JPH05252049A - Data compression processing system and data decoding processing system - Google Patents

Abstract

PURPOSE:To realize high reliability for data compression processing system and data decoding processing system employing a dynamic dictionary type Zib-Lembel code. CONSTITUTION:Dictionaries 11, 21 managing cross reference between character strings and reference numbers are formed according to the external hash method using a reference number of a character string as a hash address. In this case, the system is provided with addition means 14, 24 adding additional information indicating that a character string is already registered to a reference number with cross reference when the character string is registered in the dictionary, deciding means 15, 25 deciding the correctness of a retrieved character string/ reference number from the dictionaries 11, 21 according to the additional information added by the addition means 14, 24 at the retrieval of the dictionary.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression processing method using a dynamic dictionary type Jib-Lempel code and a data decompression processing method thereof. In particular, high reliability is ensured when a dictionary is constructed according to an external hash method. The present invention relates to a data compression processing method and a data decompression processing method that can be realized.

In recent years, various kinds of data such as character codes, vector information, and images have been handled by computers, and the amount of data handled has been increasing rapidly. When dealing with such a large amount of data, it is necessary to omit the redundant part of the data and compress the data amount in order to reduce the storage capacity and realize high-speed transmission.

Various methods of data compression have been proposed, but as a method of realizing data compression without knowing the statistical properties of the data to be compressed,
Universal codes have been proposed. As a typical method of this universal code, Ziv-Lempe (Ziv-Lempe
There is a l) code, and in this Jib-Lempel code, two algorithms, a slide dictionary type and a dynamic dictionary type, are roughly proposed. As an improvement of this slide dictionary type algorithm, LZSS code is proposed, and as an improvement of the dynamic dictionary type algorithm, LZSS code is proposed.
The W code has been proposed. In order to make the data compression / decompression process using such a universal code practical, it is necessary to construct a configuration that can appropriately deal with an error.

[0004]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a data compression processing method using a dynamic dictionary type Jib-Lempel code and an improvement in a data decompression processing method using a dynamic dictionary type Jib-Lempel code. Now, here, a conventional technique for the LZW code will be described. In the following description, one word of data will be referred to as a character, and one obtained by connecting arbitrary words of data will be referred to as a character string, following the name used in information theory.

FIG. 6 shows a processing flow of data encoding processing by the LZW code, and FIG. 7 shows a processing flow of data restoration processing by the LZW code. First, the data encoding process using the LZW code will be described with reference to FIG. In the data encoding process using the LZW code, the input data is decomposed into character strings different from each other, and the character strings are managed in a dictionary together with a reference number (dictionary number). By replacing the longest matching character string registered in the dictionary with the reference number string, a long character string is represented by a short reference number to perform encoding.

Here, as for the method of registering a character string in the dictionary, as shown in FIG. 8, for a newly registered character string other than a single character, the reference number registered up to that point and the subsequent one character are used. The composition is expressed by the combination of and registered.

That is, when performing the data encoding process by the LZW code, as shown in the process flow of FIG. 6, first, in step 1, there is a possibility that it appears in the character string to be encoded. All certain single characters are registered in the dictionary, and the maximum number N of reference numbers is set to the number of single character types. Next, in step 2, the first character K of the input data to be encoded is input, and the reference number of the character K is set as the initial character string ω. Then, in step 3, the next character K is read from the input data to be encoded, and in step 4, the character string ωK that is a combination of the initial character string ω representing the reference number and the read character K is It is determined whether it is registered in the dictionary.

When it is determined in step 4 that the character string ωK is registered in the dictionary, the process proceeds to step 5, where the reference number of this character string ωK is set as a new initial character string ω, and the subsequent step In step 6, it is determined whether or not the processing has been completed for all the characters of the input data to be encoded. If it is determined that the processing has not been completed, the process returns to step 3 to search for the longest matching character string. When it is determined that the process has ended, the process proceeds to step 8, the reference number of the initial character string ω is output, and the process ends.

On the other hand, when it is determined in step 4 that the character string ωK is not registered in the dictionary, the process proceeds to step 7 in response to the character string indicated by the initial character string ω being the longest matching character string. Then, the reference number of the initial character string ω is output, and this character string ωK is added to the reference number and registered in the dictionary. Then, the reference number of one character K following the initial character string ω up to that point is set as a new initial character string ω, and the maximum value of the reference number is incremented by 1 before proceeding to step 6. Then, the search for the next longest matching character string is started.

According to FIG. 9, LZ according to this processing flow
The W code generation will be specifically described. According to the process of step 1, the character a is registered in the dictionary with the reference number 1, the character b is the reference number 2, and the character c is registered with the reference number 3. Next, according to the process of step 2, the first character a of the input data is read and its reference number 1 is set as the initial character string ω. Then, the second character b of the input data is read according to the processing of step 3, and step 4
It is determined that the character string 1b (= ab) is not registered in the dictionary according to the processing of step 1, and the reference number 1 of the initial character string ω is output according to the processing of step 7, and
The character string 1b is registered in the dictionary together with the reference number 4, and the reference number 2 of this character b is set as a new initial character string ω.

Then, the third character a of the input data is read out according to the process of step 3, and it is determined according to the process of step 4 that the character string 2a (= ba) is not registered in the dictionary. Then, according to the process of step 7, the reference number 2 of the initial character string ω is output, and
The character string 2a is registered in the dictionary together with the reference number 5, and the reference number 1 of this character a is set as a new initial character string ω.

Subsequently, the fourth character b of the input data is read according to the processing of step 3, and it is determined that the character string 1b is registered in the dictionary according to the processing of step 4, and step 5 is executed. In accordance with the process of, the reference number 4 of this character string 1b is set as a new initial character string ω. Then, the fifth character c of the input data is read according to the process of step 3, and it is determined that the character string 4c (= abc) is not registered in the dictionary according to the process of step 4, According to the process of 7, the reference number 4 of the initial character string ω is output and the character string 4
c is registered in the dictionary together with the reference number 6, and the reference number 3 of this character c is set as a new initial character string ω. By repeating the same processing thereafter, encoding as shown in FIG. 9 is executed.

Next, referring to FIG. 7, a data restoration process using the LZW code will be described. The data restoration processing by the LZW code has a configuration of performing the data restoration by executing the inverse conversion processing of the data encoding described in FIG.

That is, when executing the data restoration processing by the LZW code, as shown in the processing flow of FIG. 7, first, in step 1, all the characters that may appear in the restored character string are displayed. Is registered in the dictionary, and the maximum number N of reference numbers is set to the number of single character types. Next, in step 2, the first code (CODE) of the input data (reference number string) to be restored is read, and OLDcod
Set it as e and search the dictionary for this CODE
The character K pointed to by is searched for and output. Here, the output character K is set to char for later exception processing.

Then, in step 3, the next code (CODE) is read from the input data to be restored and set as NEWcode. Then, in step 4, it is checked whether the CODE read in step 3 is registered in the dictionary. When it is determined in step 4 that CODE is registered in the dictionary, the process proceeds to step 5, the character string ωK pointed to by this CODE is read from the dictionary, and in step 6 that follows,
The character K of this character string ωK is stored in the stack, the reference number ω of this character string ωK is set as a new CODE, and the process returns to step 5.

When it is determined that CODE reaches the state of indicating one character by recursively executing the processes of steps 5 and 6, the process proceeds to step 7, and the character string stacked in step 6 is deleted. LILO (Last In Fast O
ut) format and pops up and outputs, and OLDcod
A reference number is added to the character string ωK, which is a combination of the previously used reference number ω set in e and the first character of the character string restored this time, and is registered in the dictionary. Then, the first character of the restored character string is set to char for later exception processing, the CODE of NEWcode is set as OLDcode, and the maximum value of the reference number is incremented by one.

When the process of step 7 is completed, the process proceeds to step 8 to judge whether the process has been completed for all codes of the input data to be restored, and when it is judged that the process is not completed, By returning to step 3, the next code restoration processing is executed, and when it is judged that the processing is completed, the processing is completed.

When it is determined in step 4 that the CODE read in step 3 is not registered in the dictionary, the process proceeds to step 9 and the following exception processing is executed. Here, such a state occurs when the immediately preceding reference number is referred to in encoding. From this, in step 9, char is output, OLDcode is set as CODE, a combination of OLDcode and char is set as NEWcode, and the process proceeds to step 5.

According to FIG. 10, L according to this processing flow
The decoding process of the ZW code will be specifically described. According to the process of step 1, the character a is the reference number 1 and the character b is
Is registered in the dictionary together with the reference number 2 and the character c together with the reference number 3. Next, according to the process of step 2, the leading code 1 of the input data is read and the character a pointed to by the code 1 is read.
Is output. Then, the second code 2 of the input data is read according to the processing of step 3, and step 5
Through the process of step 7, the character b pointed to by the code 2 is output, and the character string 1b, which is a combination of the previously processed code 1 and the first character b of the character string restored this time, is also stored in the dictionary together with the reference number 4. Be registered with.

Then, the third code 4 of the input code string is read out according to the processing of step 3, and the character string ab pointed to by the code 4 is output according to the processing of steps 5 to 7, and A character string 2a, which is a combination of the code 2 processed last time and the first character a of the character string restored this time, is registered in the dictionary together with the reference number 5. By repeating the same processing thereafter, the encoding as shown in FIG. 10 is executed, but when the sixth code 8 of the input data is read, the code 8 is the dictionary at the time of this restoration. Not registered in. From this, the exception process is executed according to the process of step 9, and the character string 5b obtained by adding the leading character b of the previously restored character string ba to the previously processed code 5 is obtained, and the character string 5b is restored to obtain the code 8 The character string bab pointed to by is obtained and output, and the character string 5b obtained by adding the first character b of the character string restored this time to the code 5 processed last time is registered in the dictionary together with the reference number 8. ..

However, if LZW encoding / LZW decoding / is executed according to the procedure shown in the processing flow of FIG. 6 / FIG. 7, in the worst case, the entire dictionary must be searched every time one character string is searched. However, there is a problem that the dictionary search takes time. Therefore, conventionally, a configuration has been adopted in which a dictionary search is executed using the external hash method.

When considering a set S composed of a plurality of character strings, a high-speed dictionary search can be realized if the storage position of the character string x of the set S can be directly calculated from the character string x itself. The hash method realizes this.
If addresses from 0 to (m-1) are assigned to the memory location (hash table), the function h: S → [0, 1, 2, ...
], And the address of the character string x of the set S is obtained by h (x). The function h is called a hash function, and the value h (x) is called a hash address of x. The hash method is usually used when the size of S is much larger than m. Therefore, no matter how h is selected, there is a possibility that h (x1) = h (x2) for the character strings x1 and x2 having different S. This is called a collision, and the external hash method is used as one of the countermeasures against this collision.

In this external hash method, as shown in FIG. 11, a linked list is prepared for each hash address i indicated by an index, and a hash address h (x) =
The character string x of i corresponds to the character string that is stored in order from the head of the linked list. Each linked list having the same hash address h (x) is called a bucket.

FIG. 12 shows a processing flow of LZW encoding using a list structure of the external hash method for dictionary retrieval, and FIG. 13 shows the dictionary data of FIG. 8 as an example and a conventional dictionary used at this time. The structure is shown in FIG.
The conventional dictionary memory corresponding to the dictionary structure of FIG.

As shown in FIG. 13, each node constituting the dictionary has an address (register position of the dictionary) shown at the upper left of the node.
And the registered symbol of the extended memory shown in the node (the character registered at that address), the address of the next first memory shown at the lower left of the node (the address of the next character following the depth of the tree structure), It manages the address of the next memory shown at the lower right of the node (the address of the next character in the horizontal direction of the tree structure). According to this dictionary structure, the dictionary data AB of FIG. 8 is managed as having a reference number 4 according to the linked list of A of address 1 → B of address 4, and AA of the dictionary data of FIG. It is managed as having the reference number 10 according to the linked list of A → B of address 4 → A of address 10.

The processing flow shown in FIG. 12 is such that, when a character string to be encoded is given, the encoding processing is executed while executing the dictionary memory search processing and registration processing of FIG. The processing of each step is as follows. Here, steps 4 to 8 define the dictionary search process, and steps 10 to 14 define the dictionary registration process.

[Step 1] A single character is registered in a dictionary together with an address. Number of registered characters → number of characters currently registered in dictionary nn → NMIN Input first character K → Initial character string ω Dictionary search array first [1, NMAX], next [1, NMAX], ext [1, N
MAX] is initialized to 0.

[Step 2] The next character input K is read. [Step 3] It is determined whether or not there is a letter K.

[Step 4] The letter K is determined by the judgment in step 3.
When it is determined that there is, the following processing is performed. here,
i is the address value of the dictionary memory specified by the stored value of the first memory, and j is the address value of the dictionary memory specified by the stored value of the next memory. i → ω; 0 → j; [Step 5] ext [i] → symbol; first [i] → i; [Step 6] It is determined whether or not i = 0.

[Step 7] i = 0 in the judgment of Step 6
When it is judged that it is not, the following processing is performed. symbol ＝
It is determined whether or not K.

[Step 8] Symbol is judged by the judgment in Step 7.
When it is determined that not = K, the following processing is performed. ext [i] → symbol; next [i] → i; [Step 9] The following processing is performed when it is determined in step 6 that i = 0. Output code [ω).

[Step 10] n → i; n + 1 → n; [Step 11] It is determined whether or not j = 0.

[Step 12] j is judged in step 11
When it is determined that = 0, the following processing is performed. i → first [ω]; K → ext [ω]; [Step 13] When it is determined in step 11 that j = 0 is not satisfied, the following processing is performed. i → next [j]; K → ext [j]; [Step 14] K → i; [Step 15] When it is determined in step 3 that there is no character K, the following processing is performed. Output code [ω).

FIG. 15 shows a processing flow of LZW decoding using a list structure of an external hash method for dictionary retrieval, and FIG. 16 shows the dictionary data of FIG. 8 as an example and a conventional dictionary used at this time. The structure is shown in FIG.
The conventional dictionary memory corresponding to the dictionary structure of FIG.

As shown in FIG. 16, each node constituting the dictionary has an address (registered position in the dictionary) shown at the upper left of the node.
And managing the expanded memory registration symbol (the character registered at that address) inside the node and the before memory address (the address of the next character following the root of the tree structure) at the bottom left of the node Is. According to this dictionary structure, it is managed that the reference number 6 sent from the compression side is the character string ABC according to the linked list of address 6 → address 4 → address 1.

The process flow shown in FIG. 15 is for executing the restoration process while performing the search process and the registration process of the dictionary memory of FIG. 14 when the code word string to be decoded is given. The processing of each step is as follows.

[Step 1] A single character is registered in a dictionary together with an address. Number of registered characters → number of characters currently registered in dictionary nn → NMIN Initialize the dictionary search array before [1, NMAX], ext [1, NMAX] to 0.

[Step 2] The first code is read. CODE → OLD code CODE → K Outputs character K. K → FIN char [Step 3] Read the next code input code. CODE → IN code [Step 4] It is determined whether or not there is a new code.

[Step 5] When it is determined in step 4 that there is a new code, the following processing is performed. CODE
It is determined whether or not is registered.

[Step 6] When it is determined in step 5 that the CODE is not registered, the following processing is performed. CODE ≧ NMIN; [Step 7] When CODE ≧ NMIN is determined in the determination in step 6, the following processing is performed. ext [CODE] → Stack; before [CODE] → CODE; [Step 8] When it is determined in step 6 that CODE ≧ NMIN is not satisfied, the following processing is performed. CODE → K Outputs character K. K → FINchar DO while (until stack is empty) [Stack top is output; POP stack;] OLDcode → before [n] K → ext [n] n + 1 → n INcode → OLDcode [Step 9] FINchar → Stack; OLDcode → CODE;

[0041]

As described above, the conventional L
In the ZW code, while a dictionary is created in the area secured in the storage device, the encoding / decoding process is performed using the dictionary. As described above, the contents of the dictionary are composed of a linked list, and when the dictionary is searched, the linked list is followed to encode a code word and restore it to a character.

By the way, such a dictionary search process is established under the precondition that the data on the storage device is not destroyed by an external factor or the like. However, when it is configured as a compression / decompression device, problems such as damage to the storage device and data corruption occur. In this case, in the conventional technique, even if erroneous data is read from the dictionary and used for the encoding process or the restoration process, it is not possible to determine whether the data is correct data or erroneous data. was there.

The present invention has been made in view of the above circumstances, and realizes high reliability in constructing a dictionary according to an external hash method when executing data compression / decompression processing by the LZW code. It is an object of the present invention to provide a new data compression processing method that can be performed and a new data decompression processing method that can achieve high reliability.

[0044]

FIG. 1 shows the principle configuration of a data compression apparatus for implementing the data compression processing of the present invention, and FIG.
FIG. 1 illustrates the principle configuration of a data restoration device that implements the data restoration processing of the present invention.

In FIG. 1, 10 is a data compression device for implementing the data compression processing of the present invention, and 11 is a dictionary included in the data compression device 10. This data compression device 10
The input data is decomposed into different character strings, and this character string is managed in the dictionary 11 together with the reference number, and the data being input is used by using the reference number of the longest matching character string registered in the dictionary. It adopts a configuration for encoding.
The dictionary 11 manages the correspondence between the character string and the reference number according to the external hash method in which the reference number of the character string is used as the hash address.

That is, the dictionary 11 manages character information at the current reference number according to the extended memory, and manages reference number information (address information) of the next character following the depth memory of the tree structure according to the first memory. According to the next memory, the reference number information of the next character continuing in the horizontal direction of the tree structure is managed. For example, if the dictionary data shown in FIG. 1 is taken as an example, the character string AB
Is managed as having a reference number 4 according to the depth direction of the tree structure of A of address 1 → B of address 4,
In addition, the character string AA is A at address 1 → B at address 4.
→ It is managed as having the reference number 10 according to the depth / horizontal direction of the tree structure of A at address 10.

Reference numeral 12 is a dictionary search means provided in the data compression apparatus 10 for searching whether or not a character string to be decomposed from input data is registered in the dictionary 11, and 13 is dictionary registration provided in the data compression apparatus 10. It is a means for registering in the dictionary 11 a character string of a combination of the longest matching character string registered in the dictionary, which is decomposed from the input data, and one character following it.

Reference numeral 14 denotes additional means developed in the dictionary registration means 13, for example, when the dictionary registration means 13 registers a character string in the dictionary 11, additional information indicating that the character string has been registered. Is added in association with the reference number. The adding means 14 may add additional information, such as parity information, which can specify whether the dictionary data is correct or incorrect, instead of the registered additional information.

Denoted by 15 is a determination means developed in the dictionary search means 12, for example, when the dictionary search means 12 searches the dictionary 11, a character string searched from the dictionary 11 according to the additional information added by the adding means 14. / Determines whether the reference number is correct.

In FIG. 2, reference numeral 20 is a data restoration device that implements the data restoration processing of the present invention, and reference numeral 21 is a dictionary included in the data restoration device 20. This data restoration device 20
The restored data is decomposed into different character strings, this character string is managed in a dictionary together with a reference number, and the corresponding character string in the dictionary 21 is specified from the reference number specified by the codeword. Then, the codeword being input is replaced with the specified character string to restore the codeword. The dictionary 21 hashes the reference numbers of the character strings.
The correspondence between the character string and the reference number is managed according to the external hash method that uses the address.

That is, the dictionary 21 manages the character information at the current reference number in accordance with the expanded memory, and manages the reference number information of the next character following the root of the tree structure in accordance with the before memory. .. For example, in FIG.
If the dictionary data shown in is taken as an example, the reference number 6 sent from the compression side is address 6 → address 4
→ The character string ABC is managed according to the linked list of address 1.

Reference numeral 22 is a dictionary retrieval means provided in the data decompression device 20, which retrieves a character string pointed to by a code word sent from the compression side from the dictionary 21, and 23 is a dictionary registration means provided in the data decompression device 20. That is, the character string of the combination of the longest matching character string registered in the dictionary generated from the restored data and the following one character is registered in the dictionary 21.

Reference numeral 24 denotes additional means developed in the dictionary registration means 23. When the dictionary registration means 23 registers a character string in the dictionary 21, additional information indicating that the character string is already registered is displayed. Is added in association with the reference number. The adding unit 24 may add additional information, such as parity information, which can specify whether the dictionary data is correct or incorrect, instead of the registered additional information.

Reference numeral 25 is a judgment means which is developed in the dictionary search means 22, for example, and when the dictionary search means 22 searches the dictionary 21, a character string searched from the dictionary 21 according to the additional information added by the adding means 24. / Determines whether the reference number is correct.

[0055]

In the data compression apparatus 10 of the present invention, the addition means 1
Reference numeral 4 indicates that, when the dictionary registration means 13 newly registers a character string not registered in the dictionary 11 with a reference number, the character string to be registered is already registered. The additional information (which may be the additional information that can specify whether the registered dictionary data is correct or incorrect) is registered in association with the reference number.

On the other hand, the judging means 15 is the dictionary searching means 12
Is searching for the longest matching character string by searching the dictionary data of the dictionary 11, and the character string / reference number searched by the dictionary searching means 12 is sure according to the additional information added by the adding means 14. Confirm that the registered dictionary data has been registered (if the additional information can identify the correctness of the registered dictionary data, confirm that the search dictionary data is correct), / When entering a reference number search (wrong search search dictionary data search if additional information allows identification of correct / wrong registered dictionary data), the dictionary search means 12 is notified of the error. Go

As described above, the data compression apparatus 10 of the present invention
Then, the input data is decomposed into character strings different from each other, and this character string is managed in the dictionary 11 together with the reference number. The data being input is referred to as the reference number of the longest matching character string registered in the dictionary. Since it is possible to prevent erroneous encoding processing due to erroneous dictionary data at the time of adopting a configuration for encoding by using it, it becomes possible to perform data compression processing that can realize high reliability. ..

In the data restoration device 20 of the present invention, when the dictionary registration means 23 newly adds a character string not registered in the dictionary 21 to the dictionary 21 by adding a reference number, the addition means 24 The additional information indicating that the character string to be registered is already registered (may be additional information that can specify the correctness of the registered dictionary data) is registered in association with the reference number.

On the other hand, the judging means 25 is the dictionary searching means 22.
When the character string is restored by searching the dictionary data of the dictionary 11, the character string / reference number searched by the dictionary searching means 22 is surely registered according to the additional information added by the adding means 24. Check that the additional information is correct (if the additional information can identify the correctness of the registered dictionary data, confirm that the search dictionary data is correct), and check the character string / reference other than registered When the search for a number (when the additional information makes it possible to specify whether the registered dictionary data is correct or incorrect) is searched, an error is notified to the dictionary search means 22.

Thus, the data restoration device 20 of the present invention
Then, the restored data is decomposed into different character strings, the character string is managed in a dictionary together with a reference number, and the corresponding character string in the dictionary 21 is specified from the reference number specified by the codeword. It is possible to prevent the execution of erroneous restoration processing due to erroneous dictionary data when the configuration is adopted in which the codeword being input is replaced with this specified character string. Therefore, it becomes possible to execute a data restoration process that can realize high reliability.

[0061]

EXAMPLES The present invention will be described in detail below with reference to examples. FIG. 3 illustrates a configuration example of the data compression / decompression device of the present invention.

In FIG. 3, original data (character data or codeword data) to be processed is input via the DMA control circuit 100. MPU as a control procedure
Reference numeral 200 is a reading circuit 402 of the dictionary search circuit 400 for inputting the input original data by using one character and the reference number of the character string up to now.
Then, the dictionary search circuit 400 is activated.

Thereafter, the dictionary search circuit 400 reads the candidate character of the character string extended by one character from the dictionary memory 300, the match check circuit 405 checks the match between the input character and the candidate character, and the connection detection circuit 404 selects the candidate. The presence / absence of characters is detected. Here, it is confirmed by referring to the in-use flag in the dictionary memory 300 whether or not the data of the matching search target is already registered data. Continue processing if in use,
If it is not used, the processing is interrupted because the search is invalid.

The pipeline control circuit 403 executes the next candidate character in the dictionary memory 300 in parallel with the matching between the input character and the candidate character by the match checking circuit 405 and the presence or absence of the input character and the candidate character by the concatenation detection circuit 404. To read. In this way, by performing pipeline processing by the pipeline control circuit 403, it is possible to perform search and collation processing for each plurality of candidate characters within the cycle time of the dictionary memory 300.

Further, the dictionary search circuit 400 is provided with a continuous address circuit 401 for generating continuous addresses,
Processing is performed so that the reading circuit 402 reads out the hash address and the candidate character registered in the continuous address of the dictionary memory 300.

In encoding the LZW code, the dictionary memory 30 is used.
The character string with the maximum length in 0 is found. Therefore, it is understood that the input character is added and the character string is sequentially extended one character at a time, and when there are no candidate characters, the character string has the maximum matching length. At this time, up to the longest match character string is represented by reference numbers using the address ω, and the MPU 20
0 outputs the reference number ω as a code compressed from the input / output port 500 to the outside. Also, a set of the reference number ω and the last input character is registered in the dictionary memory 300. At this time, a flag indicating the use of the target address is written as 1 in the dictionary memory 300.

On the other hand, in order to restore the LZW code, the dictionary memory 300 is accessed by the input code to restore the characters one by one while tracing the linked list forward. Outputs a character string as character string data. At this time, similar to the compression process, by checking whether or not the code word (reference number) to be decompressed is already registered in the dictionary memory 300, an error in the code word due to storage / transmission is detected. I will go.

FIG. 4 shows the data compression device 1 described with reference to FIG.
0 illustrates an example of a processing flow executed by 0. In the figure,
Steps that perform the same processing as the conventional processing flow shown in FIG. 12 are represented by the same step numbers.

As can be seen by comparing the processing flow of FIG. 4 with the processing flow of FIG. 12, the data compression apparatus 1 of the present invention is
0 means that between step 1 and step 2, step 1-1
As a result, the process of initializing FLG [1, NMAX] to 0 is performed, and between step 6 and step 7, as step 6-1,
When it is determined whether FLG [i] = 1, and when it is determined that FLG [i] = 1, the process proceeds to step 7 and FLG [i]
When it is judged that = 1 is not satisfied, the processing is judged to be an error and the processing is interrupted.
In step 12-1, the process of 1 → FLG [ω] will be performed between the first and second steps.

That is, when registering dictionary data,
In step 12-1, the usage flag indicating that the dictionary data has been registered is set to ON, and when searching the dictionary data, in step 6-1 it is confirmed that the usage flag is ON. While performing the search process, the process is performed.

FIG. 5 shows the data restoration device 2 described with reference to FIG.
0 illustrates an example of a processing flow executed by 0. In the figure,
Steps that execute the same processing as the conventional processing flow shown in FIG. 15 are represented by the same step numbers.

As can be seen by comparing the processing flow of FIG. 5 with the processing flow of FIG. 15, the data restoration device 2 of the present invention can be used.
0 means that between step 1 and step 2, step 1-1
As a result, the process of initializing FLG [1, NMAX] to 0 is performed, and between step 6 and step 7, as step 6-1,
FLG [CODE] = Judge whether FLG [CODE] = 1
When it is determined that the value is 1, the process proceeds to step 7. When it is determined that FLG [CODE] = 1 is not satisfied, the process is determined to be an error and the process is interrupted. As 1, the processing of 1 → FLG [n] is performed.

That is, when registering dictionary data,
In step 8-1, the usage flag indicating that the dictionary data has been registered is set to ON, and when searching the dictionary data, it is confirmed in step 6-1 that the usage flag is ON. While performing the search process, the process is performed.

Although the illustrated embodiment has been described, the present invention is not limited to this. For example, in the example,
The present invention has been disclosed by adding a use flag indicating that the character string is already registered when registering the character string in the dictionary. However, instead of the use flag, registration such as parity information is performed. It is also possible to add additional information that enables correctness of the dictionary data. Further, in the embodiment, L
Although the present invention has been disclosed according to the ZW code, it can be applied as it is to the one according to the encoding / restoring processing of the dynamic dictionary type Dibulenpel code.

[0075]

As described above, according to the present invention,
The input data is decomposed into different character strings, and this character string is managed in a dictionary together with a reference number.The data being input is encoded using the reference number of the longest matching character string registered in the dictionary. Since it is possible to prevent erroneous encoding processing due to erroneous dictionary data at the time of adopting a configuration for encoding, it is possible to perform data compression processing that can realize high reliability.

Further, the restored data is decomposed into character strings different from each other, this character string is managed in the dictionary together with the reference number, and the corresponding character string in the dictionary is specified from the reference number designated by the code word. To prevent the execution of erroneous restoration processing due to erroneous dictionary data when there is a configuration in which the codeword being input is replaced with this specified character string for restoration. Therefore, it becomes possible to execute a data restoration process that can realize high reliability.

[Brief description of drawings]

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

FIG. 2 is a principle configuration diagram of the present invention.

FIG. 3 is a configuration diagram of an embodiment of the present invention.

FIG. 4 is an explanatory diagram of a data encoding process using an LZW code according to the present invention.

FIG. 5 is an explanatory diagram of a data restoration process using an LZW code according to the present invention.

FIG. 6 is an explanatory diagram of a data encoding process using an LZW code.

FIG. 7 is an explanatory diagram of a data restoration process using an LZW code.

FIG. 8 is an explanatory diagram of a dictionary registration configuration in LZW code.

FIG. 9 is an explanatory diagram of a data encoding process using an LZW code.

FIG. 10 is an explanatory diagram of a data restoration process using an LZW code.

FIG. 11 is an explanatory diagram of a list structure of an external hash method.

FIG. 12 is an explanatory diagram of data encoding processing by a conventional LZW code using the external hash method.

FIG. 13 is an explanatory diagram of a conventional dictionary configuration.

FIG. 14 is an explanatory diagram of a conventional dictionary memory.

FIG. 15 is an explanatory diagram of a data restoration process using a conventional LZW code using the external hash method.

FIG. 16 is an explanatory diagram of a conventional dictionary configuration.

FIG. 17 is an explanatory diagram of a conventional dictionary memory.

[Explanation of symbols]

 10 data compression device 11 dictionary 12 dictionary search means 13 dictionary registration means 14 addition means 15 determination means 20 data decompression device 21 dictionary 22 dictionary search means 23 dictionary registration means 24 addition means 25 determination means

Front page continued (72) Inventor Shigeru Yoshida 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited

Claims (4)

[Claims] 1. A structure in which input data is decomposed into character strings different from each other and the character strings are managed in a dictionary together with a reference number, and the input data is referred to the longest matching character string registered in the dictionary. In the data compression processing method that employs a configuration of encoding using numbers, the dictionary employs a configuration that manages the correspondence relationship between the character string and the reference number according to the external hash method that uses the reference number of the character string as the hash address. At the same time, when the character string is registered in the dictionary, an adding means (14) for adding the additional information indicating that the character string has been registered in association with the reference number, and the adding means for searching the dictionary. A data compression processing method, comprising: a determination unit (15) for determining whether a character string / reference number to be searched from a dictionary according to the additional information added by (14). 2. The data compression processing method according to claim 1, wherein the adding means (14) adds additional information for identifying whether the dictionary data is correct or incorrect, instead of the registered additional information, and the determining means. (15) is a data compression processing method characterized by performing processing to determine whether the character string / reference number searched from the dictionary is correct or not according to the additional information. 3. The restored data is decomposed into character strings different from each other, the character strings are managed in a dictionary together with a reference number, and a corresponding character string in the dictionary is specified from a reference number designated by a codeword. In the data restoration processing method, in which the codeword being input is replaced with this specified character string, the dictionary uses the reference number of the character string as the hash address. According to the external hash method, the correspondence between the character string and the reference number is managed, and when the character string is registered in the dictionary, the additional information indicating that the character string has been registered is added to the reference number. And an addition means (24) for associating with the above, and a determination means (25) for determining whether the character string / reference number to be searched from the dictionary is correct according to the additional information added by the addition means (24) at the time of dictionary search. Equipped with A data restoration processing method characterized by that. 4. The data restoration processing method according to claim 3, wherein the adding means (24) adds additional information for identifying whether the dictionary data is correct or incorrect, instead of the registered additional information, and the determining means. (25) is a data restoration processing method characterized by performing processing so as to determine whether the character string / reference number searched from the dictionary is correct according to the additional information.