JP3143029B2 - Data compression method and device, and data decompression method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression method and apparatus for compressing stored data in various storage devices and expanding read data, or compressing and expanding transferred data in digital communication equipment and the like, and an apparatus and data expansion method. A method and an apparatus therefor.

[0002]

2. Description of the Related Art As a code used for encoding for this kind of compression, a "data compression algorithm and its realization" are known.
(Interface, CQ Publisher, Aug. 199
2. , Pp. 105-117) and JP-A-3-68219.
There is a so-called adaptive code as shown in Japanese Unexamined Patent Application Publication No. H10-260, pp. 176-181. This adaptive code has the characteristic that, while reading a character string (input data string), the statistical properties thereof are extracted and coding is performed. A representative universal code which is a kind of the adaptive code is a Lempel-Zip code (also called a slide dictionary method because it is easy to grasp the concept of the encoding).

In this slide dictionary method, a buffer for storing a number of characters in the vicinity of a character currently being encoded from a given character string is prepared, and this buffer is used as a dictionary. It is a method of performing conversion. As the encoding progresses, the character string stored in this buffer slides in a given character string, and is therefore called the slide dictionary method.

Reference to a dictionary in encoding is based on searching for the longest longest matching sequence in a character string in a buffer that matches a character string starting from a position to be encoded at the present time. If such a character string is found, the 1-bit match flag indicating match / mismatch is set to "1", and the position of the found character string in the buffer and its length are represented by an appropriate number of binary digits. The code word is used as the code word. If such a character string does not exist in the buffer, the match flag is set to "0", and the character to be encoded is output as it is (for example, in ASCII code) as a code word.

Here, in the slide dictionary method, data (length indicator) representing the length of the matching data string in the dictionary has a fixed structure corresponding to each length on a one-to-one basis. Therefore, the number of bits is fixed or the number of bits is merely increased by using the extended format when the length becomes a certain value or more.

[0006]

However, in such a conventional technique, when the data before compression is known in advance, the number of remaining data can be represented by a certain number or less, that is, a length indicator. When it is less than 1/2 of the upper limit,
The data size of the length indicator becomes redundant. For example, assuming that the length indicator is 8 bits, the upper limit that can be represented by 8 bits is 255 of “11111111”, so that when the number of remaining data is 1/2 or less, ie, 127, it is “01111111”. Thus, the upper 1 bit becomes redundant. Further, when it becomes 1/2 or less, ie, 63, it becomes "00111111", the upper 2 bits become redundant, and thereafter, every time the number of remaining data becomes 1/2, The upper bits become redundant one bit at a time, and the bits are wasted. This is the compression method (slide dictionary method)
Was also a cause of poor compression ratio for small-size data strings.

Accordingly, the present invention has been made to solve such a problem, and the data at the end of the input data string is equal to or less than 1/2 of the upper limit represented by the length indicator. Remove redundant parts of the compressed data for
It is an object of the present invention to provide a data compression method and its device and a data decompression method and its device with higher compression efficiency.

[0008]

According to the first aspect of the present invention, a predetermined number of data strings in the vicinity of data to be coded at each time point in an input data string are stored in a storage means. The data string in the dictionary is used as a dictionary to refer to for encoding for the encoding, and matches the data string starting from the position to be encoded at each time,
The longest matching data string is searched below the predetermined upper limit within the range of the predetermined quantity, and if the matching data string is found, a flag indicating a match, a pointer indicating the position of the found data string in the dictionary, and A codeword is composed of a length indicator indicating the length, and if the matching data string is not found, a codeword is composed of a flag indicating mismatch and the data to be encoded itself. In the data compression method for compressing data, the number of remaining data in the input data string is counted, and when the number of remaining data is equal to or less than the upper limit, the minimum number of remaining data that can be represented by a binary number is calculated. The number of bits of the length indicator is set as the number of bits, and encoding is performed using the set number of bits of the length indicator.

Similarly, according to the present invention, the input data sequence stores a predetermined number of data sequences in the vicinity of the data to be encoded at each time in the storage means, and stores the stored data sequence. A data string in the dictionary that matches a data string starting from the position to be encoded at each point in time, using the string as a dictionary to be referenced for encoding, and a predetermined upper limit within the range of the predetermined quantity. The longest matching data string is searched below, and if the matching data string is found, a match indicating flag, a pointer indicating the position of the found data string in the dictionary, and a length indicator indicating its length are obtained. If the code word is formed and the above-mentioned matching data string is not found, the code word is formed from the flag indicating the mismatch and the data to be coded by itself to thereby compress the data. In the compression device, a counting means for counting the number of remaining data of the input data sequence, and based on the number of remaining data counted by the counting means, when the number of remaining data is equal to or less than the upper limit, the number of data is converted to a binary number. Setting means for setting the number of bits of the length indicator to the minimum number of bits that can be represented by: and encoding means for performing encoding with the number of bits of the length indicator set by the setting means. It is a thing.

According to a third aspect of the present invention, in the data compression apparatus of the second aspect, a program or data written in the nonvolatile semiconductor memory is compressed.

According to a fourth aspect of the present invention, there is provided the data compression method according to the first aspect of the present invention or the data compressed by the data compression apparatus according to the second aspect of the present invention, wherein the flags included in the codewords do not match. Indicates that other data included in the code word is taken out and decoded, the data is input to a predetermined number of dictionaries to be referred for decoding, and the flag included in the code word matches. Is indicated, the corresponding data is taken out from the dictionary using the pointer and the length indicator included in the code word and decoded, and the data is input to the dictionary to obtain compressed data. A data decompression method for decompression, wherein the number of remaining data of decompressed data is counted, and when the number of remaining data falls below a predetermined upper limit within a range of the predetermined quantity, the remaining data is counted. The number over data to determine the number of bits of the minimum of the length from the number of bits indicator that can be represented by a binary number, in which to perform decoding using the number of bits of the discriminated length indicator.

Similarly, according to a fifth aspect of the present invention, the compressed data compressed by the data compression method according to the first aspect or the data compression apparatus according to the second aspect is input, and the flags included in the code words do not match. Indicates that other data included in the codeword is taken out and decoded,
The data is input to a predetermined number of dictionaries to be referred to for decoding, and when a flag included in the codeword indicates a match, the above data is obtained by using a pointer and a length indicator included in the codeword. A data decompression device for extracting and decoding corresponding data from the dictionary and inputting the data to the dictionary, thereby decompressing the compressed data, and a counting means for counting the number of remaining data of the decompressed data. ,
Based on the number of remaining data counted by the counting means, when the number of remaining data is equal to or less than a predetermined upper limit within the range of the predetermined quantity, the number of data is calculated from the minimum number of bits that can be expressed in a binary number. It is provided with a discriminating means for discriminating the number of bits of the length indicator, and a decoding means for performing decoding using the number of bits of the length indicator discriminated by the discriminating means.

According to a sixth aspect of the present invention, in the data decompression apparatus of the fifth aspect, the compressed data read from the nonvolatile semiconductor storage means in which the compressed data of the program and the data is stored is expanded. .

[0014]

According to the data compression method of the first aspect, the size (the number of bits) of the length indicator indicating the length of the longest matching data string in the compressed data is determined by the remaining data number of the input data string. Is set to the minimum number of bits that can be represented by a binary number when the number of bits becomes less than or equal to the upper limit of the length of the input data string. Remove redundant parts of compressed data,
Compression efficiency can be improved.

According to the second aspect of the present invention, the data compression apparatus has the same function as that of the first aspect and can perform compression at a high speed.

According to the third aspect of the present invention, since the program and data written in the nonvolatile semiconductor memory are compressed, an expensive storage element can be used efficiently.

On the other hand, according to the data decompression method of the fourth aspect, when the number of remaining data of the decompressed data is equal to or less than a predetermined upper limit, the number of bits of the length indicator in the compressed data is set to two. Since the decoding is performed by determining the minimum number of bits that can be represented by a base number, the data compressed as described above can be expanded.

Further, according to the data decompression device of the fifth aspect, it is possible to perform the decompression at a high speed while having the same operation as that of the fourth aspect.

Further, according to the data decompression device of the present invention, it is possible to decompress a program or data read from the nonvolatile semiconductor storage means at high speed.

[0020]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a data compression device according to one embodiment of the present invention. In the figure, reference numeral 1 denotes an input data buffer for temporarily storing an input data string. This input data buffer 1 has a data bus width of 8 in this embodiment.
It is composed of a shift register of 256 bytes in total, and has an address FFh (h is 16 bits).
In this case, by inputting data at address 255, that is, the highest address, the data at each address is shifted toward the lower address. When a search end signal (trigger) is received from the longest match data string search device 3 to be described later, when the match data string presence / absence flag is “1”, new data is added by the number of bytes indicated by the match data string length. And the data in the buffer is sequentially shifted by the number of bytes. When the match data string presence / absence flag is "0", new data is input by one byte, and the data in the buffer is shifted by one byte.

Reference numeral 2 denotes a dictionary buffer for storing a slide dictionary based on the slide dictionary method described above. In this embodiment, the dictionary buffer 2 also has a data bus width of 8 bits and an address of 8 bits, and has a shift register of 256 bytes in total. It is composed of When input data is empty in the dictionary buffer 2, that is, dictionary size (dic_size) <2
At the time of 56, the dictionary size (dic_size) is latched as an address (D (dic_size) =
Input data). On the other hand, when the dictionary buffer 2 is full, that is, when the dictionary size (dic_size) = 256,
First, the dictionary data is shifted toward address 00h (D
(N) = D (n + 1), n = 1, 2,..., 254)
After that, it is latched at the address FFh. At the time of the longest match data string search described later, data indicating an address is output under the control of the longest match data string search device 3, and dictionary data is searched.

The longest matching data string search device 3 extracts a data string of 2 bytes or more that matches the data string starting from the address 00h of the input data buffer 1 longest in the 256-byte dictionary buffer 2 as described later. Dictionary size (dic_siz) output from the dictionary size counter 4
The search is performed from the area indicated by e). As a result of this search, if a match data string is found, the match data string presence / absence flag is set to “1”, and a match data string pointer (crspd_point) indicating the dictionary position (address) of the first data of the match data string ) And a match data string length (crspd_len), which is a length indicator indicating the length (the number of bytes) of the match data string, to the encoding device 7 described later, and outputs the match data string to the dictionary buffer 2. . If not found, the match data string presence / absence flag is set to “0” and one byte of raw data is output to the encoding device 7 and the dictionary buffer 2.

In this embodiment, when the longest match data string search device 3 outputs a data string to the dictionary buffer 2, an increment signal is output to the dictionary size counter 4 every time one byte is output, and the remaining data is output. The decrement signal is output to the number counter 5. When the output is completed, the longest match data string search device 3 outputs a search end signal (trigger) to the encoding device 7 and ends one longest match data string search process. Further, upon receiving an encoding end signal (trigger) from the encoding device 7, the longest match search process described above is started again.

The dictionary size counter 4 is composed of an up counter having a maximum count value set to 256, and is reset by a system reset signal (not shown) at the start of compression. Then, it is counted up by the increment signal from the maximum matching data string search device 3 described above,
When the maximum value (256) is reached, that value is retained thereafter.
The output of this dictionary size counter 4 (dictionary size dic_
size) is output to the longest match data string search device 3.

The remaining data number counter 5 corresponds to the counting means provided in the present invention, and is constituted by a down counter having a bit number capable of setting the maximum data number assumed in the field of application of the compression apparatus. Before the start of compression, the number of data in the input data sequence is set. Thereafter, each time the decrement signal is received from the longest matching data string search device 3, the countdown is performed, and the number of remaining data is output to the length indicator data size generator 6.

Length indicator data size generator 6
Is equivalent to the setting means provided in the present invention, is constituted by an encoder composed of a gate circuit, and outputs an output value from the remaining data number counter 5 (remaining data number r
A corresponding length indicator data size (len_size) is generated based on (em_size) and output to the encoding device 7. This length indicator data size generator 6
FIG. 2 shows a correspondence table between the input (remaining data number rem_size) and the output (length indicator data size). Even when the number of remaining data is 257 or more, the upper limit of the maximum matching data string length is 256, so the maximum value of the length indicator is 255. When the number of remaining data is 1, there is no matching data string of 2 bytes or more, so that the matching data string presence / absence flag is always 0, and therefore the length indicator data size is also 0.

The encoding device 7 corresponds to the encoding means in the present invention, and is obtained by adding a function specific to the present application described later to a conventional encoding device. The encoding device 7 starts the encoding process by using the search end signal from the longest matching data string retrieval device 3 as a trigger, and outputs the encoded compressed data bit stream together with the output data synchronization clock. That is, first, a 1-bit flag indicating whether a matching data string is present or not is output. Next, when this flag is "0", 8 bits (1 byte) of raw data are sequentially output from the LSB (least significant bit). When the flag is “1”, first, the matching data string length (crspd
_Len) 8-bit (1-2) obtained by subtracting 1 from 2-256
55), the output is sequentially performed from the LSB by the number of bits represented by the length indicator data size output from the length indicator data size generator 6. Next, 8 bits of a matching data string pointer (crspd_point) are sequentially output from the LSB. When the output of these compressed data bit streams ends, an encoding end signal is output to the longest match data string search device 3.

Next, the operation of the data compression apparatus configured as described above will be described. An input data string to be compressed is temporarily stored in a 256-byte input data buffer 1. Here, for example, “abcabd... Ab
The following description is made by taking an example of a 256-byte input data string of c ″. At the start of compression, the dictionary size counter 4 is reset to 0 by a system reset signal.
In addition, 256 is set in the remaining data counter 5 as the number of input data.

The longest match data string search device 3 executes the longest match data string search processing according to the algorithm shown in the flowchart of FIG. That is, first, step 10
1, the match data string length (cr) of 9 bits (0 to 256)
spd_len) is initialized to 0, and an 8-bit (0-255) dictionary pointer (dic_point) indicating the address of the dictionary buffer 2 is initialized to 0, that is, to the beginning of the dictionary. In step 102, the temporary (temporary) match data string length (comp_len) is set to 0.
To 8 indicating the address of the input data buffer 1
The input data buffer pointer (str_point) of the bit (0-255) is initialized to 0, that is, initialized to the beginning of the input data sequence.

Next, in step 103, the input data (S) corresponding to the input data buffer pointer (str_point), that is, "a" and the dictionary pointer (di)
It is determined whether or not the dictionary data (D) corresponding to (c_point) matches. Since data has not yet been input to the dictionary buffer 2, the data does not match and the process proceeds to step 104, where the dictionary pointer (dic_point) is incremented. At step 105, the dictionary pointer (dic_point) outputs the output of the dictionary size counter 4. It is determined whether or not the dictionary size (dic_size) is equal to or larger than the dictionary size (dic_size). Since the dictionary size counter 4 is reset to 0 by the system reset signal at the start of compression, the determination in step 105 is (Y) and the process proceeds to step 113.

In step 113, the matching data string length (c
rspd_len) is 2 or more. That is, if the length of the matched data string is 1 byte or less, there is no point in compressing the data.
The no-flag (crspd_flag) is set to 0, and if the matching data string length is 2 bytes or more, a valid matching data string is detected, and the routine proceeds to step 114, where the flag is set to 1. Since the matching data string length (crspd_len) is still 0 while being initialized in step 101, the process proceeds from step 113 to step 115 to set the matching data string presence / absence flag (crspd_flag) to 0. At this time, one byte of raw data "a", which is the head data of the input data sequence, is output to the encoding device 7 and the dictionary buffer 2, and an increment signal is output to the dictionary size counter 4 to output the remaining data number counter 5 To output a decrement signal. Then, the process proceeds to step 116, where a search end signal is output to the encoding device 7 and the input data buffer 1, and the first search process ends.

The encoding device 7 uses the search end signal from the longest matching data string search device 3 described above as a trigger, and
The encoding process shown in the flowchart is started. That is, in step 201, the match data string presence / absence flag 1 bit from the longest match data string search device 3 is output as a compressed data bit stream. Next, at step 202, it is determined whether or not the coincidence data string presence / absence flag is 1.
Now, since the flag is 0, the process proceeds to step 203, where 8 bits (1 byte “a”) of the raw data from the longest matching data string search device 3 are sequentially output from the LSB (least significant bit). Then, the process proceeds to step 206, where an encoding end signal is output to the longest matching data string search device 3, and the first encoding process is ended.

The longest match data string search device 3 is started with the coding end signal from the coding device 7 as a trigger, and the processing shown in the flowchart of FIG. 3 is executed. Here, also in the second and third longest match data string search processing, the leading data of the input data string is “b”,
Since "c" is set, the data matching the head data of the input data string is searched from the dictionary in steps 103 to 103.
No matching data string is found in the processing of No. 5, and the same search processing and encoding processing as described above are performed. Then, the dictionary buffer 2 stores 3 bytes “ab” from the beginning of the input data string.
c ”is stored, the dictionary size (dic_size) output from the dictionary size counter 4 indicates 3, and the remaining data number (rem_si) output from the remaining data counter 5
ze) is 253, and the length indicator data size (len__) output from the length indicator data size generator 6
size) remains at 8 (see FIG. 2).

In the fourth search process, the head data of the input data string is "a". In step 103 of the flowchart of FIG. 3, the input data (S) corresponding to the input data buffer pointer (str_point), that is, "a" and the dictionary pointer (dic_p
oint), that is, the dictionary data (D), that is, "a" matches. Therefore, the process proceeds to step 106, where the input data buffer pointer (str_point)
t), the dictionary pointer (dic_point), and the temporary match data string length (comp_len) are each incremented to 1, respectively. Next, in step 107, it is determined whether or not the dictionary pointer (dic_point) has become equal to or larger than the dictionary size (dic_size). Since the dictionary pointer is 1 and the dictionary size is 3, the process proceeds to step 108. In this step 108,
The input data (S) corresponding to the input data buffer pointer (str_point) and the dictionary pointer (dic_p)
oint) is determined to be identical to the dictionary data (D) corresponding to "b", so that the process returns to step 106, where the above-described increment is performed, and the input data buffer pointer (str_point) and the dictionary pointer (Dic_point) and the temporary match data string length (comp_len) are 2 each.

In the next step 107, the dictionary pointer is set to 2
Does not become 3 or more of the dictionary size.
Proceed to. Here, the input data buffer pointer (str)
_Point) is "d", and the dictionary data (D) corresponding to the dictionary pointer (dic_point) is "c".
09, the temporary match data string length (comp_l
en) is longer than the matching data string length (crspd_len). The temporary match data string length (comp_len) is 2, and the match data string length (crs
pd_len) is 0 as it is initialized in step 101, so the process proceeds to step 110. Step 110
Then, the value 2 of the temporary match data string length (comp_len) is substituted for the match data string length (crspd_len), and the dictionary pointer (dic) is assigned to the 8-bit (0-255) match data string pointer (crspd_point).
_Point) to the match data string length (crspd_le)
n) is substituted by 0, and the dictionary pointer (di
c_point) is the matching data string length (crspd_le)
The value is decremented by 1 (here, 1) obtained by subtracting 1 from n), and the process returns to step 102. If the dictionary pointer becomes equal to or larger than the dictionary size in step 107, steps 111 and 112 execute step 10
Update processing of the matching data string length similar to that of steps 9 and 110 is performed. At the time of the determination in step 109, a value has already been substituted for the matching data string length (crspd_len).
This value is the temporary match data string length (comp_le
If the value is equal to or more than the value of n), the process returns to step 102 without performing the process of step 110.

In step 102, the temporary match data string length (comp_len) is initialized to 0, and the input data buffer pointer (str_po) is set.
int) is initialized to the beginning of the input data sequence. At this point, the leading data of the input data string is “a”, the data corresponding to the value 1 of the dictionary pointer (dic_point) is “b”, and there is no matching data thereafter. When the value of the dictionary pointer becomes 3 in step 104 after making one round of the loop of 105, the value becomes equal to the dictionary size value 3 in step 105, and the process proceeds to step 113. In step 113, since the value of the matching data string length (crspd_len) is 2, the condition of 2 or more is satisfied, and the routine proceeds to step 114, where the matching data string presence / absence flag is set to "1". At this time, the value 0 of the match data string pointer (crspd_point) indicating the position (address) of the head data of the match data string in the dictionary
And the value 2 of the matching data string length (crspd_len) indicating the length (the number of bytes) of the matching data string is output to the encoding device 7, and the matching data string "ab" is output to the dictionary and one byte is output. , An increment signal is output to the dictionary size counter 4 and a decrement signal is output to the remaining data counter 5. Then, the process proceeds to step 116 to output a search end signal to the encoding device 7 and the input data buffer 1, and ends the current search process.

Similarly, the encoding device 7 starts the encoding process shown in the flowchart of FIG. 4 by using the search end signal as a trigger. In step 201, 1 bit of a match data string presence / absence flag from the longest match data string search device 3 is output as a compressed data bit stream. Next, at step 202, it is determined whether or not the coincidence data string presence / absence flag is 1. Now, since the flag is 1, step 2
04, the 9-bit matching data string length (2 to 25)
6) minus 1 and lower 8 bits (1 to 255) are LS
B is sequentially output for the number of bits of the length indicator data size. Now, since the length indicator data size is 8, all 8 bits are output. Next, the routine proceeds to step 205, where 8 bits of a coincidence data string pointer (crspd_point) are sequentially output from the LSB. And step 206
And the encoding end signal is sent to the longest matching data string search device 3
, And the current encoding process ends. At this time, the compressed data bit stream output from the encoding device 7 has a total of 17 bits including a match data string presence / absence flag of 1 bit, a match data string length of 8 bits, and a match data string pointer of 8 bits. In this case, compression is meaningless, but, for example, in the case where the input data string matches 3 bytes where "abcabc...", 24 bits are compressed to 17 bits.

When the number of remaining data in the input data string becomes 128 bytes or less by repeating the above processing, the number of remaining data (r
em_size) is 128 or less, and the length indicator data size (lem_size) output from the length indicator data size generator 6 is 7 or less (see FIG. 2). Therefore, the compressed data bit stream in the case where the matching data string is found is the flag 1 indicating whether the matching data string is present or not.
The number of bits, the matching data string length is 1 to 7 bits, and the matching data string pointer is 8 bits, for a total of 10 to 16 bits. More specifically, if "abc" at the end of the input data string matches "abc" from the beginning of the dictionary buffer 2, the number of remaining data becomes three and the length indicator data size becomes two. 2), a match data string presence / absence flag is 1 bit, a match data string length is 2 bits, and a match data string pointer is 8 bits, for a total of 11 bits. That is, 3 bytes (24 bits) are compressed to 11 bits, and in the related art, 17 bits remain even in this case. Therefore, the compression processing of this embodiment has a much better compression ratio. I understand.

The format of the compressed data bit stream output from the encoding device 7 by repeating the longest matching data string search process and the encoding process as described above is as shown in FIG. That is, in FIG.
Fn is one bit of a compression flag (corresponding data string presence / absence flag) for the following data Dn. "0" represents non-compression data, and "1" represents compression data.
Data Dn is data for the flag Fn.
When the flag Fn is "0", the raw data is 1 byte (LSB 8 bits), and when the flag Fn is "1", the compressed data is in the range of 8 to 16 bits. When the flag Fn = 1, the compressed data Dn has a matching data string length (cr
spd_len) are arranged in order from the LSB, followed by an 8-bit matching data string pointer (crspd_poi).
nt) are arranged in order from the LSB. The 8-bit match data string pointer (crspd_point) indicates the start address of the match data string in the dictionary.
The bit match data string length (crspd_len) is 1
255 represents the length of 2-256, and corresponds to the number of remaining data as shown in FIG. That is, when the number of remaining data becomes 128 bytes or less, the matching data string length (cr
Only the corresponding lower bits of (spd_len) are required, and the other upper bits are redundant bits and become unnecessary, and the length indicator data size is 1 to 7 bits. Therefore, the compression ratio is improved by that much as compared with the prior art.

Therefore, by applying this data compression device (method) to a communication device or a storage device in which the number of data is known in advance, the amount of transfer data and the amount of storage data are smaller than in the prior art. It is possible to improve the speed and reduce the storage capacity. In particular, the present invention is effective when compressing a data string of a small size (compared to a dictionary size) that the slide dictionary method targeted by the present invention is not good at.
As for the reduction of the storage capacity, an expensive storage element can be efficiently used in, for example, a game cassette using a nonvolatile semiconductor storage element such as a ROM.
Further, in this game cassette, a large number of small-sized programs and data are stored, which is more effective.

In the above embodiment, the data compression apparatus is realized in order to increase the speed of compression. However, in a field where high speed is not required much, for example, writing of a program or data to a ROM for a game cassette, the above-mentioned data compression apparatus is used. The above processing may be realized by software and executed by a computer.

FIG. 6 is a block diagram showing a data decompression device according to one embodiment of the present invention, which decompresses a compressed data bit stream compressed and transferred or stored by the above-described data compression device.

Reference numeral 11 denotes a demultiplexer corresponding to the decoding means in the present invention, which is a conventional demultiplexer to which a function specific to the present application described later is added. The demultiplexer 11 receives the compressed data bit stream shown in FIG. 5 as input, determines whether the data is compressed or uncompressed based on a flag Fn in the first bit of the input data, and inputs data (Dn) of a required number of bits. . That is, if the data is uncompressed (Fn = 0), the next 8 bits are stored and output as byte data to the decompressed data string, and are written into the dictionary buffer 12. The output of the decompressed data string is performed together with the output data synchronization clock, and the writing to the dictionary buffer 12 is performed by the write clock and the shift clock. Writing to the dictionary buffer 12
If there is free space in the dictionary buffer 12, (dic_size <
256), the write address is the dictionary size dic_siz
e, and after writing, an increment signal is output to the dictionary size counter 13 and the remaining data number counter 14
To output a decrement signal. Dictionary buffer 1
2 is full (dic_size = 256), the dictionary data is first shifted to address 00h (D
(N) = D (n + 1), n = 0, 1,..., 254)
Then, write to the address FFh. On the other hand, compression (F
n = 1), the number of bits of the data Dn following this is
The length indicator data size output from the length indicator data size generator 15 is referred to, and “length indicator data size (len_size) +8” is determined. Then, first, input is performed for the number of bits of the length indicator data size (len_size), and the upper bits (8-len_less) that are less than 8 bits
The “size” bit is supplemented with “0” to form an 8-bit matching data string length (crspd_len). The subsequent 8 bits are input to form a 1-byte matching data string pointer (crspd_point). The matching data string in the dictionary buffer 12 is referred to based on the matching data string length (crspd_len) and the matching data string pointer (crspd_point) thus obtained.
First, the match data string pointer (csrpd_point)
Dictionary data D (csrpd_po) at the address indicated by t)
int). This data is output as decompressed data, and the matched data string length (crspd_len) is decremented. The decompressed data D (csrpd_
point) is written in the dictionary buffer 12. As described above, if there is a space in the dictionary buffer 12, (dic_siz
e <256), the write address is the dictionary size dic_s
size (D (dic_size) = D (crsp
d_point). Then, while outputting an increment signal to the dictionary size counter 13 and a decrement signal to the remaining data counter 14, the coincidence data string pointer (crspd_point) is incremented. On the other hand, if the dictionary buffer 12 is full (di
c_size = 256).
0h (D (n) = D (n + 1), n =
0, 1,..., 254) and then write to the address FFh (D (255) = D (crspd_point)
t)). Then, a decrement signal is output to the remaining data number counter 14, and the coincident data string pointer (crspd) is output.
_Point) is not incremented. If a borrow occurs in the decrementing process of the matching data string length (crspd_len), one decoding process ends, and the next flag of the input data is checked. That is, since the matching data string length (crspd_len) represents the value of (actual matching data string length) -1, the actual matching data string does not reach 00h but reaches a borrow and reaches FFh. It is a long process.

The dictionary buffer 12 is the same as the dictionary buffer 2 of the data compression device, and is composed of a shift register of 8 bytes for the data bus and 8 bits for the address and a total of 256 bytes. As described above, the input data from the demultiplexer 11 is used when the dictionary buffer 12 has a space, that is, the dictionary size (dic_size) <
In the case of 256, this dictionary size (dic_size) is latched as an address (D (dic_size)
= Input data). On the other hand, when the dictionary buffer 12 is full,
That is, when the dictionary size (dic_size) = 256, the dictionary data is first shifted toward the address 00h (D (n) = D (n + 1), n = 1, 2,..., 25
4) and then latched at address FFh. Further, under the control of the demultiplexer 11, the data at the specified address (crspd_point) is read and output to the demultiplexer 11.

The dictionary size counter 13 is constituted by an up counter having a maximum count value set to 256, and is reset by a system reset signal (not shown) at the start of decompression. Then, the count is incremented by the increment signal from the demultiplexer 11 described above, and when the count reaches the maximum value (256), the value is held thereafter. The dictionary size (dic_s) from the dictionary size counter 13
ize) is output to the demultiplexer 11.

The remaining data number counter 14 corresponds to the counting means provided in the present invention, and is constituted by a down counter having a bit number capable of setting the maximum data number assumed in the application field of the present decompressor. , The number of data in the input data string is set before the start of decompression, and thereafter, the countdown is performed by the decrement signal from the demultiplexer 11, and the number of remaining data (rem_size) is output to the length indicator data size generator 15.

Length indicator data size generator 15
Is equivalent to the discriminating means provided in the present invention, is constituted by an encoder composed of a gate circuit, and has a corresponding length indicator based on the output value (remaining data number) from the remaining data number counter 14. Data size (le
n_size), that is, the number of bits of the matching data string length is generated and output to the demultiplexer 11. The correspondence between the input (the number of remaining data) and the output (length indicator data size) of the length indicator data size generator 15 is as shown in FIG.

Next, the operation of the data decompression device configured as described above will be described. Here, a description will be given assuming that the compressed data bit stream according to the specific example used in the description of the operation of the data compression apparatus is expanded. This compressed data bit stream is represented by Fn and Dn in FIG.
F0 = 0, D0 = “a”, F1 = 0, D1 = “b”, F
2 = 0, D2 = “c”, F3 = 1, D3 = 000000
010000000000, F4 = 0, D4 = “d”,.
.., Fmax = 1 and Dmax = 10000000000. At the start of decompression, the dictionary size counter 13
Is reset to 0 by the system reset signal, and 256 is set in the remaining data counter 14 as the number of input data.

When the data decompression device is started and the above-described compressed data bit stream is input to the demultiplexer 11, the demultiplexer 11 starts the processing shown in the flowchart of FIG. That is, step 301
To determine whether the first bit (flag F0) of the input data string is 1 or not. Since the first bit F0 of the input data is 0, the process proceeds to step 302, where the next 8 bits (D0 =
“A”) is input and output to the decompressed data string as byte data. Next, at step 303, it is determined from the dictionary size (dic_size) whether or not there is a free space in the dictionary buffer 12, and if there is a free space (dic_size <25).
6) Proceeding to step 304, the dictionary size counter 1
The byte data is written in the dictionary buffer 12 using the dictionary size (dic_size) indicated by No. 3 as an address.
Now, since the dictionary size (dic_size) is 0,
Data D0 = at address 00h of dictionary buffer 12
“A” is written. Thereafter, in step 305, an increment signal is output to the dictionary size counter 13 to count up, and in step 307, a decrement signal is output to the remaining data counter 14 to count down, and the process returns to step 301. Note that the dictionary buffer 12
If there is no free space (dic_size = 256), the process proceeds to step 306, where the dictionary data is shifted toward the address 00h and then written to the address FFh.

Next input data flag F1 = 0, data D1 = "b", next flag F2 = 0, data D2 =
Since it is “c”, steps 301 to 305 and 307 are repeated twice, and “b”, “b”,
“C” is output. Now, in the dictionary buffer 12, "a", "b", "c" are stored in order from address 00h, the dictionary size (dic_size) output from the dictionary size counter 13 is 3, and the remaining data number counter 14 (rem_s
(size) is 253, and the length indicator data size (len_size) output from the length indicator data size generator 15 remains at 8 (see FIG. 2).

Since the flag F3, which is the first bit of the next input data, becomes 1, the flow advances from step 301 to step 308 to input the number of bits of the length indicator data size, that is, 8 bits (00000001). ,
The 8-bit matching data string length is configured as it is. In the next step 309, 8 bits (00000000) are input to form a 1-byte matching data string pointer. Next, the routine proceeds to step 310, where the data "a" of the coincidence data string pointer 00h is read from the dictionary buffer 12 and output as decompressed data. And step 31
In step 1, the matching data string length (00000001) is decremented. In step 312, a decrement signal is output to the remaining data counter 14. Next, step 3
At 13, it is determined from the dictionary size (dic_size) whether or not there is an empty space in the dictionary buffer 12.
_Size) as an address to write the decompressed data to the dictionary buffer 12. Now, the dictionary size (dic_siz
Since e) is 3, the address 03 of the dictionary buffer 12 is
Decompressed data "a" is written to h. Thereafter, in step 315, an increment signal is output to the dictionary size counter 13 to count up. In step 316, the coincidence data string pointer (00h) is incremented.
Proceed to step 318. If there is no space in the dictionary buffer 12 (dic_size = 256), step 3
17, the dictionary data is shifted toward the address 00h, and then written to the address FFh.
Proceed to 8.

At step 318, it is determined whether or not the matching data string length has a borrow. Matching data string length is 00
Since no borrow has yet been made at h, the process returns to step 310. Then, similarly to the above, the data “b” of the matching data string pointer 01h is read from the dictionary buffer 12, and
Output as decompressed data (step 31
0), decrement the coincident data string length 00h (step 311), and output a decrement signal to the remaining data counter 14 (step 312) to write the data "b" into the dictionary buffer 12 (step 311). 314). Then, an increment signal is output to the dictionary size counter 13 (step 315), the coincidence data string pointer is incremented (step 316), and the routine proceeds to step 318. Here, the matching data string length has a borrow due to the decrement in step 310, and
Since it is Fh, the process returns to the first step 301.
Then, the next leading bit of the input data string (flag F4)
Is 0, the process proceeds to step 302, and the same processing as described above is repeated.

When such processing is repeated and the first bit Fmax of the input data becomes 1, the process proceeds from step 301 to step 308. In step 308, the number of bits corresponding to the length indicator data size is input. Here, the number of remaining data (rem_size) is 3, and
Since the length indicator data size (len_size) from the length indicator data size generator 15 is 2, the next 2 bits, ie, “10”, are input, and “0” is added to the upper bits that are less than 8 bits. This is supplemented to form an 8-bit matching data string length (00000010). In the next step 309, 8 bits (00000000)
To form a 1-byte matching data string pointer. Next, the routine proceeds to step 310, where the data "a" of the coincidence data string pointer 00h is read from the dictionary buffer 12 and output as decompressed data. Then, in step 311, the matching data string length (00000010) is decremented. In step 312, a decrement signal is output to the remaining data counter 14. Next, in step 313, it is determined from the dictionary size (dic_size) whether or not the dictionary buffer 12 has an empty space. Size (di
Write the decompressed data to the dictionary buffer 12 using c_size) as an address. Now, the dictionary size (dic_siz
Since e) is 253, the decompressed data "a" is written to the address FDh of the dictionary buffer 12. Thereafter, in step 315, an increment signal is output to the dictionary size counter 13 to count up. In step 316, the coincidence data string pointer (00h) is incremented, and the flow advances to step 318. Now, the match data string length is 01
h, the above steps 310 to 316 are repeated twice, and when the matching data string length becomes FFh (borrow), decoding processing of all input data ends, and “abcabd. Is obtained, and the input data sequence before being compressed by the above-described data compression device can be restored.

In the above embodiment, the data decompression device is implemented to speed up the decompression process. However, in a field where high speed is not required much, the above-described process is realized by software and executed by a computer. You may do it.

Further, since the number of data is known in advance, the RO
In a game cassette or the like in which a large number of small-sized programs and data are stored in expensive nonvolatile semiconductor storage means such as an M or a battery-backed RAM and the like and high-speed reading is required, the above-described data decompression device should be incorporated. Is particularly effective because the above conditions are satisfied.

Further, in each of the above embodiments, the case where shift registers are used for the dictionary buffers 2 and 12 of the data compression device and the data decompression device has been shown. May be sequentially updated to sequentially update the contents of the dictionary by overwriting, thereby realizing a slide dictionary.

[0057]

According to the first aspect of the present invention, in the data compression method for compressing data using the slide dictionary method, the number of remaining data in the input data string is counted, and the remaining data is in the range of the dictionary capacity. If the number of remaining data is less than the specified upper limit, set the number of bits of the length indicator to the minimum number of bits that can represent the remaining data in binary, and set the number of bits of the set length indicator to The encoding is performed at the end of the input data sequence.
It is possible to remove a redundant portion of the compressed data for data having a length equal to or less than / 2, and to increase the compression ratio as compared with the related art. Therefore, by applying the present invention to a communication device or a storage device of which the number of data is known in advance, the transfer data amount and the storage data amount are reduced as compared with the conventional technology, so that the transfer speed is improved and the storage capacity is reduced. Can be achieved. In particular, the present invention is effective when compressing a data string of a small size (compared to a dictionary size) that the slide dictionary method targeted by the present invention is not good at.

According to a second aspect of the present invention, there is provided a data compression apparatus for compressing data using the slide dictionary method, wherein a counting means for counting the number of remaining data in the input data sequence, The number of bits of the length indicator is set to the minimum number of bits that can represent the number of data in a binary number when the number of remaining data is equal to or less than a predetermined upper limit within the range of the dictionary capacity based on the number of remaining data that has been set And setting means for setting the number of bits of the length indicator set by the setting means, so that the same effect as in claim 1 can be obtained, and This can be performed at a high speed.

According to the third aspect of the present invention, since programs and data written in the non-volatile semiconductor storage means are compressed, there is an effect that an expensive storage element can be used efficiently.

According to a fourth aspect of the present invention, there is provided a data decompression method for decompressing data compressed by the data compression method according to the first aspect or the data compression apparatus according to the second aspect. The remaining data number is counted, and when the remaining data number becomes equal to or less than a predetermined upper limit within the range of the dictionary capacity, the remaining data number is calculated from the minimum number of bits that can be represented by a binary number to a length indicator. Is determined, and decoding is performed using the determined number of bits of the length indicator, so that the data compressed as described above can be expanded.

According to a fifth aspect of the present invention, there is provided a data decompression device for decompressing data compressed by the data compression method according to the first aspect or the data compression device according to the second aspect. Counting means for counting the number of remaining data of the data; and, based on the number of remaining data counted by the counting means, when the number of remaining data falls below a predetermined upper limit within the range of the dictionary capacity, the number of data is reduced by 2 Discriminating means for discriminating the number of bits of the length indicator from the minimum number of bits that can be represented by a hexadecimal number; and decoding means for decoding using the number of bits of the length indicator discriminated by the discriminating means. With this arrangement, the same effect as that of the fourth aspect is obtained, and also, there is an effect that the extension can be performed at a high speed.

Further, according to the invention of claim 6, there is an effect that the compressed data of the program or data read from the nonvolatile semiconductor storage means can be decompressed at high speed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a data compression device according to one embodiment of the present invention.

FIG. 2 is a diagram showing a correspondence table between a remaining data number and a length indicator data size in the embodiment.

FIG. 3 is a flowchart showing the operation of the longest match data string search device in the embodiment.

FIG. 4 is a flowchart showing an operation of the encoding device in the embodiment.

FIG. 5 is an explanatory diagram of a compressed data bit stream in the embodiment.

FIG. 6 is a block diagram showing a data decompression device according to one embodiment of the present invention.

FIG. 7 is a flowchart showing the operation of the demultiplexer in the embodiment.

[Explanation of symbols]

 2,12 Dictionary buffer 3 Longest matching data string search device 4,13 Dictionary size counter 5,14 Remaining data number counter 6,15 Length indicator data size generator 7 Encoding device 11 Demultiplexer

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-1-314430 (JP, A) JP-A-2-233025 (JP, A) JP-A-3-68219 (JP, A) JP-A-3-32419 204232 (JP, A) JP-A-3-204233 (JP, A) JP-A-3-78322 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H03M 7/40

Claims (6)

(57) [Claims] 1. A dictionary for storing a predetermined number of data strings in the input data string in the vicinity of data to be encoded at each time in a storage unit, and referencing the stored data strings for encoding. The data string in the dictionary that matches the data string starting from the position to be encoded at each point in time is searched for the longest matching data string that is equal to or less than a predetermined upper limit within the range of the predetermined quantity. If the matching data string is found, a code word is formed from a flag indicating a match, a pointer indicating the position of the found data string in the dictionary, and a length indicator indicating the length of the matching data string. If a column is not found, the data compression method for compressing data by constructing a codeword from a flag indicating a mismatch and the data to be coded itself. Of counting the number of remaining data, if the number of remaining data is equal to or less than the upper limit, the number of remaining data 2
A data compression method comprising: setting the number of bits of the length indicator to the minimum number of bits that can be expressed in a base number; and performing encoding using the set number of bits of the length indicator. 2. A dictionary for storing a predetermined number of data strings in the input data string in the vicinity of data to be encoded at each time in a storage unit, and referencing the stored data strings for encoding. The data string in the dictionary that matches the data string starting from the position to be encoded at each point in time is searched for the longest matching data string that is equal to or less than a predetermined upper limit within the range of the predetermined quantity. If the matching data string is found, a code word is formed from a flag indicating a match, a pointer indicating the position of the found data string in the dictionary, and a length indicator indicating the length of the matching data string. If the column is not found, the data compression apparatus performs data compression by forming a codeword from a flag indicating a mismatch and the data to be encoded. Counting means for counting the number of remaining data, and, based on the number of remaining data counted by the counting means, when the number of remaining data is equal to or less than the upper limit, the smallest bit capable of expressing the number of data in a binary number. Data comprising: setting means for setting the number of bits of the length indicator to a number; and encoding means for performing encoding with the number of bits of the length indicator set by the setting means. Compression device. 3. The data compression apparatus according to claim 2, wherein a program or data written in the nonvolatile semiconductor memory is compressed. 4. A compressed data compressed by the data compression method according to claim 1 or the data compression apparatus according to claim 2 is input, and when a flag included in a code word indicates a mismatch, the data is included in the code word. The other data is extracted and decoded, and the data is input to a predetermined number of dictionaries to be referenced for decoding.If the flag included in the codeword indicates a match, the data is included in the codeword. A data decompression method for decompressing compressed data by extracting and decoding corresponding data from the dictionary using the pointer and the length indicator to be input, and inputting the data into the dictionary. The number of remaining data of the decompressed data is counted, and when the number of remaining data is equal to or less than a predetermined upper limit within the range of the above-mentioned predetermined amount, the minimum number of data that can be represented by a binary number is calculated. Data decompression method characterized by determine the number of bits of the length indicator from the number of bits, performs decoding using the number of bits of the discriminated length indicator. 5. A compressed data compressed by the data compression method according to claim 1 or the data compression apparatus according to claim 2, and when a flag included in a code word indicates mismatch, the data is included in the code word. The other data is extracted and decoded, and the data is input to a predetermined number of dictionaries to be referenced for decoding.If the flag included in the codeword indicates a match, the data is included in the codeword. A data decompression device for extracting corresponding data from the dictionary using the pointer and the length indicator to be decoded and decoding the data, and inputting the data to the dictionary, thereby expanding compressed data, Counting means for counting the number of remaining data of the decompressed data; and, based on the number of remaining data counted by the counting means, the number of remaining data is a predetermined upper limit within the range of the predetermined quantity. Discriminating means for discriminating the number of bits of the length indicator from the minimum number of bits capable of expressing the number of data in binary when the number of data becomes lower, and the number of bits of the length indicator discriminated by the discriminating means A data decompression device, comprising: decoding means for performing decoding by using. 6. The data decompressor according to claim 5, wherein the decompressed data read from the non-volatile semiconductor storage means in which the compressed data of the program or data is stored.