2 registers the input data sequence to be compressed in the N-th column of the registered data sequence memory 3 in the process of step S7, and in the subsequent process of step S8, sets the value of the registered data counter N to the registered data sequence. It is determined whether or not the maximum address value n (the maximum number of columns n) of the column memory 3 is reached. Then, as a result of the determination, if the value of the registered data string N is not the maximum address value n, the compression algorithm executing unit 2 adds 1 to the value of the registered data counter N as the processing of step S9, and proceeds to the processing of step S1. Then, the process returns to the next input data sequence.
[0022]
On the other hand, if the result of the determination in step S8 indicates that the value of the registered data string N is the maximum address value n (maximum number of columns n), the compression algorithm execution unit 2 executes the processing of step Sa to change the value of the registered data counter N The value is reset to 1, and the process returns to step S1 to shift to the process for the input data string in the first column of the next row. The compression algorithm execution unit 2 executes the above processing for all input data input by the data input unit 1.
[0023]
As is apparent from the above description, in the data compression system according to the present embodiment, the compression algorithm execution unit 2 compares the input data sequence with the registered data sequence having a correlation positional relationship, and arbitrarily sets the divided data sequence. The fixed-width code and the bit string of the mismatched portion according to the determination result of the individual match between the input data string and the registered data string at the data width of are output as compressed data, and the input data string before compression is Is registered at a corresponding column position as a registered data string to be compared. According to such a configuration, it is possible to efficiently and quickly compress even data having a correlation for each fixed data unit while inputting data.
[0024]
<Data restoration system>
Next, the configuration of the data restoration system according to the present embodiment will be described with reference to FIGS. As shown in FIG. 5, the data restoration system includes a data input unit 11 for inputting data including a compressed data string compressed by the data compression system according to the above-described embodiment, and a registration data to be compared having a predetermined width. , A registration data string memory 13 for dividing and storing the data, a decompression algorithm executing unit 12 for performing decompression processing on the input compressed data, and a data output unit 14 for outputting the decompressed data.
[0025]
The decompression algorithm execution unit 12 outputs the registered data sequence stored in the column position correlated with the compressed data sequence from the fixed-width code table 15 based on the fixed-width code included in the compressed data sequence input from the data input unit 11. Calculate a matching portion of the data string before compression with each data width arbitrarily divided, extract a matching portion from the registered data sequence, and determine the extracted matching portion and a bit portion after the fixed width code of the compressed data sequence. Are combined and output to the data output unit 14 as a restored data string having a predetermined bit width. Further, the restoration algorithm execution unit 12 registers the restored data string as the next comparison target registration data at the corresponding column position in the registered data string memory 13.
[0026]
The data restoration system shown in FIG. 5 will be described as a system configuration separate from the data compression system shown in FIG. 1. For example, a data input unit 11, a registered data string memory 13, a fixed width code table 15, A part or all of the components such as the data output unit 14 are combined with the components such as the data input unit 1, the registered data string memory 3, the fixed width code table 5, and the data output unit 4 in the data compression system shown in FIG. You may comprise by a common system.
[0027]
Next, the data restoration process executed by the restoration algorithm execution unit 12 will be described with reference to the flowchart shown in FIG. In the process of step S10, the restoration algorithm execution unit 12 resets a registered data counter, which is an internal counter, to 1 as an initial setting process, and further sets the registered data to be compared to a column having a predetermined bit width. And stored in the registered data string memory 13.
[0028]
Here, as the registered data to be registered in the registered data string memory 13, if the data of the first row of the input data is output without being compressed in step S0 of the data compression system described with reference to FIG. Also in this step S10, the uncompressed input data of the first row is divided into columns having a predetermined bit width and registered, and the decompression process is started from the data of the second row of the compressed data. When predetermined dummy data is registered in the registered data string memory 3 in step S0 of the data compression system and compression processing is performed from the first row of the input data, the predetermined dummy data is also stored in step S10. Is registered in the registered data string memory 13, and a restoration process is performed from the data in the first row of the compressed data.
[0029]
The decompression algorithm execution unit 12 acquires the Nth input data sequence of the decompression target compressed data input by the data input unit 11 in the process of step S11, and receives the compressed data input in the subsequent process of step S12. Extract the fixed-width code from the column. In the processing of the subsequent step S13, the decompression algorithm executing unit 12 refers to the fixed width code table 15 and extracts a plurality of widths set during compression from the registered data string stored in the Nth column of the registered data string memory 13. In the data of (i), data corresponding to the extracted fixed-width code is extracted, and in the subsequent processing of step S14, based on the data width of the data extracted in step S13, mismatched data, that is, uncompressed data, Is extracted.
[0030]
The decompression algorithm execution unit 12 decomposes the original data by combining the data extracted from the Nth registered data sequence and the uncompressed data extracted from the input data sequence in the subsequent step S15. In the subsequent step S16, the restored data is registered in the Nth column of the registration data column memory 13.
[0031]
The restoring algorithm execution unit 12 determines whether or not the value of the registered data counter N is the maximum address value n (the maximum number of columns n) of the registered data string memory 13 in the subsequent processing of step S17. If the result of the determination is that the value of the registered data string N is not the address value n, the restoration algorithm execution unit 12 adds 1 to the value of the registered data counter N as the processing of step S18, and returns to the processing of step S11. Then, the process proceeds to the next input data string. On the other hand, if the result of the determination indicates that the value of the registered data counter N is the maximum address value n (maximum number of columns n), the restoration algorithm execution unit 12 resets the value of the registered data counter N to 1 as the process of step S19. Then, the process returns to the process of step S11, and shifts to the process for the next input data sequence of the first column.
[0032]
As is apparent from the above description, in the data decompression system according to the present embodiment, the decompression algorithm execution unit 12 decompresses the coded part of the compressed data with reference to the fixed width code included in the compressed data sequence. Then, by combining the restored data with data other than the encoded part of the input data, the data to be restored is sequentially restored. According to such a configuration, data compressed by the data compression system according to the present embodiment can be efficiently and rapidly restored.
[0033]
Although the embodiments of the present invention have been described in detail, the present invention can be implemented in various other forms without departing from the spirit or main features. For example, in the present embodiment, data having a data structure in which 32-bit data is arranged in eight columns in one row and has a correlation (similarity) between adjacent rows for each column is used as a model. Although described, the data structure of the data to be compressed is not limited to this. The number of columns per row may be less than eight columns or eight or more columns, and the data length of one column may be less than 32 bits or more than 32 bits.
[0034]
Also, an example is shown in which the fixed width code is divided into 4 bits, 8 bits, and 20 bits, and the setting pattern is determined by the degree of coincidence in the data width. For example, the divided data width is 8 bits, 8 bits, and 16 bits. The set pattern may be determined according to the degree of coincidence in, and various values are assumed for the set value of the fixed width code.
[0035]
【The invention's effect】
As described above, according to the present invention, an input data sequence is compared with a data sequence at a position correlated with the input data sequence, and the matched position is encoded into a code having a fixed bit width, and the code does not match the code. Since the data sequence at the position is synthesized and compressed as compressed data, data having a correlation for each fixed data unit can be efficiently compressed and restored.
[Brief description of the drawings]
FIG. 1 is a block diagram showing one embodiment of a data compression system according to the present invention.
FIG. 2 is a flowchart for explaining a compression algorithm of a compression algorithm execution unit in FIG. 1;
FIG. 3 is an explanatory diagram illustrating a compression algorithm of a compression algorithm execution unit in FIG. 1;
FIG. 4 is an explanatory diagram showing an encoding example.
FIG. 5 is a block diagram showing one embodiment of a data restoration system according to the present invention.
FIG. 6 is a flowchart for explaining a restoration algorithm of a restoration algorithm execution unit in FIG. 5;
FIG. 7 is a configuration diagram of data having a correlation every fixed period.
[Explanation of symbols]
1, 11 Data input unit 2 Compression algorithm execution unit 3, 13 Registered data string memory 4, 14 Data output unit 5, 15 Fixed width code table 12 Restoration algorithm execution unit

Claims (2)

Means for dividing data to be compressed among the input data into input data strings having a predetermined compression unit bit number,
A storage unit that stores an input data sequence that has been input once to be compared with the input data sequence as a registered data sequence,
Each divided data string obtained by further dividing the input data string divided into the predetermined compression unit bit number into an arbitrary bit number width, and a divided data string at a position correlated with the divided data string stored in the storage means Are compared, the divided data string at the matched position is encoded into a code having a fixed bit width, and the divided data string at a position that does not match the code is synthesized and compressed as compressed data. Compression means for registering in the storage means as a registration data sequence,
Data compression system. A data restoration system for restoring data compressed by the data compression system according to claim 1, wherein a storage unit stores a previous restored data sequence as a registered data sequence,
Extracting a divided restored data string stored in the storage unit based on the encoded data in the compressed data string compressed by the data compression system, and extracting the divided restored data string at a position that does not match the extracted divided restored data string And generating a restored data string by synthesizing the data, and restoring means for registering the generated restored data string as the next registered data string in the storage means.
Data recovery system having.

Images