JP3038234B2 - Dictionary search method for data compression equipment - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

【Overview】

Regarding a dictionary search method of a data compression device using LZW coding as an improvement of the incremental decomposition type, which is a kind of universal coding, the dictionary search time is made possible by enabling high-speed reading of a dictionary memory using a list structure of an external hash method. The dictionary memory has a list structure according to the external hash method consisting of a first memory (index memory), a next memory (concatenated memory), and an extended memory storing candidate characters, and a continuous index address of the nested memory. Configured in the address,
The configuration is such that the search is performed by a continuous address following the first search based on the input character, thereby speeding up.

[Industrial applications]

The present invention relates to a dictionary search method for a data compression device using 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, the storage capacity can be reduced by omitting redundant portions in the data and compressing the data amount.
It can be transmitted quickly. Universal coding has been proposed as a method for compressing such various data in a single 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. As a typical method of the universal code, there is a Ziv-Lempel code (for example, for example, for example,
Munakata "Ziv-Lempel Data Compression Method", Information Processing, Vol.2
6, No. 1, 1985). Two algorithms of the universal type and the incremental parsing type have been proposed for the dihurempel code. Furthermore, as an improvement of the universal algorithm, LZ
There is an SS code (TCBell, “Better OPM / L Text Compres
sion ", IEEE Trans.on Commun., Vol.COM-34, No. 12, DEC.
1986). In addition, as an improvement of the incremental decomposition algorithm, LZW
(Lempel-Ziv-Welch) code (TAWelch, "A Te
chnique for High-Performance Data Compression ", Co
mputer, June 1984). Among these codes, the LZW code is used for file compression of a storage device or the like because of its high speed processing and the simplicity of the algorithm.

2. Description of the Related Art FIG. 7 shows a conventional encoding process flow using an LZW code, and FIG. 8 shows a decoding process flow. First, the LZW encoding process has a rewritable dictionary,
The input character string is divided into different character strings (subsequences), and the character strings are registered in the dictionary with reference numbers in the order of appearance, and the character string currently input is registered in the dictionary. It is represented by the reference number of the longest matching character string and encoded. FIG. 9 is an explanatory diagram of LZW encoding, and FIG.
An explanatory diagram of the decoding is shown, and FIG. 11 shows an example of a dictionary configuration created at the time of the decoding. In FIGS. 9, 10, and 11, for simplicity of explanation, abc 3
An example of compressing and decompressing data composed of character combinations is described. In the LZW encoding process of FIG. 7, first, in step S1 (hereinafter, "step" is omitted), a character string composed of one character for every character is registered in the dictionary as an initial value, and then encoding is started. In the encoding of S1, a dictionary is searched using the first character K input to obtain a reference number ω, and this is used as the initial character string. Next, the next character K of the input data is read in S2, and it is checked whether the character input is completed in S3. Then, the process proceeds to S4, and the character K read in S2 is added to the initial character string ω obtained in S1. A search is performed to determine whether the added extended character string (ωK) exists in the dictionary. If the character string (ωK) is not in the dictionary in S4, the process proceeds to S6 and S1
Is output as a code word code (ω), a new reference number is added to the character string (ωK) and registered in a dictionary, and the input character K of S2 is further referred to as a reference number ω. And increment the dictionary address n to S2
To read the next character K. On the other hand, if the character string (ωK) is found in the dictionary in S4, the character string (ωK) is replaced with the reference number ω in S5, and the process returns to S2 again and the maximum matching length is reached until the character string (ωK) cannot be searched from the dictionary in S4. Continue searching. The following specifically describes the LZW encoding with reference to FIGS. First, the input data input of FIG. 9 is read from left to right.
When the first character a is entered, there is no matching character string in addition to the character a in the dictionary, so OUTPUT CODE 1 (reference number ω)
Is output as a code word. Then, the character a is set to the initial character string ω. Next, assuming that the second character b is input, since the extended character string ωK = ab obtained by adding the input character to the initial character string ω is not in the dictionary, the OUTPUT CODE 2 of the character b is output as a code word. . Then, a reference number 4 is added to the extended character string ωK = ab and registered in the dictionary. The actual dictionary registration is registered as a character string 1b as shown on the right side of FIG. And the letter 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 b is not in the dictionary, so the OUTPUT CODE 1 of the character a is output as a code word. After that, the extended character string ωK = ba is represented by 2a, and is added to the reference number 5 and registered in the dictionary. Then, the character a becomes a new initial character string ω. For the fourth input character b, the expanded character string ωK = ab is
Since the code word 4 of 1b has already been registered in the dictionary, 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. Since this extended character string ωK = abc is not registered in the dictionary, OUTPUT CODE 4 of the character string ab = 1b is output as a code word, and the extended character string ωK = abc is written in the dictionary as code word 6 in the form of 4c. sign up. Hereinafter, similarly, this processing is continued. The decoding process in FIG. 8 performs the reverse operation of the encoding in FIG. In the LZW decoding of FIG. 8, similarly to the encoding, a character string consisting of one character for every character is registered in the dictionary as an initial value before decoding starts. First, the first code (reference number) is read in S1, and the current CO
DE is OLDcode, and since the first code corresponds to one of the reference numbers of one character already registered in the dictionary, a character code (K) that matches the input code CODE is searched for and the character K is output. Note that the output character K is set in FINchar for later exception processing. Next, the process proceeds to S2, where the next code is read and set as CODE in CODE. In S3, it is checked whether or not there is a new code, that is, whether or not the code input is completed, and the process proceeds to S4, where it is checked whether or not the code CODE input in S3 is defined (registered) in the dictionary. Normally, the input code word has been registered in the dictionary in the previous processing, so the process proceeds to S5, where the character string code (ωK) corresponding to the code CODE is read from the dictionary, and the character K
And temporarily stack the reference number CODE (ω) with the new code
Returning to S5 again as a CODE, the steps of S5 and S6 are recursively repeated until the reference number ω reaches one character K. Finally, the process proceeds to S7 and the characters stacked in S6 are LIFO (Last In Fast Ou).
t) Pop up and output in the format. At the same time, in S7, the code ω used last time and the first 1
A new reference number is added to the character string representing the character K as a set (ωK) and registered in the dictionary. The LZW decoding process will be specifically described with reference to FIG. First, in FIG. 11, 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 a character string a having a reference number that matches the code word 1 by referring to the dictionary and output. Similarly, the next code word 2 is replaced with the character b and output. At this time, a character string ωK obtained by combining the code word 1 processed last time and the first character b of the character string decoded this time.
Add a new reference number 4 to = 1b and register it in the dictionary. The third code word 4 is the character string 1b obtained by searching the dictionary
And replace it with the string ab to output the string ab. At the same time, a new reference number 5 is added to the character string ωK = 2a (= ba), which is a combination of the code word 2 processed last time and the first character a of the character string decoded this time, and registered in the dictionary. Hereinafter, similarly, this processing is repeated. In the LZW decoding of FIG. 11, 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, a character string 5b obtained by adding the first character b of the previously decoded character string ba to the previously processed code word 5
, And perform exception processing, replacing 5b = 2ab = bab and outputting. Then, after outputting 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 corresponds to steps S4 and S8 in the decryption processing flow of FIG.
Finally, the output of the character string in S7 and the registration in the dictionary in which the reference number is added to the new character string are performed in S7. In the LZW decoding of FIGS. 8 and 11, the case where the dictionary is created in real time while decoding the code on the decoding side has been described, but the dictionary created at the time of encoding is used as it is as a copy to the decoding side May be encoded. In this case, the exception processing on the decoding side becomes unnecessary. When LZW encoding is performed according to the procedure shown in the processing flow diagram of FIG. 7, the entire dictionary must be searched at the worst every time a character string is searched. There was such a problem. Therefore, in the conventional dictionary search method, the processing speed is increased by using an external hashing method (open hashing or chaining). First, in general dictionary search by hash method,
When considering a set S composed of a plurality of character strings, 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. . Assuming that a storage location of a character string, that is, an address from 0 to m−1 is assigned to a hash table, the function h: S → [0,1,... And set the address of the character string x of the set S to 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. In the hash method, the size of the set S is usually the number of addresses m
Used when it is much larger than. 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. 12, a linked list is prepared for each hash address i indicated by an index (directory), and the hash address h (x) =
The character string x of i is stored in order from the head of the linked list. Each linked list with the same hash address h (x) is called a bucket. LZ using list structure of external hash method for dictionary search
FIG. 13 shows a flowchart of the W encoding process. FIG. 14 shows the configuration of a dictionary memory according to the external hash method, and specifically shows a search procedure and a registration procedure of LZW encoding using the tree structure of the encoded character string shown in FIG. 15 as an example. ing. First, in FIG. 14, the dictionary memory includes a first memory (First Memory) 100 and a next memory (Next Memory).
y) Extension memory of 200 and next memory 200
Memory) 300. Here first memory 10
0 corresponds to the index (directory) of the external hash method shown in FIG. 12, the next memory 200 corresponds to “next” in the linked list in FIG. 12, and the extended memory 300 corresponds to “name” in FIG. Corresponding to The tree structure in FIG. 15 is composed of the characters K ₁₀ , K ₂₁ , K ₂₂ ,.
·, K ₄₁ is already registered, K ₄₂ indicated by a broken line shows the case where the newly registered. The hierarchy in the tree structure is indicated by an i counter in the process of FIG. 13, and the number of characters in the same hierarchy is indicated by a j counter. Therefore, the registration address of each character is represented as ω _ij . String now included in the registered tree structure of FIG. 15 _{"K 10, K 22, K 32} , K 42 " LZW coding and registration by the dictionary search in accordance with the processing flow of FIG. 13 when the input Can be described as follows. In FIG. 13, first, the following initialization processing is performed in S1. Initialize the dictionary to include the first character. For example, if the alphabet is 26 characters, the character code is registered as it is in the first memory in FIG. 14 as a hash address. For Figure 15, it means the state in which the character K ₁₀ in the tree top is registered in the address omega _10. The current character registration number n in the dictionary is set to the number of characters registered above. For 26 alphabets, n =
It becomes 26. The first character K input is assumed to be the initial character string i. For Figure 15, the prefix string i = 1 since the first input character is K _10. In the following processing flow, the initial character string i is described as an i counter. 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], these are initialized to 0. After the initialization of S1, the process proceeds to S2 and reads the next character “K ₂₂ ”. Next, in S3, it is checked whether there is any unprocessed character. When all processing is completed, the process proceeds to S16, where the codeword c
ode (ω) is output and the process ends. At this time, since there are unprocessed characters, the process proceeds to the dictionary search steps shown in S5 to S9. In the dictionary search step, first, the value of the initial character string i = 1 at that time is set in the address ω _ij in S5, and the j counter is set to j = 0. As a result, the address ω _ij = ω ₁₀ of the first memory is generated. Then since reading the contents of the address omega ₁₀ First memory 100 address ω _ij = ω ₂₁ is obtained in S6, and sets the i counter i = 2. Then, the process proceeds to S7, where it is checked whether i = 0. At this time, since i = 2, the process proceeds to S8, and the first memory of S6 is determined.
The character “K ₂₁ ” is read out with reference to the extended memory 300 at the address ω ₂₁ obtained from 100, and the input character “K ₂₂ ” obtained in S2 is read.
Is determined. In this case, since they do not match, the process proceeds to S9, and the value i = 2 of the i counter at this time is set to j.
Set in the counter, j = 2, and next memory
Address ω _ij = ω stored at 200 addresses ω _21
22 is set as i = 2 in the i counter. Thus, a new address ω _ij = ω ₂₂ is created. Subsequently, the process returns to S7, where i = 0 is checked.
Since it is 2, the flow advances to S8 again to read out the registered character “K ₂₂ ” in the extended memory 300 at the address ω ₂₂ and input the character “K ₂₂ ”.
Is determined. At this time, because they match
Returns to the S2, reads the next character "K _32". Similarly, S
By repeating the processing of 5～S9, dictionary search is performed in the order indicated by arrows in solid line in FIG. 14, the search processing already up registered letter "K _41" is performed. When the search for the registered character “K ₄₁ ” is completed and a mismatch is determined for the last input character “K ₄₂ ” in S 8, i = 2 in S 9.
Together with the set, the next memory 20 of the address ω ₄₁
Since 0 is 0, i = 0 is set. Therefore, when returning to S7, it is determined that i = 0, the process exits the dictionary search step and proceeds to S10, where the character string "K ₁₀ , K
_22, the K ₃₂ "address omega ₃₂ illustrating the output as a code word code (omega), the process proceeds to the dictionary registration step of S11～14. In the dictionary registration step, first, in S11, the currently registered character string n is set to n = i, that is, n = 4, and n is further incremented by one. And extended memory 3 the letter "K _42"
Register at address ω _ij = ω ₄₂ of 00. Next, in S12, it is checked whether or not j = 0.
Proceeds from it is a 2 to S14, write the address ω ₄₂ that registered the letter "K _42" to address ω ₄₁ of the next memory 200. On the other hand, if j = 0 in S12, i.e., if the state has shifted to the register of the first memory 100, the address omega ₁₁ of first memory 100 in FIG. 14, omega _22, as shown in omega _32, extend The character registration address of the memory 300 is stored. The registration of the character "K _42" in the character registration step, the next memory 200 and the extended memory 300 in FIG. 14
The address omega ₄₁ shown is partitioned by a broken line in the lower part becomes the registration state of the omega _42, so that the address omega ₄₂ of the new character "K _42" in the tree structure shown in FIG. 15 has been added. still,
In Figure 14, for convenience of explanation about the address omega _41, it is shown in an overlapping search and registration. When S11~S14 dictionary registration step is finished, a new prefix string i the letter "K _42" registered in S15, i.e., set to the value of the i counter, the letter "K _42" back to S2 again As the tree top, the processing shifts to dictionary search for the character string that follows.

[Problems to be solved by the invention]

As described above, in the conventional LZW encoding, when encoding is performed by executing the processing flow shown in FIG. 7 by software, it takes a lot of time for the dictionary search process.
Using the external hash method, the speed of dictionary search is increased by the processing flow of FIG. However, in the dictionary search using the external hash method, the reading of the candidate characters, the matching of the candidate characters with the input characters, and the determination of the match / mismatch are sequentially performed, so that the dictionary search time is shorter than the entire time. It accounts for about 80%, and higher speeds are needed. In addition, since a list structure using an external hash method is used for reading candidate characters, there is not much relation between the storage address of the current candidate character and the storage address of the next candidate character, and reading is performed at any time. However, it was not possible to advance the address, and it was not possible to make the most of the performance of the elements constituting the dictionary memory. For example, when a DRAM is used as a dictionary memory, there is no continuity in the address.
s) fixed to a row address ((Colum Adress) fast reading of the page mode or the like in which only the change is difficult. For example, in the case of FIG. 14, address omega ₃₂ of the next memory 200, the omega ₃₃ address Since there is no continuity of
As shown in the figure, a normal read mode is used in which a column address and a row address are individually specified each time, and there is a problem that the speed cannot be increased. The present invention has been made in view of such a conventional problem, and a dictionary search method of a data compression apparatus capable of shortening a dictionary search time by enabling high-speed reading of a dictionary memory using a list structure of an external hash method. The purpose is to provide.

[Means for Solving the Problems]

FIG. 1 is a diagram illustrating the principle of the present invention. First, according to the present invention, encoded data is divided into different sub-sequences, different reference numbers are added to the respective sub-sequences and registered in a dictionary, and input data is stored in the sub-sequences in the dictionary. Data to be encoded by specifying the reference number of the subsequence that matches the maximum length
Data compression apparatus, for example, a data compression apparatus that performs LZW encoding. According to the present invention, as a dictionary search method of such a data compression device, a next memory 200 having a first memory 100 and an extended memory 300 according to a list structure of an external hash method is provided, and an external hash address based on input data is provided. The link address is partially stored in the next memory 200.
And a dictionary search means 16 for continuously generating the address of the next memory 200 based on the input data and searching for candidate data of the extension memory 300 that matches the input data. It is characterized by. Here, the dictionary search means 16 includes a pipeline control means 26 which performs a parallel check of input data and candidate data, the presence or absence of candidate data, and the reading of the next candidate data in parallel. In addition, a high-speed page mode is used as an access mode of the dictionary memory 20.

[Action]

According to the dictionary search method of the data compression device according to the present invention having such a configuration, in the LZW encoding dictionary search, the index addresses of the dictionary memory having the list structure based on the external hash method are configured by continuous addresses. If one hash address is determined, the next address can be predicted. Therefore, the search access of the candidate character can be further speeded up, the encoding can be performed by reading the dictionary memory at high speed, and the encoding processing time can be shortened.

【Example】

FIG. 2 is a configuration diagram showing an embodiment of a data compression apparatus (encoding apparatus) provided with the dictionary search method of the present invention shown in FIG. In FIG. 2, original data 10 to be processed is a DMA (D
irect Memory Access) input via the control circuit 12. The MPU 14 as a control means receives the input raw data 10
Dictionary search circuit for one character and the reference number of the previous character string
After setting in the 16-character reading circuit 18, the dictionary search circuit 16 is activated. Thereafter, the dictionary search circuit 16 reads the candidate character of the character string extended by one character from the dictionary memory 20, performs a match check (collation) between the input character and the candidate character in the match check circuit 22, and a candidate character Is detected. The pipeline control circuit 26 reads 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 check circuit 22 and the detection of the presence or absence of the candidate character by the connection detection circuit 24. Thus, the pipeline control circuit
By performing the pipeline processing at 26, 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, and the multiple character reading circuit 18 reads a hash address and a candidate character registered in the continuous address of the dictionary memory 20. I will put it 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 by one character, and 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. 3 is a block diagram showing an embodiment of the dictionary search circuit 16 of the present invention shown in FIG. In FIG. 3, the address register 18-1, the register 18
-2 and the register 18-3 are the multi-character reading circuit of FIG.
18, the register 22-1 and the comparator 22-2 correspond to the coincidence checking circuit 22 in FIG. 2, the NOR circuit 24-1 corresponds to the connection detecting circuit 24 in FIG. 1 corresponds to the continuous address circuit 28 in FIG. Next, the dictionary search according to the embodiment of FIG. 3 will be described with reference to the explanatory diagram of the search procedure and the registration procedure of FIG. 4 and the explanatory diagram of the tree structure showing the registered state of the dictionary memory 20 of FIG.
In the following description, the memory address ω is represented as ω _ij by the upper address i and the lower address j. It is assumed that a character string “K ₁₀ , K ₂₂ , K ₃₂ , K ₄₂ ” included in the tree structure in FIG. First MPU14 the first character K ₁₀ of the first type character string
1 reference number omega ₁₀ characters of while set in the address register 18-1 to specify the higher address, and input 2
To set the th character K ₂₂ to the register 18-2. Next, the pipeline control circuit 26 is instructed to start the dictionary search circuit 16. The pipeline control circuit 26 first sets a counter 28-1 for generating a continuous address to 0, and then reads the dictionary memory 20. The contents of the counter 28-1 specify the least significant two bits (LSB) of the address of the dictionary memory 20. Therefore, the contents of the address register 18-1 ω _ij = ω
The first memory 100 shown in FIG. 4 is accessed by the upper address of the dictionary memory 20 according to ₁₀ and the address (ω ₁₀ +0) which specifies the lower address of the dictionary memory 20 by the content j = 0 of the counter 28-1. to set the ω ₂₁ read, in the address register 18-1 Te. Next, the next memory 20 of the dictionary memory 20 is determined by an address (ω ₂₁ +0) using the contents ω ₂₁ of the address register 18-1 as the upper address and the contents of the counter 28-1 as the lower address.
0 and the extended memory 300 are accessed, and the first candidate character is
It reads the K ₂₁ and link address omega ₂₂ of the second candidate character K _22. Read out the first candidate character K ₂₁ registers 18-
Is set to 2, link address omega ₂₂ of the second candidate character K ₂₂ are set in the register 18-3. And the register
Input character K ₂₂ and registers are set to 22-1 18-
Comparing the first candidate character K ₂₁ which is set to 2 units 22-
A comparison is made in step 2 to determine the match or mismatch. Since they do not match, a disagreement is determined,
But to read the next candidate character K _22, this time counter 28-1
One of the value of the increment to the next memory 200 of the address was changed only the lower address of the dictionary memory 20 (ω ₂₁
+1) generates and reads in the access of the next memory 200 the next candidate characters K ₂₂ to the register 18-2. The contents ω ₂₂ of the address register 18-1 that specify the higher address this time is as it is. Hereinafter, similarly, this operation is repeated.
Generating consecutive addresses blindly using 1 is
In order to increase the size of the dictionary memory 20, this embodiment considers generating four consecutive addresses. For example, if the character code is 8 bits, an address exceeding 9 bits is meaningless. Therefore, once every four searches, the concatenated address ω _ij obtained by accessing the first memory 100 is used instead of the continuous address of the next memory 200. That is, if the lower address “00,01,10,11” is generated four times in the counter 28-1 while the upper address is fixed, the register 18-3 is simultaneously stored in the register 18-3 at the same time as switching to the next continuous address “00”. 4
At the second access, the connection address of the first memory 100 stored in the registers 18-3 is stored in the address register 1
Set to 8-1. For example, the upper address ω of the next memory 200 in FIG.
Taking ₃₁ as an example, by incrementing the lower address by the counter 28-1, ω ₃₁ +0 (= ω ₃₁ ) ω ₃₁ +1 (= ω ₃₂ ) ω ₃₁ +2 (= ω ₃₃ ) ω ₃₁ +3 (= ω ₃₄ ) Is generated as a continuous address, and the fifth time, the concatenated address ω ₃₅ to the next continuous address stored in the next memory 200 is read out, and the generation of the continuous address is repeated again from the beginning as the upper address. When the comparison between the input character and the candidate character matches in the comparator 22-2 by such a dictionary search, the NOR circuit 24-
In step 1, it is checked whether the contents of the register 18-3 (the link address of the next memory 200) is all 0s.
Repeat the dictionary search until. If the register 18-3 is all 0s, it is detected that there are no more candidate characters to be searched. In this case, the MPU 14 and the pipeline control circuit 26 terminate the search processing of the dictionary search circuit 16, and output the address of the last candidate character that has been matched by the dictionary search up to that time as a codeword code (ω). For Figure 4, the next memory 200 in the input character "K _41"
Is zero, the dictionary search is terminated at this stage, and the address (ω ₄₁ +0) of the last candidate character “K ₄₁ ” that matches is output as the codeword code (ω). Subsequently, the MPU 14 registers the address (ω ₄₂ +1) of the last remaining input character “K ₄₂ ” in the extension memory 300 and the registration of the connection address ω _{42 in} the address (ω ₄₁ +0) of the next memory 200. after, to migrate to the new dictionary search for the input character "K _42" as a prefix string i. As described above, according to the present invention, since addresses can be continuously generated to search for candidate characters and concatenated addresses, only the row addresses are changed while the column addresses are fixed as shown in FIG. A high-speed page mode using continuous addresses can be used, and candidate characters and their concatenated addresses can be read at high speed, so that high-speed dictionary search can be realized.

【The invention's effect】

As described above, according to the present invention, in the dictionary search of LZW encoding, a linked list using the external hash method is configured by continuous addresses, so that if one address is determined, it is possible to precede the address by predicting the address. For example, by realizing a high-speed page mode when a DRAM is used, the performance of the memory element can be fully exhibited and the speed of dictionary search can be increased.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention; FIG. 2 is a diagram illustrating the configuration of an embodiment of the present invention; FIG. 3 is a diagram illustrating the configuration of an embodiment showing details of a dictionary search circuit according to the present invention; FIG. 5 is a diagram showing a tree structure showing dictionary registration contents of the present invention; FIG. 6 is a DRAM structure using the high-speed page mode of the present invention;
FIG. 7 is a flowchart of a conventional LZW encoding process; FIG. 8 is a flowchart of a conventional LZW decoding process; FIG. 9 is an explanatory diagram of LZW encoding; FIG. FIG. 11 is an explanatory diagram of LZW encoding; FIG. 12 is an explanatory diagram of a list structure of an external hash method; FIG. 13 is a flowchart of a conventional LZW encoding process using an external hash method; FIG. Explanatory diagram of LZW code search procedure and registration procedure of FIG. 13; FIG. 15 is a tree structure diagram showing dictionary registration contents of FIG. 14; FIG. 16 is a timing chart of DRAM read mode in which high-speed page mode cannot be used It is. In the figure, 10: raw data 12: DMA control circuit 14: MPU 16: dictionary search means (dictionary search circuit) 18: multiple character reading circuit 18-1: address revaster 18-2, 18-3: register 20: dictionary Memory 22: Match check circuit 22-1: Register 22-2: Comparator 24: Link detection circuit 24-1: NOR circuit 26: Pipeline control circuit 28: Continuous address circuit 28-1: Counter 30: I / O circuit 100 : First memory 200: Next memory 300: Extended memory

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Yasuhiko Nakano 1015 Uedanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Inside Fujitsu Limited (56) References JP-A-59-231683 (JP, A) JP-A-60-116228 (JP, a) JP Akira 64-7230 (JP, a) JP Akira 61-13340 (JP, a) JP Akira 58-155589 (JP, a) (58 ) investigated the field (Int.Cl. ⁷ , DB name) H03M 7/42

Claims (3)

(57) [Claims] An encoded data is divided into different sub-sequences, a different reference number is added to each sub-sequence and registered in a dictionary, and input data is stored in a sub-sequence in the dictionary. A data compression apparatus for coding by designating a reference number of a subsequence that matches the maximum length has a first memory ((100)) and an extended memory (300) according to a list structure of an external hash method. A dictionary memory (20) comprising a next memory (200), wherein a concatenated address of an external hash address based on input data is partially constituted by a continuous address of the next memory (200)
And the next memory (200) based on the input data
And dictionary search means (16) for searching for candidate data of the extension memory (300) which continuously generates the above address and matches the input data. 2. A dictionary search method for a data compression apparatus according to claim 1, wherein said dictionary search means (16) checks whether the input data matches the candidate data, determines whether or not the candidate data exists, and reads the next candidate data. A dictionary search method for a data compression device, comprising: a pipeline control means (26) for performing the above in parallel. 3. A dictionary search method for a data compression device according to claim 1, wherein a high-speed page mode is used as an access mode of said dictionary memory (20).