JP2952067B2 - Data compression method - 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 system based on LZW coding as an improvement of an incremental decomposition type, which is a kind of universal coding. In recent years, various types of data such as character codes, vector information, and images have been handled by computers, and the amount of data handled has rapidly increased. When dealing with a large amount of data, by compressing the data amount by omitting redundant portions in the data, the storage capacity can be reduced or the data can be transmitted faster.

[0002] Universal encoding has been proposed as a method capable of compressing such various data in one system. Here, the field of the present invention is not limited to character code compression, and can be applied to various data. In the following, one word of data will be used, following the name used in information theory. The unit is called a character, and the data obtained by combining a plurality of words is called a character string.

[0003] As a typical method of the universal code,
There is a Ziv-Lempel code (for more information,
For example, see Munakata "Ziv-Lempel Data Compression Method", Information Processing, Vol. 26, No. 1, 1985). Two algorithms of a universal type and an incremental parsing type have been proposed for dihurempel codes.

Further, as an improvement of the universal type algorithm, there is an LZSS code (TCBell, “Better OPM”).
/ L Text Compression ”, IEEE Trans. On Commun., Vol.
COM-34, No. 12, DEC.1986). Further, as an improvement of the incremental decomposition type algorithm, LZW (Lempel-Ziv-Welch)
(TAWelch, “A Technique for High-Perfo
rmance Data Compression ”, Computer, June 1984).

[0005] Among these codes, the LZW code has come to be used for file compression of a storage device or the like because of its high speed processing and the simplicity of the algorithm.

[0006]

2. Description of the Related Art FIG. 9 shows a flow of a conventional LZW code, and FIG. 10 shows a flow of a decoding process. First, the LZW encoding process has a rewritable dictionary, divides an input character string into different character strings (substrings), assigns reference numbers to the character strings in the order in which they appear, and registers them in the dictionary. The character string currently input is represented and encoded by the reference number of the longest matching character string registered in the dictionary.

FIG. 11 is an explanatory diagram of LZW encoding, FIG. 13 is an explanatory diagram of LZW decoding, and FIG. 12 shows an example of a dictionary configuration created at the time of encoding and decoding.
In FIGS. 11, 12, and 13, for the sake of simplicity, an example is described in which data consisting of a combination of three characters of abc is compressed and decompressed. In the LZW encoding process shown in FIG. 9, first, in step S1, a character string consisting of one character for all characters is registered in the dictionary as an initial value, and then encoding is started.

In the encoding in step S1, a dictionary is searched by the first character K inputted to obtain a reference number ω, which is used as the initial character string. Next, in step S2, the next character K of the input data is read. In step S3, it is checked whether or not the character input has been completed.
A search is performed to determine whether an extended character string (ωK) in which the character K read in step S2 is added to the initial character string ω obtained in step 1 is in the dictionary.

If the character string (.omega.K) is not found in the dictionary in step S4, the flow advances to step S6 to output the reference number .omega. Of the character K obtained in step S1 as a code word code (.omega.). Is added to the dictionary and registered in the dictionary, the input character K in step S2 is replaced with the reference number ω, the dictionary address n is incremented, and the process returns to step S2 to read the next character K.

On the other hand, if the character string (ωK) is found in the dictionary in step S4, the character string (ωK) is referred to in step S5 by the reference number ω
And returns to step S2 again to continue searching for the maximum matching length until the character string (ωK) cannot be searched from the dictionary in step S4.

The LZW encoding will be described below in detail with reference to FIGS. First, the input data input of FIG. 11 is read from left to right. When the first character a is input, there is no matching character string other than the character a in the dictionary, so that OUTPUT CODE 1 (reference number ω) is codeworded and output. Then, the character a is set to the initial character string ω.

Next, assuming that the second character b is input,
This input character is added to the initial character string ω.
Since ab is not in the dictionary, OUTPUT CODE 2 of character b is output as a codeword. Then, the extended character string ωK = ab
With reference number 4 and registered in the dictionary. The actual dictionary registration is registered as a character string 1b as shown on the right side of FIG. Then, the character b becomes the initial character string ω.

Subsequently, if a third character a is input, an extended character string ωK = b obtained by adding the initial character string ω to the character b
Since a = 2a is not in the dictionary, the OUTPUT CODE of character a
After outputting 1 as a code word, the extended character string ωK = ba is represented by 2a, and is registered in the dictionary with the reference number 5 attached. Then, the character a becomes a new initial character string ω. Regarding the fourth input character b, the extended character string ωK = ab is already registered in the dictionary as the code word 4 of 1b, so the character string ωK is set as a new initial character string ω, and the fifth character c is input. To create an extended character string ωK = 4c = abc. This extended string ω
Since K = abc is not registered in the dictionary, OUTPUT CODE4 of the character string ab = 1b is output as a code word, and the extended character string ωK = abc is registered in the dictionary as code word 6 in the form of 4c. Hereinafter, similarly, this processing is continued.

The decoding process of FIG. 10 performs the reverse operation of the encoding of FIG. In the LZW decoding of FIG. 10, as in the case of encoding, decoding is started after a character string consisting of one character for every character is previously registered in a dictionary as an initial value. First, in step S1, the first code (reference number) is read, and the current CODE is set.
OLDcode, and the first code corresponds to one of the one-character reference numbers already registered in the dictionary.
Search for character code (K) that matches DE and output character K.

The output character K is used for later exception processing.
Set to FINchar. Next, the process proceeds to step S2, where the next code is read and set as INcode in CODE. In step S3, it is checked whether there is a new code, that is, whether or not the code input has been completed, and the process proceeds to step S4. In step S3, it is checked whether the input code CODE is defined (registered) in the dictionary. . Normally, since the input code word is registered in the dictionary in the processing up to the previous time, step S
The process proceeds to step S5, where a character string code (ωK) corresponding to the code CODE is read from the dictionary, the character K is temporarily stacked in step S6, and the reference number CODE (ω) is set as a new code CODE, and the process returns to step S5 again. The steps S5 and S6 are recursively repeated until the reference number ω reaches one character K. Finally, the process proceeds to step S7, where the characters stacked in step S6 are popped up and output in LIFO (Last In Fast Out) format. I do. At the same time, in step S7,
A new reference number is added to a character string representing a combination (ωK) of the code ω used last time and the first character K of the character string restored this time and registered in the dictionary.

The LZW decoding process will be specifically described with reference to FIG. First, in FIG. 13, the first input code word (INPUT CODE) is 1, and the characters a, b, and c are already registered in the dictionary as reference numbers 1, 2, and 3 as shown in FIG. Is replaced with the character string a of the reference number that matches the code word 1 and output.

Similarly, for the next code word 2, the character b
And output. At this time, code word 1 processed last time
A new reference number 4 is added to the character string ωK = 1b, which is a combination of the first character b of the character string decoded this time and the first character b, and registered in the dictionary. For the third code word 4, the character string ab is output by replacing the character string 1b obtained by the dictionary search with the character string ab. At the same time, a character string ωK = 2a, which is a combination of the code word 2 processed last time and the first character a of the character string decoded this time.
A new reference number 5 is added to (= ba) and registered in the dictionary.

This process is repeated in the same manner.

In the LZW decoding of FIG. 13, there is the following exception processing. This exception handling occurs in the decoding of the sixth input codeword 8. Codeword 8 is not defined in the dictionary at the time of decoding and cannot be decoded. In this case, an exception process is performed in which a character string 5b is obtained by adding the first character b of the previously decoded character string ba to the previously processed codeword 5 and further replaced with 5b = 2ab = bab. Then, after the output of the character string, the reference number 8 is added to the character string 5b obtained by adding the first character b of the character string decoded this time to the previous code word 5 and registered in the dictionary.

This exception processing is performed through the processing of steps S4 and S8 of the decoding processing flow of FIG. 10. Finally, in step S7, the character string is output to the dictionary in which the reference number is added to the new character string. Is registered in step S7. In the LZW decoding of FIGS. 10 and 13, a case has been described where a dictionary is created in real time while decoding the code on the decoding side, but the dictionary created at the time of encoding is used as a copy on the decoding side as it is. May be encoded. In this case, the exception processing on the decoding side becomes unnecessary.

However, if LZW encoding is performed according to the procedure shown in the flowchart of FIG. 9, every time a dictionary is searched for one character string, the entire dictionary must be searched at the worst. There was a problem. Therefore, in the conventional dictionary search method, the external hash method (open has
hing or chaining) to increase the processing speed.

First, in a general dictionary search by the hash method, when a set S composed of a plurality of character strings is considered,
The storage position of the character string x of the set S can be directly calculated from the character string x itself, and the address indicating the storage position can be directly calculated, so that a high-speed dictionary search can be performed. String storage location,
That is, assuming that addresses from 0 to m-1 are assigned to the hash table, in the hash method, one function h: S → [0, 1,..., M-1] is determined, and the set S The address of the character string x of h (x)
Asking. This function h is a hash function, and the value h (x)
Is referred to as a hash address of the character string x.

The hash method is usually used when the size of the set S is much larger than the number m of addresses.
However, no matter how the hash function h is selected, h (x1) = h (x2) hash addresses may coincide with different character strings x1 and x2 of the set S.
This is called collision, and an external hashing method (open hashing, or chaining) is used as one of the measures against collision.

In the external hash method, as shown in FIG. 14, a linked list is prepared for each hash address i indicated by an index (directory), and the character string x of the hash address h (x) = i that has caused a collision is Store in order from the top of the linked list. Each linked list with the same hash address h (x) is called a bucket.

FIG. 15 shows a processing flow of LZW encoding using a list structure of the external hash method for dictionary search. FIG. 16 shows a configuration example of a conventional dictionary, and FIG. 17 shows an arrangement on a dictionary memory corresponding to the dictionary configuration. First, Figure 1
7, the dictionary memory is the first memory (first
st) 100, next memory (next) 200, and extension memory (extension; abbreviated as "ext") 30
0. Here, the first memory 100 is shown in FIG.
14 corresponds to the index (directory) of the external hash method, the next memory 200 corresponds to “next” in the linked list in FIG. 14, and the extended memory 300 corresponds to “name” in FIG.

In the dictionary configuration shown in FIG. 16, as shown at the lower right, one node indicates the following information. (1) Inside the node; Registered symbol of the extended memory (2) Upper left of the node; Address (3) Lower left of the node; Address of the next first memory (4) Lower right of the node; Address of the next memory It is shown that.

The LZW encoding process shown in FIG. 15 will be described below by taking as an example a case where three characters A, B, and C are targeted for the sake of simplicity. First, step S
In step 1, the following initialization processing is performed. (1) Initialize the dictionary to include the first character.
Here, since three characters of alphabets A, B, and C are targeted, the character codes of A, B, and C are directly used as hash addresses as addresses 1 and 2 in the dictionary memory of FIG.
Register in 2 and 3.

(2) The current character registration number n in the dictionary is set to the number of characters registered in (2). In the case of three alphabets, n = 3. (3) The first character K that is input is defined as a first character string i. In this case, since the first input character is “A”, the initial character string i = 1. (4) The dictionary search array is initialized to 0. That is, the first, next, and extended memory search arrays are first
Since they are represented by [1, Nmax], next [1, Nmax] and EXT [1, Nmax], they are initialized to zero.

After the initialization process in step S1 is completed, the process proceeds to step S2, and as a result, it is assumed that the dictionary shown in FIGS. 16 and 17 has been created. In this state, a description will be given of a case where the character string “AAAAA” is input and encoded.

Since the initialization in step S1 has been completed,
The first input character “A” is set to the initial character string ω = 1, the first input character “A” is set to the initial character string ω = 1 in step S1, and the second input character “A” is read in step S2. . Subsequently, it is determined in step S3 that there is an unprocessed character, and the process proceeds to the dictionary search step shown in steps S5 to S9.

In the dictionary search step, first, in step S5
Sets the initial character string ω = 1 to the counter i as i = 1, and sets the j counter to j = 0. Here, the counter i is the address value of the dictionary memory specified by the value stored in the first memory, and the counter j is the address value of the dictionary memory specified by the value stored in the next memory.

Next, in step S6, the contents of address 1 in the dictionary memory of FIG. 17 designated by the i counter are read, and "A" is read from the extension memory 300 as a symbol (smbol).
And reads the next first address “4” from the first memory 100 and sets the i counter to i = 4. Subsequently, the process proceeds to step S7, where it is checked whether or not i = 0 to determine whether or not to proceed to the dictionary registration step. Since i = 4 at this time, the process proceeds to step S8, and the extension of the address 1 in step S6 is performed. The symbol “A” obtained by referring to the memory 300 and the first input character “A”
Is determined. In this case, since they match, the process returns to step S2 to read the third input character "A".

Subsequently, step S5 is performed via step S3.
Then, the value i = 4 of the counter i at that time is set to the address ω of the dictionary memory, and the address 4 of the dictionary memory is referred to. Next, in step S6, the contents of address 4 of the dictionary memory are read, and the symbol (s
mbol), and reads the next first address “6” from the first memory 100 and sets the i counter to i = 6.

Subsequently, the flow advances to step S7 to check whether i = 0 or not. At this time, since i = 6, step S7 is executed.
8, the extended memory 30 at the address 4 in step S6
It is determined whether the symbol “B” obtained from 0 matches the input character “A” obtained in step S2. In this case, the process proceeds to step S9 because they do not match. Step S9
First, the value of j = 10 obtained from the next memory 200 by referring to the address 4 of the dictionary memory is set in the i counter to make i = 10. The replacement of the i counter and the j counter is performed so that the determination in step S7 is performed only for the i counter, so that the determination can also be performed for the j counter.

Then, referring to address 10 of the dictionary memory designated by the replaced i counter, address 1
The symbol “A” stored in the extended memory 300 of “0” is read, and the address value 11 of the next first memory stored in the first memory 100 of the address 10 is set to i.
Set to counter. Next, returning to step S7, since i = 11 at this time, the symbol "A" at address 10 obtained in step S9 is compared with the input character "A", and if they match, the flow proceeds to step S2. Go to the processing of the third character.

With respect to the third and fourth input characters "A", the same processing as that for the first input character is performed.
When the processing of the address 12 is completed, there are no more characters to be processed in step S3, so that the processing proceeds to step S16, where the final address ω = 12 is output as a codeword code (ω), and a series of processing is completed.

Next, the processing of the dictionary registration step of steps S11 to S15 will be described. Dictionary registration is performed when i = 0 in the search of the first memory 100 or the next memory 200 in the dictionary search step. That is, if i = 0 is determined in step S7, dictionary search can no longer be performed, so the dictionary address ω at that time is output as a code word code (ω) in step S10, and the dictionary registration step is entered.

In the dictionary registration step, first, in step S1
In step 1, the number n of characters currently registered in the dictionary memory at that time is set in the i counter, and n is incremented by one.
Subsequently, in step S12, it is checked whether or not j = 0, and j =
If it is not 0, i = 0, and the process proceeds to step S13 to perform the registration processing of the first memory 100. If j = 0, the process proceeds to step S14 to perform registration processing of the next memory.

The first memory 100 in step S13
(1) In the first memory 100 of the memory address n specified by the i counter, the next registration destination is indicated in the first memory 100 (n +
(2) The extended memory 10 of the next memory address (n + 1) is stored.
Register the input character K as a symbol at 0. Specifically, in the case where the input character “A” is registered following the address 11 in FIGS. 16 and 17, the next registration destination is stored in the first memory 100 of the memory address 11 specified by the i counter. Is stored, and the input character “A” is registered as a symbol in the extended memory 100 at the next memory address 12.

On the other hand, the next memory 2 in step S14
In the registration process of 00, (1) the value of (n + 1) indicating the next registration destination is stored in the next memory 200 of the memory address specified by the i counter, and (2) the next memory address (n + 1) Extended memory 10
Register the input character K as a symbol at 0.

Specifically, in the case where the input character "A" is registered after the address 11 in FIGS.
The address value 10 indicating the next registration destination is stored in the next memory 200 at the memory address 11 specified by the counter, and the input character “A” is registered as a symbol in the extended memory 100 at the next memory address 10. When the above-described registration processing is completed, the registered character K is set in the i counter, and the process returns to the dictionary search step from step S2.

[0042]

SUMMARY OF THE INVENTION Such a conventional LZ
Since the W code requires a lot of time for dictionary search processing when encoded by software, the dictionary search is speeded up using an external hash method. However, in the book search by the external hash method, since the collation between the input character and the candidate character is performed sequentially, the dictionary search time is about 80% of the entire time.
% And it is difficult to increase the speed.

On the other hand, the inventor of the present invention has made it possible to speed up the search by dividing the linked list into a plurality of parts using the information of input characters which have already been encoded when performing a dictionary search. An encoding scheme is proposed. However, in actual encoding, a usable memory capacity is predetermined, and depending on the size of input data, encoding may be completed without using the entire dictionary memory. Further, depending on the application, there may be a case where it is desired to perform encoding by giving priority to the processing time over the compression rate.

However, the conventional coding method has a problem that it is difficult to perform coding by properly combining a request for high speed and a request for high compression. The present invention has been made in view of such a conventional problem, and an object of the present invention is to provide a data compression method capable of performing encoding appropriately corresponding to a demand for high speed and high compression.

[0045]

FIG. 1 is a diagram illustrating the principle of the present invention. First, the present invention divides an encoded character string into different substrings, adds a different reference number to each substring, registers it in the dictionary 1, and stores the input character string in the substring in the dictionary 1. Encoding means 2 for compressing data by encoding specified by the reference number of the one that matches the maximum length, and input character as the reference number i of the subsequence registered in the dictionary 1 using the external hash method for searching the subsequence Dictionary search means 3 for generating a hash address to which information Km extracted from the element of K is added to generate a linked list of the number of divisions according to the number of bits of the additional information Km and searching the dictionary 1 Target data compression method.

According to the present invention for such a data compression method, the number of bits of the information Km extracted from the element of the input character K added to the reference number i of the subsequence is appropriately changed, and the number of divisions of the linked list is changed. Is provided. Here, the number-of-division designating means 4 determines the number of bits of the information Km extracted from the element of the input character K added to the reference number i of the subsequence by the number of bits suitable for high-speed processing (see FIG. The number of bits (see FIG. 3A) suitable for the compression process is specified.

More specifically, the number of bits is specified based on the number of divisions determined by the information added to the head of the input character string data by the search division number determination means 5. Further, the division number designation means 4 outputs information Km extracted from the element of the input character K.
May be specified as the number of bits corresponding to the pre-specified slag.

[0048]

According to the data compression system of the present invention having such a configuration, the following operations can be obtained. As processing conditions for data compression, there are a case where a high compression ratio is required even though it takes time, and a case where it is desired to perform processing at a high speed by giving priority to the processing time over the compression ratio.

Although such high compression ratio and high speed are conflicting processing conditions, in the present invention, the reference number of the subsequence,
That is, the hash value is obtained by adding the information Km of the element of the input character K to the address i, that is, the number of bits Km of the input character code, so that the hash value is divided according to the number of bits Km of the additional information. By arbitrarily specifying the effective bit number Km of the character code used as the additional information for determining the number, dynamic data compression processing suitable for each of the high compression rate and high-speed processing can be appropriately performed.

[0050]

FIG. 2 is a block diagram showing an embodiment of a data compression system having a dictionary search function according to the present invention. FIG.
In FIG. 1, original data (character data or codeword data) 10 to be processed is a DMA (Direct Memory Access).
It is input via the control circuit 12. M as control means
The PU 14 inputs the input original data 10 to one character K and a hash address obtained by adding the element bit Km of the character code of the one character K to the reference number i of the character string so far in the dictionary search circuit 16. After setting in the character reading circuit 18, the dictionary search circuit 16 is activated. The element bits of the character code of one character K determine the number of search divisions of the linked list in the external hash method. If the character code is 8 bits, the effective bit number of the element bit Km is selected from, for example, the upper bit. The following nine types of linked list search division numbers are obtained. In the number of divisions of such a linked list, when the number of divisions 256 matches all 256 character types to be processed, a complete hash is obtained, and only one dictionary search is required.

In the present invention, the encoding is performed by designating the number of divisions that meets the designated processing condition from among the nine division numbers. The specific number of divisions can be specified by setting the number of element bits Km of the next one character K to be added to the reference number (dictionary address) i of the character string immediately before being encoded. The function as the division number designating means 4 shown in the principle explanatory diagram of FIG. 1 for this purpose is realized by program control of the MPU 14. FIG. 3 shows the number of search divisions of the linked list with respect to the size (size) of the input data. FIG. 3 (a) shows characteristics suitable for high compression processing conditions, and FIG. Shows characteristics suitable for processing conditions.

In other words, when it is desired to obtain a high compression ratio even though it takes time, the number of divisions which is inversely proportional to the input data size shown in FIG. In this case, the larger the input data size, the smaller the number of divisions of the linked list, and the longer the maximum length of the matching subsequence in the dictionary, so that a high compression rate can be obtained. However, as the length of the matching subsequence increases, the number of dictionary searches increases, and the processing time increases. In addition, when the memory capacity is constant, unused memory can be effectively used.

On the other hand, when it is desired to reduce the processing time even if the compression ratio cannot be obtained, the number of divisions is specified in proportion to the input data size shown in FIG. In this case, the number of divisions increases according to the input data size, and the maximum number of divisions is 256.
Since it is a complete hash, it can be encoded by one dictionary search. Further, when the memory capacity is constant, it means that regardless of the input data size, all processing can be performed in a fixed time at the expense of the compression ratio.

The designation of the number of divisions based on the input data size meeting any of the conditions shown in FIGS. 3A and 3B is performed by the operator knowing the data size to be processed and the MPU.
The number of divisions may be directly specified for 14. A value indicating the size of the data at the head of the input data is set in advance, and the size of the data is read by the MPU 14 to
According to the division characteristics shown in FIG. 3A or FIG. 3B, the number of divisions can be automatically converted to the optimal division number for the size of the input data. The function of the MPU 14 for determining the size of the input data is a function as the search division number determining means 5 shown in the principle explanatory diagram of FIG.

As shown in FIG. 4 (a), the input data format for this has a value indicating the size of the input data at the head, followed by the original data sequence. Further, as shown in FIG. 4B, the encoded data has, at the beginning, the size of the dictionary used for encoding, that is, the used dictionary size (the number of divisions and the used size of each divided dictionary). The converted data follows.

Therefore, at the time of restoration, it is possible to determine the largest dictionary to be used for restoration based on the size of use of the leading dictionary, and perform restoration. Referring again to FIG. 2, the dictionary search path 16 reads the candidate character of the character string extended by one character from the dictionary memory 20, and the match check circuit 22 checks (matches) the match between the input character and the candidate character. The connection detection circuit 24 detects the presence or absence of a candidate character.

The pipeline control circuit 26 reads out the next candidate character from the dictionary memory 20 in parallel with the matching between the input character and the candidate character by the match checking circuit 22 and the detection of the presence or absence of the candidate character by the connection detection circuit 24. Multiply. By performing the pipeline processing in the pipeline control circuit 26 in this manner, the search and collation processing for each of a plurality of candidate characters can be executed in the cycle time of the dictionary memory 20.

Further, the dictionary search circuit 16 is provided with a continuous address circuit 28. The continuous address circuit 28 generates a continuous address.
The hash address and the candidate character registered at the consecutive addresses are read out. In encoding the LZW code, a character string that matches the maximum length in the dictionary memory 20 is obtained. Therefore, the input character is added and the character string is sequentially extended one character at a time. When the candidate character disappears, it is understood that the character string has the maximum matching length. At this time, the character string up to the maximum matching length character string is represented by a reference number using the address ω, and the reference number ω is output from the input / output port 30 as an externally compressed codeword code (ω). . FIG. 5 is a block diagram of an embodiment showing details of the dictionary search circuit 16 of FIG.

In FIG. 5, the MPU 1 firstly sets, for example, the most significant bit Km of the 8-bit character code of the first character reference number i and the second character K of the character string (Km, i).
Is set in the address register 18-1 and the input second character K is set in the register 22-1. Next, the dictionary search circuit 16 is added to the pipeline control circuit 26.
Command to start.

First, the pipeline control circuit 26
After resetting 8-1 to Km = 0, the dictionary memory 20 is read. The FF 24-2 is the most significant bit (MSB) of the address of the dictionary memory 20, and the address register 1
The area corresponding to the array First of the dictionary memory 20 is read with the contents of 8-1 as the lower address. Here, FIG. 6 shows an example of the configuration of the dictionary memory 20, and FIG. 7 shows an arrangement of the dictionary memory 20 corresponding to FIG. 6 and 7
Exemplifies the encoding of three characters a, b, and c for the sake of simplicity.

In this memory array, first0,
1 is determined by the most significant bit Km of the next one character added to the original hash address i, and Km = 0 designates first1 and Km = 1 designates first1.

Therefore, in FIG. 5, the address (reference number) i of the first character and the most significant bit K of the second character
When the dictionary memory 20 is accessed using m as an address,
At this time, since Km = 0 due to the reset of the FF 28-1, the array first0 in the dictionary memory of FIG.
An area corresponding to (first memory) and extension (extended memory) is read. Dictionary memory 20
A portion (first0) corresponding to the linked list address in the contents of one word read from the above is set in the address register 18-1, and a portion (e) corresponding to the candidate character K 'is set.
xtention) is set in register 18-2.

At the same time, the address register 18-1
Of the character K excluding the most significant bit Km of the character K is transferred to the register 18-3. Also, F
The content Km of F28-1 is moved to FF24-2. In parallel with the reading of the dictionary memory 20, the input characters in the register 22-1 and the candidate characters in the register 18-2 are compared and matched by the coincidence comparison circuit 22-2.

The input character K and the candidate character K 'are obtained by comparison and collation.
Match, the pipeline control circuit 26 sets the next input character in the register 22-1, and at this time the FF 28
Since −1 remains after reset, Km = 0
The array first0 and extend in the dictionary memory 20 at the address specified by the address register 18-1
The area (first 0) corresponding to the linked list address is read from the address register 18 in the content of one word read from the dictionary memory 20.
-1 and the part (ext
ention) is set in the register 18-2, comparison and collation are performed, and so on.

At the time of such comparison and collation, it is determined whether or not the contents of the address register 18-1 read from the dictionary memory 20 and stored at the same time as the comparison and collation are all 0s. If all 0s are found, it is detected that there are no more candidate characters. The output of the NOR circuit 24-1 when there are no more candidate characters is MPU14.
And MPU 14
Outputs the value of the memory address at the end of the comparison and collation as a codeword, and proceeds to search for the next input character. Similarly, in the search for the next character, a set of the contents i of the register 18-3 (the immediately preceding reference number already encoded) and the most significant bit Km of the input character K is set in the address register 18-1. , The input character is set in the register 22-1, and the search processing for the second and subsequent characters is performed.

On the other hand, if the result of the comparison and collation does not match, the area of the array next having the same address is read and set in the address register 18-1, the next dictionary memory 20 is read, and the matching candidate is read. The reading of the array next is repeated until a character is obtained. Such a dictionary search of the present invention will be specifically described with reference to FIGS. 6 and 7 as follows.

Now, it is assumed that the character string "aaaa" is encoded in a state where the dictionaries are constructed as shown in FIGS. Here, it is assumed that the most significant bit of the character “a, c” is Km = 0, and the most significant bit of the character “b” is Km = 1. First, the reference number of the first character “a” is i = 1, and since the most significant bit MSB of the character code of the second character “a” is Km = 0, the address of the dictionary memory 1 in FIG. Km in 1
= 0 and the contents 10 of first0 specified by exte
The candidate character “a” of the ntion is read. in this case,
Since the input character "a" matches the candidate character "a", the next dictionary memory is accessed by the address 10 obtained from first0 to read out the candidate character "a",
Further, the content 11 of the array first0 specified by the most significant bit Km = 0 of the third character "a" is read. For the third character "a", a match with the candidate character "a" is obtained. Similarly, the fourth and fifth characters are processed, and the contents of the last fifth character "a" array first0 are Since it is 0, it is determined that there are no more candidate characters, and the final address 12 is output as the code word of the input character string "aaaa".

On the other hand, as for the character string "abc", since the most significant bit of the second character is Km = 1, the contents 4 of the array "first1" of the address 1 are read out, and on condition that they match the candidate character. Read the candidate character at address 5,
Finally, a match with the candidate character at the address 6 is obtained, and the code word 6 of the input character "abc" is output. Note that the search for next when no match with the candidate character is obtained is the same as in the related art.

FIG. 8 shows a flowchart of the encoding algorithm according to the present invention, which is basically the same as the conventional method shown in FIG. The difference is that (1) a memory address 1 is created by a combination of a reference number i of a character string coded immediately before as a memory address in step S5 and an element bit of the next character K, for example, the most significant bit Km. Point,
(2) In step S6, the dictionary memory first [l] determined by the set of l = (Km, i) is read to obtain the divided first in the next memory address i. (3)
Further, at the time of registering the candidate character to the address i in step S13, which of first0 and 1 of the previous address i-1 is registered with the address i is distinguished according to the most significant bit Km of the candidate character. ,.

In the above embodiment, the method of storing the candidate characters as they are in the dictionary memory and comparing them is described. However, in order to reduce the memory capacity, the candidate characters consist of only the bits excluding the bit Km added to the hash address. May be provided. In another embodiment of the present invention, instead of adding a specific bit Km of an input character to a hash address,
It is obvious that the same can be realized by adding bits of information created by processing input characters.

[0071]

As described above, according to the present invention, the characteristics of the number of divisions in the dictionary memory, for example, with respect to the input data size are selected according to the processing conditions of high speed or high compression, and the size of the input data is determined. By automatically or artificially specifying according to the encoding, by dynamically determining the encoding for each encoding and performing encoding, high-speed and high-compression elements are successfully fused Can realize data compression.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of a compression method according to the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention.

FIG. 3 is a characteristic diagram showing the number of divisions with respect to the input data size according to the present invention divided into processing conditions.

FIG. 4 is an explanatory diagram showing an input data format and an encoded data format according to the present invention;

FIG. 5 is a configuration diagram of an embodiment showing details of the dictionary search circuit of FIG. 2;

FIG. 6 is an explanatory diagram showing a configuration of a dictionary memory used for encoding in FIG. 2;

FIG. 7 is an explanatory diagram of a layout of a dictionary memory corresponding to FIG. 6;

FIG. 8 is a flowchart showing an encoding algorithm of the present invention.

FIG. 9 is a flowchart of a conventional LZW encoding algorithm.

FIG. 10 is a flowchart of a conventional LZW decoding algorithm.

FIG. 11 is a diagram illustrating a specific example of conventional LZW encoding.

FIG. 12 is an explanatory diagram of a dictionary configuration example.

FIG. 13 is a diagram illustrating a specific example of conventional LZW decoding.

FIG. 14 is an explanatory diagram of a list structure of the external hash method

FIG. 15 is a flowchart showing a conventional LZW code encoding algorithm using an external hash method;

16 is an explanatory diagram showing a configuration of a dictionary memory used for encoding in FIG.

FIG. 17 is an explanatory diagram of a layout of a dictionary memory corresponding to FIG. 16;

[Explanation of symbols]

 1: Dictionary 2: Dictionary search means 3: Data addition means 4: Judgment means 10: Original data 12: DMA control circuit 14: MPU 16: Dictionary search means (dictionary search circuit) 18: Multiple character reading circuit 18-1: Address Register 18-2, 18-3: Register 20: Dictionary memory 22: Match check circuit 22-1: Resistor 22-2: Comparator 24: Connection detection circuit 24-1: NOR circuit 24-2: FF 26: Pipeline Control circuit 28-1: FF

Continuation of the front page (72) Inventor Yasuhiko Nakano 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (56) References JP-A-6-83575 (JP, A) JP-A-4-219818 (JP, A) JP-A-4-156110 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 5/00 G06F 17/30 G06T 9/00 H03M 7/40 H04N 1/41

Claims (5)

(57) [Claims] An encoded character string is divided into different sub-sequences, a different reference number is added to each sub-sequence and registered in a dictionary (1), and an input character string is stored in the dictionary (1). Encoding means (2) for compressing data by designating and encoding the subsequences having the maximum length among the subsequences by the reference number, and using the external hash method to search for the subsequences, By generating a hash address in which information (Km) extracted from the element of the input character (K) is added to the reference number (i) of the substring registered in 1), the additional information (K
m) a dictionary search means (3) for generating and searching a linked list of the number of divisions according to the number of bits of the input character string to be added to the reference number (i) of the subsequence. A data compression method comprising a division number designating means (4) for designating the number of divisions of a linked list by appropriately changing the number of bits of information (Km) extracted from elements of K). 2. The data compression method according to claim 1, wherein said division number designating means (4) extracts information extracted from an element of an input character (K) added to a reference number (i) of said partial sequence. A data compression method, wherein the number of bits of (Km) is designated as a number of bits suitable for high-speed processing or a number of bits suitable for high-compression processing. 3. A data compression method according to claim 1, wherein said division number designating means (4) extracts information extracted from an element of an input character (K) added to a reference number (i) of said partial sequence. A data compression method wherein the number of bits of (Km) is specified based on information indicating the size of character string data to be encoded. 4. A data compression method according to claim 1, further comprising a search division number determining means (5) for determining a division number from information added to a head of a character string. method. 5. A data compression method according to claim 1, wherein said division number designation means (4) extracts information extracted from an element of an input character (K) added to a reference number (i) of said partial sequence. A data compression method, wherein the number of bits of (Km) is a number of bits corresponding to a predetermined number of divisions.