JP3241787B2 - 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 for compressing data using a Jib-Lempel 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 been rapidly increasing. When dealing with a large amount of data, by compressing the amount of data by omitting redundant portions in the data, the storage capacity can be reduced or the data can be transmitted faster.

2. Description of the Related Art Universal coding has been proposed as a method for compressing various data in a single system.
Here, the field of the present invention is not limited to compression of character codes,
Although it can be applied to various data, in the following, one word unit of data will be referred to as a character, and data connected with arbitrary words will be referred to as a character string, following the name used in information theory.

[0003] Typical methods of universal coding include Ziv-Lempel coding and arithmetic coding. Two algorithms have been proposed for the Jib-Lempel code, a slide dictionary type (also called universal type) and a dynamic dictionary type (also called incremental decomposition type).

Further, as an improvement of the slide dictionary type algorithm, there is an LZSS code (TC Bell, “Better O.
PM / L Text Compression ”, IEEE Trans. On Commun., V
ol.COM-34, No.12, Dec. 1986) and LHarc used in personal computers. As an improvement of the dynamic dictionary type algorithm, there is an LZW (Lempel-Ziv-Welch) code (TAWelch, “A Technique for High-Performance D
ata Compression ", Computer, June 1984).

[0005] These improved methods are used for compressing a file in an auxiliary storage device or compressing transmission data in a modem.

[0006]

2. Description of the Related Art A slide dictionary type algorithm and a dynamic dictionary type algorithm in conventional Jib-Lempel coding will be described. 1. Slide dictionary type algorithm The slide dictionary type algorithm is a method that requires a large amount of calculation but can obtain a high compression rate. That is, the encoded data is divided into the longest sequence that matches from an arbitrary position in the past data sequence (subsequence) and encoded as a copy of the past character string.

FIG. 10 shows a principle diagram of a universal type Jib-Lempel code encoder. The P buffer 12 stores encoded input data, and the Q buffer 10 receives data to be encoded. The character string in the Q buffer 10 is compared with the series in the P buffer 12, and P
Find the longest matching character substring in buffer 12,
To specify the longest character string in the buffer 12, the next set of information is encoded.

[0008]

[Table 1]

Next, the encoded character string in the Q buffer 10 is transferred to the P buffer 12 and new data is transferred to the Q buffer 1.
Enter 0. Hereinafter, the same operation is repeated, and the data is decomposed into partial strings and encoded. As described above, in the Jib-Lempel code, a current character code sequence is encoded as a copy from an encoded past sequence. When the Jib-Lempel code is used, the character code document information can be compressed to about 1/2. [QIC-122 code] QIC (Quauter Inch Cartrrige Standard In), a group of manufacturers centering on 3M
c.) is a code adopted as a standard compression method for a 1/4 inch cartridge magnetic tape.

FIG. 11 is a flowchart showing the algorithm of the QIC122 code, and performs the following processing. S1: has a history of 2048 bytes as a P buffer,
Empty the contents. Also, the input data is packed in the Q buffer. S2: Search for the longest character string S in the P buffer that matches the character string in the Q buffer.

S3: If the longest character string S searched for is two or more characters, the process proceeds to the copy mode of S5, and if it is one character, S is performed.
4 to the raw data mode. S4: In the raw data mode, a set of [flag bit 0] and [raw data 1 byte] is encoded. S5: In the copy mode, [Flag bit 1] is fixed-length coded, and a set of [Appearance position of character string S] [Match length] is coded.

S6: The encoded character string in the Q buffer is transferred to the P buffer, and the same number of characters are input to the Q buffer. At the same time, the oldest characters in the character frame transferred from the O buffer to the P buffer are discarded from the P buffer. S7: The processes of S1 to S6 are repeated until there is no more input data. FIG. 12 shows QIC- expressed in the BNF meta-language.
1 shows a codeword format of a 122 code. The meta symbols used in the BNF meta language have the meanings shown in FIG.

The codeword format of the QIC-122 code shown in FIG. 12 will be described in detail as follows. (1) A compressed stream (Compressed Stream) is composed of a compressed string (Compressed String) and an end marker. (2) The compressed string is represented by an ASCII raw byte following identification bit 0 for raw data, and a compressed byte following identification bit 1 for compressed data.

(3) The ASCII raw byte is represented by 8 bytes as 1 byte. (4) A compressed byte is a set of an offset (start position) and a length (match length). (5) The offset (start position) is represented by 7 bits in the case of the identification bit 1. The identification bit 0 is represented by 11 bits.

(6) The end marker is 11000000
0, and the offset is 0. (7) The bit b is 0 or 1. (8) The length (match length) is represented by a variable length code as shown in FIG. FIG. 14 shows a specific example of encoding of the QIC-122 code. In FIG. 14, the character string "ABAAAAAA
CABA "is input.

For the first three characters "ABA", P
Since the number of matching characters in the buffer is one or less, a bit sequence of ASCII raw bytes is output. The five "A" characters from the fourth character to the eighth character match the character "A" immediately before in the P buffer, so the compressed byte identification bit 7-bit offset identification bit Offset = 1 Length = 5 bytes The series "1 1 000 0001 110
Output as "0".

Here, the offset value indicating the start position of the substring having the maximum length coincidence indicates the number of the latest registered position (address) in the P buffer from the previous position. Since the ninth character "C" is not in the P buffer, it outputs an ASCII raw byte. The 10th to 12th characters "ABA" are P
Since the three characters have already been registered as the three characters from the head of the buffer, a bit sequence “1 1000100101” consisting of compressed byte identification bits, 7-bit offset identification bits, offset = 9 length = 3 bytes is output.

As described above, since all the input characters have been encoded, "1 100000000" is output as end data, and the processing is terminated. 2. Dynamic Decomposition (Incremental Decomposition) Algorithm This algorithm is known to have a lower compression ratio than the universal type, but is simple and easy to calculate.

In the incremental decomposition type diplen-pel code, if the input symbol sequence is X = aabababaa..., The incremental decomposition into the component sequence X = X ₀ X ₁ X ₂ . First, X ₁ is a column of the longest removing the right edge of the symbol of the preformed component, and X = a · ab · aba · b · aa ····. Thus, X ₀ = λ (empty _{string) X 1 = X 0 a X} 2 = X 1 b X 3 = X 2 a, X 4 = X 0 b X 5 = X 1 a, can be decomposed,. Each of the component sequences that have been incrementally decomposed is encoded in the following set using the existing component sequences.

[0020]

[Table 2]

In other words, the incremental decomposition type algorithm obtains a coding pattern having the same maximum length among sub-sequences that have been decomposed in the past, and encodes the same as a duplicate of the sub-sequence that has been decomposed in the past. Improvements to the dynamic dictionary algorithm include LZW (Lempel-Ziv-Welch) codes (TAWelch, "A Tech
nique for High-Performance Data Compression ", ComPu
ter, June 1984) LZJ code (M. Jakobsson, "Comperssion of Character
Strings by An Adaptive Dictionar ", BIT, Issue 25, 19
1985, pp. 593-603). Next, the LZW code will be described. [LZW Code] FIG. 15 shows a flow of an LZW code encoding process. That is, the LZW encoding has a rewritable dictionary, divides the data of the input character code into different character strings, assigns numbers to the character strings in the order in which they appear, registers them in the dictionary, and inputs the character strings. A character string is represented and encoded only by the number of the longest matching character string registered in the dictionary. The technology of the dynamic dictionary code and the LZW code is disclosed in Japanese Patent Application Laid-Open No. 59-231683, U.S. Pat.
8,302. The encoding process in FIG. 15 is as follows.

S1: Encoding is started after a character string consisting of one character for all characters is registered in advance as an initial value. The number of dictionary registrations n is set as the number of character types A. Position the cursor at the beginning of the data. S2: Find the longest character string S registered in the dictionary that matches the character string from the cursor position.

S3: The dictionary number of the character string S is [log ₂
n] bits. Where [log ₂ n]
Is the minimum integer of log ₂ n or more. The dictionary registration number n is incremented by one. S4: A character string SC obtained by adding the first character C of the cursor to the character string S is registered in the dictionary. The cursor moves to the character after the character string S. Return to S2.

FIG. 16 is a flowchart showing the decoding of the LZW code, which is the reverse of the coding. The dynamic dictionary type algorithm has a high processing speed because a sequence in the dictionary is selected only from a sequence coded (sampled) in the past. However, there is a disadvantage that a high compression ratio cannot be obtained because only a part of the series of data that appeared in the past is included.

As an improved version of the dynamic dictionary type algorithm,
There is an LZJ code in which the amount of learning to a dictionary is increased so that encoding can be performed using only an index. [LZJ Code] FIG. 1 shows a processing flow of LZJ code encoding.
7 and the processing flow of the decoding is shown in FIG.

Here, the notation of a dictionary and a character string is defined as follows. A character set is represented by A, and a character string formed by combining the characters of the set A is represented by S. The i-th character of the character string S is S (i). Further, a plurality of partial character strings S
(I), S (i + 1),..., S (j) are replaced by S (i,
j). The dictionary is represented by D _h (S), and the dictionary tree (t
All partial character strings of a fixed length h in the character string S are registered as a path from the root of the (ree) to the leaf (leaf).

The LZJ encoding process of FIG. 16 is as follows. S1: Start encoding after registering one character of all character types in the dictionary as an initial value. The registration number n of the dictionary is set to the character type number A. The cursor k is set to 0. S2 to S5: All partial character strings of the character string S (1, k) have already been registered in the dictionary D _h (S (1, k)), assuming that encoding has been completed up to the k-th input character. S (k +
1) Encode from the character string of.

This will be described in detail as follows. S2: A dictionary D _h (S (1,
k)) the substring S (k +
1, k + z). S3: Dictionary number a _{x of} partial character string S (k + 1, K + z)
Is represented by [log ₂ n] bits and output. Where n
Is the current number of entries in the dictionary, and [log ₂ n] is log
It is the smallest integer of ₂ n or more. Here, the code word a _x represents the substring _{S (i x, j x)} . Each a _x is a dictionary D
_{h (S (1, j x} -1)), (i x ≦ j x ≦ i x + h, i
_x = j _x-1 +1).

S4: Partial character string S (kh + 2, k +
1),..., S (k + z−h + 1, k + z) are added to the dictionary by adding a dictionary number while incrementing n,
Construct a dictionary D _h (S (1, k + z)). S5: Set cursor k = k + z. S6: S1 to S5 are repeated until all characters are processed.

Here, the dictionary registration of the character string in step S4 is as shown in FIG. Next, LZ in FIG.
The J decoding process is as follows. S1: One character of all character types is registered as an initial value in the dictionary as in S1 of FIG. The registration number n of the dictionary is set to the character type number A. Cursor k =
Set to 0.

S2 to S4: Dictionary number a_w Is decrypted,
Character string S (1, j_w ), And the dictionary D
_h (S (1, j_w )) Has been restructured. Then the codeword
a _{w + 1} Is decrypted. The details are as follows. S2: code word a_{w + 1} D from the dictionary number
_h(S (1, j_w ))_{w + 1} , J_{w + 1} )
To restore. The subsequence S (i_{w + 1} , J_{w + 1} ) Is the root in the dictionary
(Root) to address a_{w + 1} Sentence represented by a node
It is a character string.

S3: After decoding the character string S (1, j _{w + 1} ), the dictionary D _h (S (1, j _{w + 1} )) is constructed in the same manner as S4 in FIG. S4: Set cursor k = j _{w + 1} . S5: S1 to S4 are repeated until all characters are processed.

[0033]

As described above, in the conventional dynamic dictionary type Jib-Lempel coding, the dictionary is formed in a tree structure, and compression is performed by comparing a registered character string with an input character string. There is an advantage that a column search is quick and encoding can be performed at high speed. However, on the other hand, there is a disadvantage that the capacity of the dictionary is increased because all character string variations must be registered in the dictionary.

On the other hand, in the slide dictionary type Jib-Lempel encoding, a fixed amount of raw data which has been encoded is stored in the dictionary, so that the dictionary has a fixed capacity, and the latest character can be updated by registering and deleting the dictionary. The advantage is that the columns can be used as a dictionary. However, there is no learning function that a character string that appears more frequently as in a dynamic dictionary type is regarded as a character string having a longer regularity.

Therefore, when the same character string frequently appears, the dictionary is occupied by repetition of the same character string,
There was a disadvantage that the efficiency of the dictionary was reduced. The present invention has been made in view of such a conventional problem, and is an encoding using a slide dictionary, in which the efficiency of the dictionary when the same character string appears repeatedly is improved, and a high compression rate is obtained. An object of the present invention is to provide a data compression method as described above.

[0036]

FIG. 1 is a diagram illustrating the principle of the present invention. The present invention performs encoding according to a dynamic dictionary type algorithm as preprocessing when encoding according to a slide dictionary type algorithm, and encodes a code string obtained by this encoding with an input character string. Basically, slide dictionary type encoding is performed as processing.

That is, as shown in FIG. 1A, the input data is decomposed into sub-sequences, and this sub-sequence is represented by the reference number of the longest-matching sub-sequence registered in the dictionary and encoded.
Encoding means (dynamic dictionary-type encoding means) 10 and a dictionary number encoded by the first encoding means 10 are input as a character string, and encoded characters stored in a dictionary that longest matches the input character string A second encoding means (slide dictionary type encoding means) 20 for encoding with a column appearance position and a matching length is provided.

Here, the first encoding means 10 includes an appearance frequency counting means for counting the number of times of use of the encoding of the partial string registered in the dictionary, and the encoded partial string of the dictionary which longest matches the input character string. When the search is performed, the reference number is encoded and output only when the appearance frequency of the search subsequence is equal to or higher than a predetermined threshold, and when the occurrence frequency is smaller than the threshold, the input subsequence is output as it is.

The first encoding means 10 has a dictionary for registering an encoded character string with a reference number attached thereto, and stores the encoded substring in the dictionary which longest matches the input character substring. LZW encoding is performed in which a search is performed and specified and specified by a reference number, and after encoding, the subsequence obtained by adding the next input character to the reference number is added to a new reference number and registered in the dictionary. Further, the first encoding means 10 has a dictionary for registering encoded character strings with reference numbers, and searches for encoded subsequences in the dictionary which longest match a subsequence of the input character string. After encoding, each subsequence of the encoded character string is sequentially set as a prefix subsequence, and a subsequence of a fixed length is created by sequentially adding the subsequences in the dictionary to the prefix subsequence. LZJ encoding that registers all in the dictionary.

FIG. 1B shows a data compression method according to the present invention.
As shown in FIG. 2, a dictionary having a decoded data sequence is sequentially stored. Code data encoded by the second encoding unit 20 is input, and a dictionary number is determined based on an appearance position and a matching length of the dictionary specified by the code data. Or first decoding means 3 for decoding a character string
0 and a dictionary in which the decoded character string is registered with a reference number added thereto, and the second decoding means 40 decodes the character string by referring to the dictionary based on the decoded data decoded by the first decoding means 30.
And characterized in that:

[0041]

According to the data compression method of the present invention having such a configuration, efficient compression can be performed by combining the encoding by the dynamic dictionary type algorithm and the slide dictionary type algorithm. For example, when the same character string appears repeatedly, the repetition of the same character string is encoded into one character designated by one reference number by encoding according to the dynamic dictionary type algorithm, which is further compressed by a slide type dictionary. This allows efficient compression even when the same character string is repeated using a fixed amount of slide dictionary.

In the encoding by the dynamic dictionary type algorithm, the longest matching character string is encoded for a character string that appears frequently. It is expressed as short (small reference number), and the match length in the next slide dictionary coding can be made shorter, and a high compression ratio can be obtained by performing variable length coding on the match length.

[0043]

FIG. 2 is a block diagram showing an embodiment of an encoding apparatus used for a data compression system according to the present invention. 2, the encoding device includes a dynamic dictionary encoding unit 10 as a first encoding unit, a slide dictionary encoding unit 20 as a second encoding unit, and a control circuit 50.

The dynamic dictionary encoding unit 10 stores an input buffer 12 for storing an input character string from an input terminal 11, a collating circuit 13, a dictionary memory 14 functioning as a dynamic dictionary, and an encoded dictionary number. Register 15 is provided. The slide dictionary encoding unit 20 includes a Q buffer 21 for storing a code string of dictionary numbers encoded by the dynamic dictionary encoding unit 10, a P buffer 22 functioning as a slide dictionary, a collating circuit 23, and a variable length code. Circuit 24 is provided.

FIG. 3 is a flowchart schematically showing an encoding process of the encoding device shown in FIG. First, step S1
Performs encoding for converting input data (input character string) into a dictionary number using a dynamic dictionary. As the encoding using the dynamic dictionary, for example, an LZW code or an LZJ code can be used.

Subsequently, in step S2, an encoded longest matching character string registered in the slide dictionary is searched for the encoded character string consisting of the dictionary number obtained in the slide dictionary encoding in step S1, and It is represented by a set of [appearance position] and [match length] or [one character of raw data] and encoded. FIG. 4 is a flowchart showing processing when LZJ encoding is performed as the dynamic dictionary encoding shown in step S1 of FIG.

The LZJ encoding algorithm shown in FIG. 4 is basically the same as the conventional LZJ encoding algorithm shown in FIG.
The following points in S4 are different. That is, step S in FIG.
In the 2, occurrence frequency among the registered character strings in the dynamic dictionary so that searches a string longest matches the input character string from among more strings predetermined threshold h _f. Therefore, the string of threshold h _f is smaller than the appearance frequency is excluded from the search.

Subsequently, the dictionary number of the longest matching character string searched in step S3 is expressed as a fixed-length bit number that can represent the maximum number m that can be registered in the dictionary and output. Further, in step S4, the number of appearances is counted up by one for each character string registered in the dictionary included in the character string encoded at the same time as the registration of all the character strings as shown in FIG. Count the frequency.

FIG. 5 is an explanatory diagram showing the tree structure of character strings registered in a dictionary created by dynamic dictionary coding using the character abcd as an example and the appearance frequency of each character. FIG.
(A) shows a tree structure of a registered character string obtained by performing dynamic dictionary encoding on four characters abcd, and indicates a reference number registered in the dictionary at the upper left of each character. , The appearance frequency is shown in the lower frame of the character. Since the four characters abcd have been initially registered and coding has been started, the frequency of appearance is not particularly counted.

This example shows a registration state in which the depth h of the dictionary is h = 3, each node corresponds to a character string, and the number of appearances of the character string is counted at each node. Assuming that the character at the node is the parent, the sum of the appearances of the child's character is the appearance frequency of the parent. FIG. 5B shows a tree structure of a character string to be searched when the threshold value h _f = 3 in step S2 of FIG. Here, the dictionary reference number of a character string having a high frequency of appearance with a threshold frequency _hf = 3 or more takes discrete values. However, since the slide dictionary type coding is performed subsequently, the coding efficiency is reduced. No effect.

In the case of dynamic dictionary coding, FIG.
Since encoding is performed on a character string with a high appearance frequency shown in FIG. 2B, a character string with a low appearance frequency is not encoded. For example, the character d is represented as one character in FIG. 5B, and the character code matches the dictionary number. Further, as shown in the flowchart of FIG.
In the case of using a modification method of LZJ encoding as shown in FIG. 4 for dynamic dictionary encoding performed as preprocessing of slide dictionary encoding, the value of the reference number is determined as a character string equal to or greater than a predetermined threshold. It is possible to select a character string of an arbitrary length equal to or less than a predetermined value which is equal to or less than a predetermined value.

The dynamic dictionary type coding performed as preprocessing is not limited to the modification of the LZJ coding shown in FIG.
For example, a simpler method may be used, such as replacing a frequently used combination in a unit of two characters with a number.
In this case, the character string to be replaced has a fixed length.
Further, as a dynamic dictionary type encoding performed as preprocessing,
In the LZW encoding shown in FIG. 15, as in the LZJ encoding, the appearance frequency may be counted, and encoding may be performed on a character string having an appearance frequency equal to or higher than a predetermined threshold.

FIG. 6 is a flowchart showing an encoding algorithm of the QIC122 code as a specific example of the slide dictionary encoding performed in step S2 of FIG. The encoding of the QIC122 code shown in FIG. 6 is performed in such a manner that input data to be encoded is applied to a code string composed of dictionary reference numbers obtained by dynamic dictionary encoding. The processing itself is shown in FIG. This is the same as the encoding algorithm of the conventional QIC122 code. The difference due to processing a code string obtained by dynamic dictionary coding as a processing target is that one word of data handled by a slide dictionary is conventionally one byte unit.
In the case of the present invention, one word is the number of bits of the dictionary number.

Next, the encoding operation will be described with reference to the embodiment of FIG. The input character string given to the input terminal 11 of the dynamic dictionary encoding unit 10 is input to the input buffer 12 at regular intervals. The matching circuit 13 searches for a longest matching character string from a character string of a predetermined threshold or more in a dictionary memory 14 functioning as a dynamic dictionary, and stores the dictionary number of the searched character string via a register 15 as a slide dictionary code. Output to the Q buffer 21 of the conversion unit 20.

At this time, the control circuit 50 registers all character strings of a fixed length h in the encoded character strings in the dictionary memory 14 with reference numbers. On the other hand, in the slide dictionary encoding unit 20, the control circuit 23 searches for a character string in the P buffer 22 that longest matches the input character string of the Q buffer 21 which is the dictionary reference number. If there are two or more characters, a set of [appearance position] and [match length] is output. If the match length is one character, raw data (dictionary reference number) 1
Output a character.

At this time, the control circuit 50 transfers the encoded character string from the Q buffer 21 to the P buffer 22, discards the oldest character of a fixed length h in the P buffer,
The same number of character strings are input from the register 15 of the dynamic dictionary encoding unit 10. The code data output from the matching circuit 23 is supplied to a variable length coding circuit 24, converted into a variable length code, and output from an output terminal 25 as a compressed signal.

FIG. 7 is a block diagram of an embodiment showing an embodiment of a decompression device used for the data compression system of the present invention. 7, the restoration device includes a slide dictionary decoding unit 30 as a first decoding unit, a dynamic dictionary decoding unit 40 as a second decoding unit, and a control circuit 60. The slide dictionary decoding unit 30 is provided with a variable length decoding circuit 32, a clipping duplication circuit 33, a Q buffer, and a P buffer 35 as a slide dictionary. The dynamic dictionary decoding unit 40 has a register 41 and a dictionary memory 4 functioning as a dynamic dictionary.
2, and an output buffer 43 are provided.

FIG. 8 is a flowchart schematically showing a decoding process of the decoding device of FIG. That is, in the decoding of the present invention, the reverse procedure of the encoding shown in FIG. 3 is taken. First, in step S1, a code string of a dictionary number is obtained by decoding a compressed code obtained by the coding apparatus shown in FIG. 2 using a slide dictionary. Subsequently, the original character string is restored by referring to the dynamic dictionary based on the dictionary number decoded in step S2, so that the original data before encoding is restored.

FIG. 9 shows L as the dynamic dictionary decoding unit 40 in FIG.
FIG. 19 is a flowchart showing processing when ZJ decoding is performed, which is basically the same as the conventional LZJ decoding shown in FIG. 18, with the difference that the decoding is performed by the preceding slide dictionary decoding in step S2. That is, a dynamic dictionary is searched for the dictionary number represented by the fixed number of bits to restore the original character string.

Next, the operation of the restoration apparatus in FIG. 7 will be described as follows. The compressed code given to the input terminal is decoded by the variable length decoding circuit 32. If the decoded code is a code in the copy mode consisting of a pair of “appearance position” and “match length”, the cut-out copy circuit 33 buffers the code. A dictionary reference number of a predetermined part specified by the appearance position of 35 and the matching length is cut out.

At this time, the control circuit 60 stores the restored dictionary reference number in the Q buffer 34 and shifts the old dictionary reference number to the P buffer 35 in accordance with the storage in the Q buffer 34. The oldest dictionary reference number in 35 is discarded by the same number. On the other hand, if the code decoded by the variable length decoding circuit 32 is a raw data code, it is not necessary to refer to the P buffer 35, so that it is output as it is, and is input to the Q buffer 34 at the same time.
Perform a shift of the old data to 5.

The character string composed of the dictionary number restored by the slide dictionary decoding unit 30 is set in the register 41 of the dynamic dictionary decoding unit 40. The dictionary memory 42 is accessed by the dictionary number of the register 41, and the designated reference is specified. The character string stored in the number is restored and the character string is output from the output terminal 44 to the outside. At this time, the control circuit 60 registers a new character string formed from the character string composed of the dictionary number input to the register 41 in the dictionary memory 42.

In the above embodiment, the case where encoding and decoding are performed with a hardware configuration as shown in FIGS. 2 and 7 is taken as an example, but the present invention is not limited to this. Without limitation, the functions of the dynamic dictionary encoding unit 10, the slide dictionary encoding unit 20, the slide dictionary decoding unit 30, and the dynamic dictionary decoding unit 40 are realized by a control program, and a combination of a computer and a dictionary memory is used. Of course, it may be realized by a software configuration.

[0064]

As described above, according to the present invention, a character string which appears frequently with dynamic dictionary coding prior to slide dictionary coding can be compressed into one character. In the conversion, more data can be stored in the slide dictionary, and the longest matching character string can be searched from more candidates, so that a high compression rate can be obtained.

In the dynamic dictionary type encoding, encoding is performed on a character string which appears frequently, and in the subsequent slide dictionary type encoding, it is easy to appear. Since the part is represented short, the matching length of the longest matching character string in the slide dictionary type coding can be further shortened, and a high compression rate can be finally obtained by performing variable length coding.

Further, when the same character string appears repeatedly, it is compressed to one word in the dynamic dictionary type coding performed as preprocessing, so that the same character when only the slide dictionary type coding is performed The problem that the repetition of columns cannot be efficiently compressed can be solved and a high compression ratio can be obtained.

[Brief description of the drawings]

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

FIG. 2 is a configuration diagram of an embodiment of an encoding device according to the present invention.

FIG. 3 is a flowchart showing an encoding process according to the present invention;

FIG. 4 is a flowchart illustrating an LZJ encoding algorithm performed as dynamic dictionary encoding according to the present invention;

FIG. 5 is an explanatory diagram showing counting of appearance frequency and encoding of a character string having a frequency equal to or higher than a threshold value in the LZJ encoding of FIG. 4;

FIG. 6 is a flowchart showing an encoding algorithm of a QIC122 code according to the present invention.

FIG. 7 is a configuration diagram of an embodiment of a decoding device according to the present invention.

FIG. 8 is a flowchart showing a decoding process according to the present invention.

FIG. 9 is a flowchart showing an LZJ decoding algorithm performed as dynamic dictionary decoding according to the present invention;

FIG. 10 is a principle diagram of slide dictionary type encoding.

FIG. 11 is a flowchart showing an encoding algorithm of a conventional QIC122 code;

FIG. 12 is an explanatory diagram of a format of a QIC122 code.

FIG. 13 is an explanatory diagram of the BNF meta-language used in FIG.

FIG. 14 is an explanatory diagram showing a specific example of encoding using a QIC122 code;

FIG. 15 is a flowchart showing a conventional LZW encoding algorithm;

FIG. 16 is a flowchart showing a conventional LZW decoding algorithm;

FIG. 17 is a flowchart showing a conventional LZJ encoding algorithm.

FIG. 18 is a flowchart showing a conventional LZJ decoding algorithm.

FIG. 19 is an explanatory diagram showing registration of a character string in LZJ encoding.

[Explanation of symbols]

 10: first encoding means (dynamic dictionary encoding unit) 11, 31: input terminal 12: input buffer 13, 23: matching circuit 14, 42: dictionary memory (dynamic dictionary) 15, 41: register 20: 2 coding means (slide dictionary coding unit) 21, 34: Q buffer 22: P buffer (slide dictionary) 24: variable length coding circuit 25, 44: output terminal 30: first decoding means (slide dictionary decoding) Unit) 32: variable length decoding circuit 33: clipping duplication circuit 40: second decoding means (dynamic dictionary decoding unit) 43: output buffer

────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hirotaka Chiba 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (56) References JP-A-63-294134 (JP, A) JP-A-62-137633 (JP, A) JP-A-3-247167 (JP, A) JP-A-3-209923 (JP, A) JP-A-3-78322 (JP, A) JP-A 4-116738 (JP, A) Kaihei 3-208469 (JP, A) (58) Field surveyed (Int. Cl. ⁷ , DB name) G06F 5/00 H03M ^7/ 30-7/40

Claims (5)

(57) [Claims] 1. A first encoding means (10) for decomposing input data into sub-sequences and encoding the sub-sequences by expressing the sub-sequences with reference numbers of longest-matching sub-sequences registered in a dictionary; A second encoding for inputting the dictionary number encoded by the encoding means (10) as a character string, and encoding the encoded character string with the start position and the coincidence length of the encoded character string held in the dictionary which longest matches the input character string. Means (20). 2. The data compression method according to claim 1, wherein
The first encoding unit (10) includes an appearance frequency counting unit that counts the number of times of use of encoding of the substring registered in the dictionary, and searches for an encoded substring of the dictionary that longest matches the input character string. At this time, a data compression method characterized in that a reference number is coded and output only when the appearance frequency of the search subsequence is equal to or higher than a predetermined threshold, and the input subsequence is output as it is when the frequency is smaller than the threshold. 3. The data compression system according to claim 1, wherein said first encoding means (10) has a dictionary for registering encoded character strings with reference numbers attached thereto. A coded subsequence in the dictionary that longest matches the character subsequence is searched and encoded by specifying a reference number, and a subsequence obtained by adding the next input character to the reference number after the encoding is added to a new subsequence. A data compression method wherein a reference number is assigned and registered in the dictionary. 4. A data compression system according to claim 1, wherein said first encoding means (10) has a dictionary for registering encoded character strings with reference numbers, Searching for an encoded subsequence in the dictionary that longest matches the subsequence of the column and encoding by specifying by reference number, and after the encoding, each subsequence of the encoded character string is sequentially a prefix subsequence, A data compression method characterized in that a fixed-length subsequence is created by adding the prefix subsequence to a subsequence in the dictionary, and all the subsequences are registered in the dictionary. 5. A data compression system according to claim 1, further comprising a dictionary in which decoded data strings are sequentially stored, wherein code data encoded by said second encoding means (20) is inputted, and A first decoding unit (30) for decoding a dictionary number or a character string based on an appearance position and a matching length of the dictionary specified by the data; and a dictionary in which the decoded character string is registered with a reference number and registered. (1) a second decoding means (40) for decoding a character string by referring to a dictionary based on the decoded data decoded by the first decoding means (30).