JP3015483B2 - Data compression and decompression 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 method and a decompression method using an LZW code which is known as an improvement of an incremental decomposition type which is a kind of universal code. 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.

[0002] When a large amount of data is handled, by compressing the amount of data by omitting redundant parts in the data, it becomes possible to reduce the storage capacity or to transmit the data at high speed. As described above, universal coding has been proposed as a method capable of compressing 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 connecting arbitrary words is called a character string.

[0003] As a typical method of the universal code,
There is a Ziv-Lempel code (for details,
For example, Munakata "Ziv-Lempel's data compression method", information processing,
Vol. 26, No. 1, 1985). For the Ziv-Lempel code, two algorithms, (1) universal type and (2) Incremental parsing, have been proposed.

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. 8 shows a flowchart of a conventional LZW code encoding algorithm, and FIG. 9 shows a flowchart of a decoding algorithm. First, LZW encoding has a rewritable dictionary, divides an input character string into different character strings, assigns numbers to the character strings in the order in which they appear, registers them in the dictionary, and registers the currently input character string. The sequence is represented and encoded only by the reference number of the longest matching character string registered in the dictionary.

FIG. 10 shows a specific example of LZW encoding, FIG. 11 shows a specific example of LZW decoding, and FIG.
Shows the contents of the dictionary used for encoding and decoding. still,
FIGS. 10, 11 and 12 show a case where a character string composed of a combination of three characters of abc is compressed and decompressed for simplicity of explanation. First, the LZW encoding process of FIG. 8 will be described as follows.

Step S1: A character string consisting of one character for every character is registered as an initial value before encoding starts. In the encoding in step S1, the dictionary is searched using the first character K input to obtain a reference number ω, which is used as a prefix string. Step S2: Read the next character K of the input data.

Step S3: It is checked whether the character input has been completed. Step S4: A search is performed to determine whether or not (ωK) in which the character K read in step S2 is added to the initial character string ω obtained in step S1 is in the dictionary. Step S5: If the character string (ωK) is found in the dictionary in step S4, the character string (ωK) is replaced with the reference number ω in step S5, and the process returns to step S2 to return to the character string (ωK).
The search for the maximum matching length is continued until K) cannot be searched from the dictionary.

Step S6: If the character string (ωK) is not found in the dictionary in step S4, the process proceeds to step S6, and the reference number ω of the character K obtained in step S1 is replaced with the code word code
(Ω), add a new reference number to the character string (ωK), register it in the dictionary, replace the input character K in step S2 with the reference number ω, increment the dictionary address n, and step Returning to S2, the next character K is read.

The details will be described with reference to FIGS. 10 and 12. First, the input data input of FIG.
Reads from left to right. The first character "a" is input to be the initial character string ω. Next, if the second character b is entered,
This input character is added to the initial character string ω.
Since ab is not in the dictionary, OUTPUT CODE 1 of character a is output as a codeword.

Then, the extended character string ωK = ab is assigned a 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 the third character a is input, the extended character string ωK = ba = 2a obtained by adding the initial character string ω to the character a is not in the dictionary.
After outputting OUTPIT CODE 2 as a code word, the extended character string ωK = ba is represented by 2a, and is registered in the dictionary with a reference number 5. Then, the character a becomes a new initial character string ω.

As for the fourth input character b, since the extended character string ωK = ab is already registered in the dictionary as the code word 4 of 1b, the character string ωK is set to a new initial character string ω, and 5
Enter the third character c to expand the character string ωK = 4c = abc
make. Since this extended character string ωK = abc is not registered in the dictionary, the OUTPUT CODE4 of the character string ab = 1b
Is output as a code word, and the extended character string ωK = abc is registered as a code word 6 in the form of 4c in the dictionary. Hereinafter, similarly, this processing is continued.

Next, the decoding process of FIG. 9 will be described. In this decoding, similarly to the encoding, the decoding is started after a character string consisting of one character for every character is previously registered in the dictionary as an initial value. Step S1: The first code CODE is read and the reference number ω is decoded. The current reference number ω is assumed to be OLD ω, and since the first code corresponds to one of the reference numbers of one character already registered in the dictionary, the character D corresponding to the input reference number ω
Search for (K) and output the letter K. The output character K is set in FINchar for later exception processing.

Step S2: The next code CODE is read. Step S3: It is checked whether or not there is a new code, that is, whether or not code input has been completed. Step S4: The reference number ω is decoded from the read code CODE and set as INω.

Step S5: It is checked whether or not the code CODE input in step S4 is registered in the dictionary (ω ≧ n). Step S6: Normally, the input code word has been registered in the dictionary in the previous process, so the process proceeds to step S6, where the character string D (ω′K) corresponding to the reference number ω is read from the dictionary.

Step S7: The character string K is temporarily stacked, and the reference number ω ′ is set as a new ω, and the process returns to step S6.
And recursively repeats the procedure of step S6 with reference number ω
Is repeated up to one character. Step S8: The characters stacked in step S7 are LI
Pop up and output in FO (Last In Fast Out) format. At the same time, a new reference number n is added to a character string represented as a set (OLD ω, K) of the reference number OLD 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, the first 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. And replaces it with the character string a of the reference number. Similarly, the next code 2 is replaced with the character b and output. At this time, a new reference number 4 is added to (1b) in which the previously processed code and the first character b decoded this time are combined and registered in the dictionary.

The third code 4 outputs a character string ab by replacing d with ab by searching the dictionary. At the same time, the code 2 processed last time and the first character a of the character string decoded this time
The character string 2a (= ba) combined with
And register it in the dictionary. Hereinafter, similarly, this processing is repeated. In the decoding of FIG. 11, there is the following exception processing.

This exception processing occurs when the sixth input code 8 is decoded. The code 8 is not defined in the dictionary at the time of decoding and cannot be decoded. In this case, the previously processed code 5
Is obtained by adding the first character b of the previously decoded character string ba to the character string 5b, and further replaced with 2ab and bab and output.

Then, the previous code word 5 is added to the output word of the character string.
The reference number 8 is added to the character string 5b obtained by adding the character b of the character string decoded this time to the dictionary and registered in the dictionary. This exception processing is performed through the processing of steps S5 and S9 of the decoding processing flow of FIG. 9. Finally, in step S8, the output of the character string and the registration in the dictionary in which the reference number is added to the new character string are performed. Done.

The encoding process and the decoding process in FIGS. 8 and 9 are performed while creating the same dictionary.

[0023]

In the conventional LZW encoding method, when an input character string is divided into different character strings and encoded, each character string currently being encoded is replaced with a previous character string. Has the form of encoding as appearing independently.
In this way, there is no problem in encoding the memoryless source.
However, many data such as actual sentences are regarded as storage information sources, and the LZW code does not sufficiently utilize the history of occurrence of a character string. There was a disadvantage that redundancy remained.

According to the conventional algorithm shown in FIG.
FIG. 13 shows a search tree of the dictionary when ZW encoding is performed. In the dictionary registration of FIG. 13, encoding is started in a state where all the characters 256 to be processed are initially registered with indexes (reference numbers) 0 to 255, and for example, as shown in FIG. The column is indexed (reference number) i1,
These are represented by i2 and i3 and registered in the dictionary.

In this case, the root of the search tree of the dictionary is empty, which indicates that the history of the character string that has appeared before with respect to the character string being figure-encoded is not considered in the conventional LZW encoding. ing. The present invention has been made in view of such a conventional problem, and reduces the redundancy between character strings by incorporating the dependency of an encoded character string with the immediately preceding character string into a dictionary classification. It is an object of the present invention to provide a data compression method and a decompression method based on LZW in which the compression ratio is increased.

[0026]

FIG. 1 is a diagram illustrating the principle of the present invention. First, the present invention is directed to a data compression method in which an input character string is LZW-encoded by designating a reference number of a subsequence that matches the maximum length among already encoded subsequences registered in the dictionary 10.

According to the present invention for such a data compression method, a plurality of divided dictionary groups 10-1 are used as the dictionary 10.
-10-N are assigned and the divided dictionary numbers Di are sequentially assigned,
Further, for each of the divided dictionaries 10-1 to 10-N, at least all character types are initially registered with reference numbers assigned in character units. At the time of encoding the input character string, the specific division dictionary 10-1 is classified by a classification code based on the character code C1 of the last character of the immediately preceding encoded character string and the character code C0 of the character immediately before the final character. Is specified and encoded.

Here, part of the character code C1 of the last character of the character string that has just been encoded (for example, the value of the upper 4 bits)
A specific division dictionary 10-1 is designated and designated by a classification code based on a part of the character code C0 immediately before the last character (for example, a value obtained by shifting upper four bits to lower four bits). Become On the other hand, the present invention converts an input character string into a dictionary 1
A data restoration method for restoring an original character string from a codeword specified and designated by a reference number of a subsequence that matches the maximum length among already encoded subsequences registered in 0 is also included.

According to the present invention as such a data restoration method, a plurality of divided dictionary groups 10-1 are used as the dictionary 10.
-10-N are assigned and the divided dictionary numbers Di are sequentially assigned,
Further, for each of the divided dictionaries 10-1 to 10-N, at least all character types are initially registered with reference numbers assigned in character units. At the time of restoring the input code string, a specific divided dictionary 1 is determined by a classification code based on the character code C1 of the last character of the immediately preceding restored character string and the character code C0 immediately before the last character.
It is characterized in that restoration is performed by designating 0-i.

Here, a part of the character code C1 of the last character of the previously restored character string (for example, the value of the upper 4 bits)
A specific classification dictionary 10-i is designated and restored by a classification code based on a part of the character code C0 immediately before the final character (for example, a value shifted to the lower 4 bits by upper 4 bits). .

[0031]

According to the data compression and decompression method of the present invention having such a configuration, the following effects can be obtained. That is, at the time of encoding, a state in which the relationship between the character immediately before the last character of the encoded character string immediately before and the final character is put together, that is, the dictionary is designated by the classification code and encoded, By defining the configuration of the dictionary itself from the character string that appeared immediately before, encoding that incorporates the history of the appearance of character strings can be performed, and redundancy between character strings can be reduced and the compression rate can be improved.

[0032]

FIG. 2 is a block diagram showing an embodiment of the present invention. In FIG. 2, reference numeral 12 denotes a CP as a control unit.
U, and the program memory 14
And the data memory 30 are connected. Program memory 1
Reference numeral 4 denotes control software 16, maximum match length search software 18 for performing a maximum match length search using an LZW code, encoding software 20 for converting an input character string into an LZW code, and conversion to an LZW code by the encoding software 20. Decoding software 22 for restoring the code string to the original character string is provided.

Here, in the present invention, the encoding software 2
0 and the decoding software 22 respectively shift the upper 4 bits of the character code of the character immediately before the last character of the immediately preceding character string that has already been encoded and shift it to the lower 4 bits. And a function to specify a dictionary to be used for encoding or decoding by using an 8-bit code, which is the sum of the upper 4 bits of the last character of the character string immediately before the character string, as a classification code. The dictionary is specified by the classification code using the LUT 26.

On the other hand, a data buffer 32 for storing a character string to be encoded or a code string to be decoded from now on is stored in the data memory 30;
A dictionary 10 is provided that is created and used sequentially during encoding and decoding of codes. The dictionary 10 includes a plurality of divided dictionaries 10-1 to 10-N.

The divided dictionaries 10-1 to 10-N are obtained by shifting the upper 4 bits of the character code of the character immediately before the last character of the immediately preceding encoded character string to lower 4 bits. The LUT 26 is represented by an 8-bit code consisting of the sum of the upper 4 bits of the last character of the immediately preceding character string already encoded.
Can be specified. In this embodiment, since the dictionary is designated by an 8-bit code, the number of divided dictionaries 10-1 to 10 to 10 corresponding to all 256 character types is determined.
-256 is provided. Specifically, the data memory 30
Is divided into 256 divided dictionaries for all character types.

The number of divisions of the dictionary 10 is determined by making a part of the 8-bit code valid when configuring the LUT 26.
An arbitrary number of divisions of 256 divisions or less such as 16 divisions, 32 divisions, and 64 divisions can be used. FIG. 3 is an explanatory diagram showing preprocessing of binary image data to be subjected to LZW encoding according to the present invention.

The inventors of the present application have proposed a four-line run-length conversion method as preprocessing when universally encoding a binary image. The four-line run-length conversion method refers to a four-line run-length conversion method for an image as shown in FIG.
As shown in FIG. 3B, the change of the bit pattern for the line is represented by a continuous length of the same pattern as the pattern type, that is, a run length.

The data after the conversion is shown in FIG.
As shown in (c), the pattern is converted into an 8-bit code by combining the pattern 4 bits and the run length 4 bits. By subjecting this transformed code sequence to LZW encoding according to the present invention, a high compression rate that incorporates the characteristics of an image can be obtained. The four-line conversion code can be regarded as a storage information source due to the nature of the image. When LZW encoding is performed on a four-line conversion code sequence, the four-line conversion code is regarded as a storage information source with reference to upper four bits representing a bit pattern of an image.

In the creation of a dictionary, as shown in FIG. 4, for example, coded character strings CL0 and CL1 are taken as an example.
A dictionary is created using a bit pattern obtained by combining a part of the bit pattern of the last character C1 of the character string CL0 coded immediately before and a part of the bit pattern of the character C0 immediately before the last character C1 as a classification code. Here, when the conversion code by the line run-length conversion in FIG. 3 is targeted, the upper 4 bits of the last character C1 of the character string CL0 encoded immediately before and the bits of the character C0 immediately before the last character C1 are set. A dictionary is created using an 8-bit pattern obtained by shifting the upper 4 bits of the pattern and combining them as the lower 4 bits as a classification code.

As shown in FIG. 5, the structure of the dictionary in which the history state between the character strings according to the present invention is taken, as shown in FIG.
A history state in which the root of the search tree is collected from (1) the upper 4 bits of the character one number before the last character of the character string that appeared immediately before, and (2) the upper 4 bits of the last character of the character string that appeared immediately before. , For example, a combination of two bit patterns immediately before the 4-line Run-Lengx code of FIG. As a result, a dictionary can be created as a storage information source, the redundancy between character strings is reduced, and the compression ratio can be increased.

Next, the data compression processing of the present invention will be described with reference to the flowchart of FIG. First, the following initialization processing is performed in step S1. (1) DN dictionaries D selected based on the immediately preceding character string
1 (i = 1,…, DN) all kinds of character strings consisting of one character (256
(Seed) is registered in advance as an initial value. (2) The total number of reference numbers of each dictionary is managed by n (i), and at the time of initialization, (character type +1) is set to DN D1s. That is, n is set to 256.

(3) The history from the immediately preceding character string is PK, and since there is no immediately preceding character string, PK =
Set 0. (4) The first character K is input and set to K1, and the input character K is converted to a reference number ω and output. (5) Input the second character as K, and enter this as the initial character string ω
And

(6) A value obtained by shifting the character immediately before the last character of the immediately preceding character string by 4 bits to the right (converted to lower 4 bits)
And an 8-bit classification code K1 created by ORing the upper 4 bits of the last character of the immediately preceding character string. this is

[0044]

(Equation 1)

Is expressed as follows. (7) L corresponding to the history state PK from the classification code K1
Set the UT. That is, the history P is obtained by the LUT (K1).
K is required. When the above initialization process is completed, L is input while inputting characters one by one in accordance with steps S2 to S6.
ZW encoding is performed. This procedure is basically the same as the conventional LZW encoding shown in FIG.

The difference is that in the conventional LZW encoding, there is only one dictionary, but in the LZW encoding of the present invention, as shown in steps S4, S5, and S6, a plurality of dictionaries depend on the history state PK. Dictionary D to dictionary D
The point is that pk is selected, the character string registered in the dictionary Dpk is collated to find the maximum matching length character string, and a character string obtained by extending the maximum matching length by one character is registered in Dpk.

After registration in the dictionary Dpk in step S6,
Npk for managing the number of reference numbers of the dictionary Dpk is incremented by one. In addition, in order to select the dictionary of the next character string, a value obtained by shifting the character immediately before the last character of the immediately preceding character string by 4 bits to the right and the upper 4 bits of the last character of the immediately preceding character string was ORed. A new history state PK is obtained from the classification code K1 using the LUT.

Next, the LZW decoding of the present invention is as shown in FIG. In the decoding of FIG. 7, the operation opposite to the encoding shown in FIG. 6 is performed. First, the processing of steps S1-1, S1-2, and S1-3 relating to the initialization of the dictionary is the same as that shown in (1) to (7) in the encoding step S1 of FIG.

In the decoding processing of steps S2 to S9, after decoding the reference number ω from the input code CODE, the value obtained by shifting the character immediately before the last character of the immediately preceding character string by 4 bits to the right is used. The dictionary D with the history state PK obtained from the value created by ORing the upper 4 bits of the last character of the last character string
pk is selected, and a character string corresponding to the reference number ω is obtained from Dpk.

The registration of a new character string in the dictionary in step S8 is similar to LZW encoding, but is performed one tempo later than the encoding. That is, in the encoding, when the encoding of the target character string is completed, the character string extended by one character,
That is, a set with the next character of the eye character string can be registered in the dictionary. On the other hand, in decryption, when the target character string is extended by one character, it is registered in the dictionary together with the first character of the next character string.
Registration is performed when the restoration of the next character string is completed.

That is, as shown in step S8, the pair (OLD ω, K) of the reference number OLDω of the immediately preceding character string and the first character K1 of the restored character string is changed to the last character string of the immediately preceding character string. The value obtained by ORing the upper 4 bits of the last character of the character string immediately before the previous character string with the value of the character immediately before the character shifted right 4 bits to the lower 4 bits is calculated. Is registered in the dictionary Dpk1 (n) selected in the history state PK1.

Further, the number npk of reference numbers is incremented by one and the restoration is completed for the case where the restored character string is extended and registered next (in order to extend the character string after searching and matching). After the reference number INω is set to OLDω, a value obtained by ORing the value obtained by shifting the character K3 immediately before the last character of the immediately preceding character string four bits to the right by 4 bits and the upper four bits of the last character K2 of the restored character string Thus, a new history state PK is obtained.

The above embodiment comprises the upper 4 bits of the last character of the immediately preceding encoded character string and the lower 4 bits obtained by shifting the upper 4 bits of the character immediately before the last character. Although the dictionary to be registered is classified by 8-bit code,
The upper 4 bits of the last character may be shifted to the lower 4 bits to combine the upper 4 bits of the immediately preceding character.

The number of bits of a character code used for creating a dictionary classification code and the number of character strings to be referred to (a relationship between a plurality of character strings may be adopted) are appropriately determined as necessary.

[0055]

As described above, according to the present invention, since the history of a character string that has appeared before the character string being encoded is also adopted, the redundancy between character strings is reduced, and A higher compression ratio than the LZW code can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram of the principle of the present invention.

FIG. 2 is a diagram illustrating the principle of dictionary division according to the present invention.

FIG. 3 is an explanatory diagram of a 4-line run-length encoding method for a binary image to be pre-processed according to the present invention;

FIG. 4 is an explanatory diagram showing how to determine a classification code of a dictionary in which a history state of a character string is taken in the present invention.

FIG. 5 is an explanatory diagram of a dictionary search tree obtained by capturing a history state according to the present invention.

FIG. 6 is a flowchart of an encoding algorithm of the present invention.

FIG. 7 is a flowchart of a decoding algorithm of the present invention.

FIG. 8 is a flowchart of a conventional LZW encoding algorithm.

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

FIG. 10 is a diagram illustrating a specific example of conventional LZW encoding.

FIG. 11 is an explanatory diagram of a dictionary configuration example.

FIG. 12 is a diagram illustrating a specific example of conventional LZW decoding.

FIG. 13 is an explanatory diagram of a conventional dictionary search tree by LZW encoding.

FIG. 14 is an explanatory diagram of a character string and a dictionary index by conventional encoding.

[Explanation of symbols]

 10: Dictionary 10-1 to 10-N: Division dictionary 12: CPU 14: Program memory 16: Control software 18: Maximum matching length search software 20: Encoding software 22: Decoding software 26: Look-up table (LUT) 30 : Data memory 32: Data buffer

────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hirotaka Chiba 1015 Kamidadanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited (56) References JP-A-4-149766 (JP, A) JP-A-3-204234 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 5/00 G06F 17/22 H03M 7/40

Claims (4)

(57) [Claims] In a data compression method, an input character string is encoded by specifying a reference number of a subsequence having a maximum length matching among already encoded subsequences registered in a dictionary (10). A plurality of divided dictionary groups (10-1 to 10-N) are provided as the dictionary (10), and divided dictionary numbers (Di) are sequentially assigned.
For each (10-1 to 10-N), at least all character types are initially registered with a reference number for each character, and when encoding an input character string, the character code of the last character of the previously encoded character string A specific division dictionary (10-i) is designated and encoded by a classification code based on (C1) and a character code code (C0) of a character immediately before the last character.
Data compression method. 2. The data compression method according to claim 1, wherein a part of the character code (C1) of the last character of the character string which has been encoded immediately before and the character code (C1) immediately before the last character.
0) a specific division dictionary (10-i) is specified and encoded by a classification code based on a part of
Data compression method. 3. A method according to claim 1, wherein an input character string is designated by a reference number of a subsequence having a maximum matching length among already encoded subsequences registered in a dictionary, and an original character string is obtained from an encoded character word. In the data restoration method for restoring a column, a plurality of divided dictionary groups (10-1 to 10-N) are provided as the dictionary (10), and divided dictionary numbers (Di) are sequentially assigned. For each (10-1 to 10-N), at least all character types are initially registered with reference numbers in units of one character, and when the input code string is restored, the character code (C1) of the last character of the restored character string immediately before ) And a classification code based on the character code (C0) of the character immediately before the last character, and designates a specific divided dictionary (10-i) for restoration. 4. In the data recovery method of claim 3 Symbol placement, the last character of the character code of the restored character string immediately before (C
1) and a classification code based on a part of the character code (C0) immediately before the last character.
0-i) data recovery, characterized in that to restore specified by the
Original method.