JP3001519B1 - Data compression method and data compression method - Google Patents

Abstract

A data compression method based on the LZW encoding method, which reduces the encoding processing by all the data to be compressed and all the dedicated dictionaries performed by the conventional method, and speeds up the encoding processing. Provide a data compression scheme that can be achieved. SOLUTION: In a data compression system based on an LZW encoding system, a learning dictionary (12-1) and one or more specialized dictionaries (12-2 to 12-4) are provided, and a data dictionary to be encoded is provided. After performing compression learning on a part of the compressed data using all of the dictionaries, the dedicated dictionary in the field closest to the compressed data content is selected and fixed, and the compressed data left by the selected and fixed dedicated dictionary Compression.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a data compression method according to the LZW coding is one of the universal coding scheme
And a data compression method.

[0002]

Data compression method definitive in LZW coding incremental decomposition type which is one of the Related Art Universal coding method, in particular encoding method at the time of compression encoding process of the conventional method, the learning dictionary and several And the encoding and dictionary learning are performed in all learning dictionaries and dedicated dictionaries for all compressed data to be compressed. Then, to the compressed data encoded by the dedicated dictionary having the highest value of the compression ratio among them, dictionary selection data that allows the decoder to determine in which dedicated dictionary the compressed data was encoded is added and decoded. For example, the code decoding process is reduced by sending the data to a device (see FIG. 21).

However, conventional methods are encoded using all of the compressed data and all dictionary, for the method of determining the compression ratio of several dictionaries used upon completion of the coding operation, Dictionary There was a waste of time before the selection.

[0004]

Therefore, the present invention relates to
A data compression method and data compression method according to the ZW coding to reduce coding processing by all of the compressed data as well as all special dictionary has been performed in a conventional manner, to increase the speed of the encoding process Data compression methods
And a data compression method.

[0005] The present inventor has conducted extensive studies to achieve the above object, Oite the LZW coding is one of the universal coding scheme, dictionary and one or more for the original training By providing a specialized dictionary and improving the encoding efficiency, by learning compression of a certain amount of compressed data using all dictionaries, grasping the tendency of compressed data from a part of the compressed data to be encoded, The present inventors have conceived that it is possible to reduce the encoding processing time by selecting a specialized dictionary specializing in such a tendency and performing encoding based on the selected dictionary, and have accomplished the present invention.

Accordingly, the present invention provides a data compression method based on LZW encoding , wherein a learning dictionary and one or more specialized dictionaries are provided, and a part of the compressed data to be encoded is provided. After performing compression learning using all of the dictionaries, the dedicated dictionary in the field closest to the compressed data content is selected and fixed, and the selected and fixed compressed dictionaries remaining are compressed. A data compression method is provided.

[0007] which is one of the incremental decomposition type LZW universal coding scheme (referred to as Ziff Lempel. Hereinafter referred to as LZW) In particular encoding method at the time of compression of the data compression method definitive in coding, the sign of the conventional method As described above, the learning process is performed by providing a learning dictionary and several types of dedicated dictionaries, performing encoding and dictionary learning on all learning dictionaries and dedicated dictionaries for all compressed data to be compressed, The decoder assigns dictionary selection data indicating which dedicated dictionary the compressed data was encoded to to the compressed data encoded by the exclusive dictionary showing the highest compression rate, and sends it to the decoder. For example, the code decoding process is reduced.

In the present invention, although the dictionary form is the same, the conventional method encodes all the data to be compressed and all the dictionaries. After the compression learning, the dedicated dictionary of the field closest to the compressed data content is selected and fixed, and by compressing the remaining compressed data, all the compressed data and the compressed data that have been performed by the conventional method are obtained. The encoding processing by all dedicated dictionaries is reduced, and the encoding processing is speeded up.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG.
A description will be given using a W encoder and the LZW decoder of the present invention in FIG.

The first configuration of the present invention is a character string input buffer 10 for storing the compressed data 42 of the compressed data frame of the first embodiment of FIG. 4, a learning dictionary 12-1, and suitable for the communication field. A coding dictionary 12 including a communication dictionary 12-2, a voice dictionary 12-3 suitable for the voice field, and an image dictionary 12-4 suitable for the image field, and a learning dictionary code string corresponding to each dictionary Output buffer 14, communication-dedicated dictionary code string output buffer 15, audio-only dictionary code string output buffer 16, and image-only dictionary code string output buffer 17
A function of encoding the compressed data 42 from the input buffer 10 by the LZW method using the dictionary selector 13 and the respective dictionaries, and outputting the encoded data to the output buffer; A dictionary most suitable for encoding compressed data from the learning dictionary and the dedicated dictionary, which stores registered characters learned in the dedicated dictionary and data indicating how much the registered characters appear in the compressed data. And a LZW encoding unit 11 having a function of outputting a selection signal 18 to the dictionary selector 13.

A second aspect of the present invention, in the first configuration, as shown in the compressed data frame of FIG. 6 a second embodiment, the dictionary selection flag indicating whether the data to be compressed in advance which areas This is a data compression method including the LZW encoding unit 11 having a function of encoding the attached compressed data 43.

Generally, in the universal coding method, FIG.
A dictionary is created according to the dictionary configuration shown in FIG. Usually LZ
In the case of the W method, basic alphanumeric characters (basic registration characters 92) and some control words are registered in the dictionary. Since the scope of the present invention is limited to the compression operation, the description of the exception processing in the LZW method is omitted here. Therefore, description of the control word will not be given.

First, the operation of the LZW encoder according to the first embodiment will be described with reference to FIG. 1 using a learning dictionary for the encoder. The LZW encoding unit 11 encodes the compressed data 42 according to the encoding procedure of FIG. The operation will be described below according to this procedure.

First, the LZW encoding unit 11 has 259 (0 to 258) basic registered characters (a, b, c, d,... 0, 1, 2,...) As shown in the learning dictionary of FIG. ..) 92 are registered (step 100).

Here, it is assumed that the character string input buffer 10 of the LZW encoder holds compressed data frames (abcabcabcaabbcc...) As compressed data 42. Next, the first character a of the frame
Is read (step 101), and it is searched whether or not it is registered in the learning dictionary 12-1 (step 102). In this case, since a is registered as the basic registration character 92 and the search matches, the next character b is pointed (step 103). The next pointed compressed character b is read, and it is searched whether ab is registered in the learning dictionary 12-1 (steps 101 to 102). Since the search character string ab is not registered, it is registered in the learning dictionary 12-1 as a newly registered character 93, and the next contact number 25
9 is given (step 104).

Next, the character a matched in the previous search is shown in FIG.
2 codeword (000) shown in the communication dictionary of the LZW decoder 2
000000) and outputs it to the learning dictionary code string output buffer 14 (step 105). Next, since the compressed character string is not at the end (step 106), the process returns to the reading of the compressed character string (step 101), and the character search of the character b previously read is performed. In this case, since the character b is registered as the basic registered character 92, nothing is output,
A search is performed by pointing to the next character c to determine whether bc is registered in the learning dictionary 12-1. Since the character bc is not registered, it is registered as a newly registered character 93, and the next contact number 260 is assigned. Also in this case, the code word (0
00000001) is output to the code string output buffer 14.

In the same manner as described above, encoding of the subsequent compressed character string is performed in an incrementally decomposing type, a learning dictionary as shown in FIG. 7 is created, and a code word string (000000000,
000000001000000000010, 10000
0011,...) Are output from the code string output buffer to the opposite decoder (steps 101 to 106). The encoding operation is performed by the communication dictionary 12-2, the audio dictionary 12-3 (FIG. 10), and the image dictionary 12-4.
The same applies to FIG. 11 until a dictionary is selected.

Next, FIG. 2 shows the LZW decoder of the present invention.
For decoding, the learning dictionary 22 of the decoder is used in the same manner as the encoding.
Explanation is made using -1. The LZW decoding unit 21 decodes the compressed data according to the decoding procedure in FIG. The operation will be described below according to this procedure.

First, the LZW decoding unit 21 259 (0 to 2) as shown in the communication-dedicated dictionary in FIG.
58) basic registration characters (a, b, c, d,... 0,
, 83) are registered (step 20).
0). In this case, the difference from the coding dictionary is that the contact number sequence 8 corresponding to the code word 80 to be decoded is
1 and a character and a character string 82 corresponding to the contact number.

Here, the code string (000000000, 00000) encoded and output in the explanation of the encoding operation is output.
0001,00000000010, 100000011,
...) Are stored in the code string input buffer 20.

First, a code word (000000000) is read from the code string input buffer 20 (step 20).
1). The read codeword (000000000) is converted into a contact number (0) (step 202), and a search is performed to determine whether the codeword is registered in the learning dictionary 22-1 (FIG. 8) (step 203). In this case, since the contact number (0) is registered, the character a corresponding to the coincident contact number (0) is output to the voice dictionary character string output buffer 24 (step 204).

Subsequently, it is searched whether the character a is a new character (step 205). Character a is the basic registered character 9
Since it is 8, registration of a new character is skipped and the next code word is pointed (206, 207). The next code word (000000001) pointed to is read, and a dictionary search is performed in the same manner as the previous time. Since the code word (00000000001) corresponds to the contact number (1) and matches the basic registered character b, the character b is output to the learning dictionary character string output buffer 24 (steps 201 to 204). Here, the LZW decoding unit 21 adds the code word (10) to the character string ab since the decoded character string ab so far is not registered in the dictionary in this case.
0000011), a contact number (259) is assigned, and registered in the dictionary as a new character string 99 (steps 205 and 2).
06).

The next pointed code word (000000)
010) is the contact number (2), followed by the character c, followed by the code word (1)
000000011) corresponds to the ab of the contact number (259), and is output to the learning dictionary character string output buffer 24, respectively. Here, the character c and the following character string ab
Is not registered in the dictionary, the code word (10000000) and the contact number 260 are assigned, and the character string ca is registered as a new character string 99 in the dictionary (steps 201 to 207). In this manner, the decoding of the codeword (abcabcabcaabbcc...) And the creation of the decoding learning dictionary 22-1 are performed until the end of the compressed file.

The decoding operation is performed until the dictionary is selected for the communication dictionary 22-2, the audio dictionary 22-3, and the image dictionary 22-4 of the LZW decoder in the same manner as the description of the encoding operation. The same is done. If the code word is not registered in the decoding dictionary, an exception process (steps 203 and 209) is performed, but the description is omitted here because it is not within the scope of the present invention. As described above, LZ
Encoding and decoding of a character string by the W method are performed.

Next, the encoding and decoding operations of the present invention when a dedicated dictionary is used will be described. First, the encoding operation will be described using the communication-dedicated dictionary according to the flowchart of the encoding procedure using the dedicated dictionary in FIG. FIG. 9 shows contents in which terms in the communication field are registered in the communication dictionary 12-2 in advance, and includes a contact number 50, a registered character string 51, and an appearance counter 52.

First, the LZW encoding unit 11 initializes (here, clears to 0) a dictionary search counter, an appearance counter, and a character pointer prior to the encoding process (step 30).
1). Next, the basic registration character 53 and the special registration character 54 corresponding to each field are registered (step 302). Here, the series of compressed data shown in FIG. 3 is stored in the character string input buffer 10 of the LZW encoder 11 from the first block to the Nth block.
+2 blocks are stored. The size of the data to be compressed for each block is arbitrary.
It is assumed to be K bytes. In the first block, the compressed data frame (abca) shown in FIG. 4 is stored as the compressed data in the same manner as in the description of the encoding using the learning dictionary 12-1.
bcabcaabbcc...) 42 are held.

Subsequently, a dictionary search count value for determining a coding range for obtaining a dedicated dictionary selection value for determining which dedicated dictionary is to be used for encoding (here, 1 Kbyte · T
0) is checked (step 303), but since the initialization has been completed, the first character a is read from the compressed data frame 42 (step 304). Next, it is searched whether the character a is registered in the communication dictionary 12-2 (step 305). Since the character a is registered as the basic registered character 53, the appearance counter is counted up (+1).
(Step 306), and point to the next character (step 307). Read the next character b pointed to.
A search is made to determine whether the character string ab is registered in the communication dictionary 22-2. Since the character string ab is registered as a dedicated character string in the communication field in the communication-dedicated dictionary, the character string ab
Is counted up, and the next character is pointed (steps 304 to 307). Next, the pointed character c is read. Character string abc is communication dictionary 12
-Search whether it is registered in 2. Since the character abc is registered, the occurrence counter of the character string abc is counted up, and the next character is pointed (steps 304 to 307). The pointed character c is read, and it is searched whether the character string abca is registered in the communication dictionary 12-2. In this case, since the character string abca is not registered in the dictionary, a contact number (266) is assigned and registered in the communication dictionary 12-2 (step 304).
308). Next, the code word of the previous matching character string abc is output to the communication-specific code string output buffer 15 (step 309).

The above-described encoding of a series of character strings (steps 304 to 309) is performed in the same manner as the encoding by the learning dictionary 12-1, and is continued until the dictionary search count value reaches 1 Kbyte (step 303). ). Until the dictionary search count value reaches 1 KB T0, the learning dictionary 12-1, the communication dictionary 12-2, and the voice dictionary 12-
3 and all the image-dedicated dictionaries 12-4, new character strings are registered in the same manner, and the number of appearances of compressed data corresponding to the characters registered in each of the dictionaries is recorded in the appearance counter 52. You. When the dictionary search count value reaches 1 Kbyte (step 303), L
The ZW encoding unit 11 performs a communication dictionary hit rate calculation (step 311). As shown in the dictionary selection value and the threshold value in the embodiment of the present invention in FIG. 14, this calculation is calculated by: communication dictionary hit rate = sum of dedicated registered character appearance counts / dedicated dictionary contact number. FIG. 9 In the coded communication-dedicated dictionary, the sum total of the special registered character appearance count = 2 + 3 +
Since 6 + 8 + 7 + 5 + 4 = 35 and the number of dedicated dictionary contacts = 7, the communication dictionary hit rate = 5.0. Similarly, FIG. 14 shows the results obtained for the voice-only dictionary 12-3 and the image-only dictionary 12-4.
2-3 is 0.0, and the image-dedicated dictionary 12-4 is 1.7.

Subsequently, it is determined from the communication-specific dictionary hit rate which field the compressed data to be encoded is most suitable for. In this case, since the dictionary with the highest hit rate when the dictionary selection threshold ≥ 3.0 is the communication dictionary, the LZW encoding unit 11 performs the encoding of the remaining 9 Kbytes of compressed data using the communication dictionary 12. In addition to the output of the selection signal 18 for selecting −2 to the dictionary selector 13, as shown in the dictionary selection data and the corresponding dictionary table in FIG.
00) is also output to the communication-specific code string output buffer 15.
The dictionary selector 13 receiving the selection signal 18 makes the communication dictionary 1 as an encoding dictionary until the encoding of the current block is completed.
2-2 is fixed, and is not changed to another dedicated dictionary (step 314). Next, when the first block to be encoded ends, in preparation for obtaining the communication dictionary hit rate in the next block, the appearance counter and dictionary search pointer of each dedicated dictionary are initialized, and the dictionary selector 13 is initialized. Is reset and the selection signal 18 is terminated (step 31).
5, 316).

As described above, the data to be compressed in the first block is obtained by the special registered characters registered in the special dictionary in advance.
The encoded data is encoded at a higher compression ratio than the ordinary LZW encoding, is assembled into the compressed data frame shown in FIG. 5, and is output from the communication dictionary encoded sequence output buffer 15 to the opposite decoder. If the value is less than the threshold value 3.0, it is assumed that the compression by any of the dedicated dictionaries does not provide the effect, and the learning dictionary 12-1 does not.

Next, the decoding operation of the present invention will be described using a communication-dedicated dictionary according to the decoding procedure using the dedicated dictionary in FIG. FIG. 12 is a communication-dedicated dictionary content in which terms in the communication field are registered in advance, and includes a code word 80, a contact number 81, and a registered character string 82.

The LZW decoding unit 21 resets the dictionary selection signal 28 to the dictionary selector 23 and clears the contact number counter (Step 400). Next, the basic registration character 83 and the special registration character 84 corresponding to the communication field are registered (step 401). Since the dictionary selection data 40 is not sent until the compressed data of 1 Kbyte is compressed, the selection of the decoding dictionary is not performed here (steps 402 and 403). Until the dictionary selection data 40 is sent from the LZW encoder, decoding is performed by all of the learning dictionary 22-1, the communication dictionary 22-2, the audio dictionary 22-3, and the image dictionary 22-4. The decoded character strings are sent to the learning dictionary character string output buffer 24, the communication dictionary character string output buffer 25, the audio dictionary character string output buffer 26, and the image dictionary character string output buffer 28, respectively.
It is stored in.

When the dictionary selection data 40 arrives, the LZW decoding unit 21 encodes the compressed data after the dictionary selection data 40 by the coding dictionary from the dictionary selection data and the corresponding dictionary table shown in FIG. Is determined (step 402), and a dictionary selection signal 28 is output to the dictionary selector 23 of the LZW decoder to set a decoding dictionary. The dictionary selector 23 selects the dedicated dictionary notified from the decrypted dictionaries (step 403). in this case,
Since it has received the code word (10000000), the communication-dedicated dictionary 22-2 is selected and fixed until the next dictionary selection data 40 comes.

When the compressed data following the dictionary selection data 40 is read (step 404), it is converted into a contact number corresponding thereto (step 405), and a character string corresponding to the contact number is searched in the communication dictionary 22-2. (Step 406). Here, the codewords following the dictionary selection data 40 (... 10,0000000, 0000000000,
… ..), the code word (1000010)
00) is read, and it is searched whether or not the contact number (264) is registered in the communication dictionary 22-2. Since the character string abcc corresponding to the contact number 264 is registered in the communication dictionary, this character string is output to the character string output buffer (steps 404 to 407), and is decoded just before the dictionary selection data. The combination fabcc with a character (here, f is assumed) is the communication dictionary 22-2.
(Step 40)
8). If not registered, the code word (100101011) and the contact number (299) are newly registered characters 85.
And registers it in the communication dictionary 22-2 (steps 408 and 409). Do nothing if registered. Next, the code word (0000000000) to be pointed is read, and a character e corresponding to the contact number (4) is similarly searched. Since this is registered as a basic registered word, the character e is output to the communication dictionary character string output buffer 25, and abcse, which is a combination with the immediately preceding decoded character, is registered in the communication dictionary 22-2. Search for If it is not registered, it is registered as a newly registered character 85 in the communication dictionary 22-1. Thereafter, decoding operations of the subsequent code words (402 to 409) are performed in the same manner, and when the compressed data file ends, the decoding process ends (step 411). If the codeword is not registered in the decoding dictionary, exception processing (steps 406 and 412) is performed, but the description is omitted here because it is not the scope of the present invention.

A second embodiment of the present invention will be described with reference to FIG. Here, the configuration of the code decoder of the second embodiment is the same as that of the first embodiment. FIG. 9 shows contents in which terms in the communication field are registered in the communication dictionary 12-2 in advance, and includes a contact number 50, a registered character string 51, and an appearance counter 52.

First, the LZW encoding unit 11 initializes (here, clears to 0) a dictionary search counter, an appearance counter, and a character pointer prior to the encoding process (step 50).
0). Next, the basic registration character 53 and the special registration character 54 corresponding to each field are registered (step 501). Here, a series of compressed data shown in FIG. 6 is stored in the character string input buffer 10 of the LZW encoder 11. Similarly to the description of the encoding using the learning dictionary 12-1, the compressed data frame (abcabcabcaabbcc...) 42 shown in FIG.
Is held.

Subsequently, it is checked whether a dictionary selection flag has been added to the compressed data frame or the dictionary selection mode has already been set (step 502). In this case, since the dictionary selection flag is given, the dictionary selection flag mode is set, and the dictionary is selected from the dictionary and the corresponding dictionary shown in FIG.
Here, it is assumed that a dictionary selection flag (00) for selecting the communication dictionary 12-2 has been designated. When neither the dictionary selection flag nor the dictionary selection flag mode is set, a dictionary search count value (here, a dictionary search count value for determining a coding range for obtaining a dedicated dictionary selection value for determining which dedicated dictionary to use for coding). In this case, 1K bytes (T0) is checked (step 504), but the encoding process in this case is the same as in the first embodiment, and the description is omitted here.

In the dictionary selection flag mode, the first character a is read from the compressed data frame 42 (step 505). Next, the character a is entered in the communication dictionary 12
-2 to search whether it is registered in (Step 50
6). Since the character a is registered as the basic registered character 53, the appearance counter is counted up (+1) (step 507), and the next character is pointed (step 50).
8). Read the next character b pointed to. Character string a
A search is made to determine whether or not b is registered in the communication dictionary 22-2. Since the character string ab is registered in the communication-dedicated dictionary as a character string dedicated to the communication field, the occurrence counter of the character string ab is counted up and the next character is pointed (steps 505 to 508). Next, the pointed character c is read. A search is performed to determine whether the character string abc is registered in the communication dictionary 12-2. Since the character abc is registered, the appearance counter of the character string abc is counted up, and the character abc is further pointed to (step 5).
05-508). The pointed character c is read, and it is searched whether the character string abca is registered in the communication dictionary 12-2. In this case, since the character string abca is not registered in the dictionary, a contact number (266) is assigned and registered in the communication-dedicated dictionary 12-2 (step 50).
5-509). Next, the code word of the previous matched character string abc is output to the communication-specific code string output buffer 15 (step 510). The decoding operation according to the second embodiment of the present invention is the same as the decoding operation using the dedicated dictionary according to the first embodiment, and the description is omitted here.

As described above, by assigning a dictionary selection flag, which is dedicated dictionary selection information used for encoding, to the data to be encoded in advance, a dedicated dictionary suitable for the data to be compressed is calculated. The selection process is not required, and the encoding process when using the dedicated dictionary is reduced.

[0040]

As described above, in the LZW encoding method of the present invention, the dictionary form is the same as the conventional one as shown in the encoding range of the dictionary selection data in FIG. In contrast to encoding using data and all dictionaries, in the present invention, after a certain amount of compressed data is subjected to compression learning using all dictionaries, a dedicated dictionary in the field closest to the content of the compressed data is obtained. By selecting, fixing and compressing the remaining compressed data, the encoding processing by the conventional method with all the compressed data and all the dedicated dictionaries is reduced, and the encoding processing is accelerated. Can be achieved. Coding reduces time between the time of encoding according to the conventional method in FIG. 3 Ta, the set of block number i in the present invention 0,1,2,3, and ...... n + 2 and to represent the encoding time of each block of Ti And the following equation.

(Equation 1)

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an example of an LZW encoder according to the present invention.

FIG. 2 is a block diagram illustrating an example of an LZW decoder according to the present invention.

FIG. 3 is a diagram showing an encoding range of dictionary selection data.

FIG. 4 is a diagram illustrating a compressed data frame according to the first embodiment.

FIG. 5 is a diagram showing a compressed data frame.

FIG. 6 is a diagram illustrating a compressed data frame according to a second embodiment.

FIG. 7 is a diagram showing an encoding learning dictionary.

FIG. 8 is a diagram showing a decoding learning dictionary.

FIG. 9 is a diagram showing a dictionary dedicated to encoded communication.

FIG. 10 is a diagram showing a coded voice-only dictionary.

FIG. 11 is a diagram showing a dictionary dedicated to encoded images.

FIG. 12 is a diagram showing a communication-dedicated dictionary of the LZW decoder.

FIG. 13 is a table showing dictionary selection data and its corresponding dictionary.

FIG. 14 is an expression and a table showing a dictionary selection value and a threshold value in the embodiment of the present invention.

FIG. 15 is a flowchart showing an encoding procedure using a learning dictionary.

FIG. 16 is a flowchart showing a decoding procedure using a learning dictionary.

FIG. 17 is a flowchart showing an encoding procedure using a dedicated dictionary.

FIG. 18 is a flowchart showing a decoding procedure using a dedicated dictionary.

FIG. 19 is a flowchart showing a procedure for encoding a dictionary dedicated to the second embodiment.

FIG. 20 is a table showing flags and corresponding dictionaries in dictionary selection according to the second embodiment.

FIG. 21 is a block diagram of a conventional encoding system.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Character string input buffer 11 LZW encoding part 12 Encoding dictionary 12-1 Learning dictionary 12-2 Communication dictionary 12-3 Audio dictionary 12-4 Image dictionary 13 Dictionary selector 14 Learning dictionary code string output buffer 15 Communication-dedicated dictionary code string output buffer 16 Audio-only dictionary code string output buffer 17 Image-dedicated dictionary code string output buffer 18 Selection signal 20 Code string input buffer 21 LZW decoding unit 22 Decoding dictionary 22-1 Learning dictionary 22-2 Communication-only dictionary 22-3 Voice-only dictionary 22-4 Image-only dictionary 23 Dictionary selector 24 Learning dictionary character-string output buffer 25 Communication-only dictionary character-string output buffer 26 Audio-only dictionary character-string output buffer 27 Image-only dictionary character-string output Buffer 28 selection signal

Continuation of front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H03M ^7/ 40-7/42

Claims (6)

(57) [Claims] We claim: 1. have you <br/> the data compression method according to the LZW coding, provided with a learning dictionary and one or more specialized dictionary, said dictionary a portion of the compressed data to be encoded After performing compression learning using all of the compressed data, selecting and fixing the dedicated dictionary in the field closest to the compressed data content, and compressing the selected compressed data left by the fixed dedicated dictionary. Data compression method . 2. A character string input buffer for storing data to be compressed, an encoding dictionary including a learning dictionary and one or more specialized dictionaries, a code string output buffer corresponding to each dictionary, and a dictionary A function of encoding the data to be compressed from the input buffer using the selector and each of the dictionaries, and outputting the encoded data to the output buffer; The most suitable dictionary when encoding the compressed data is selected from the learning dictionary and the dedicated dictionary that stores data indicating how many characters appear in the compressed data, and a selection signal is sent to the dictionary selector. 2. The data compression method according to claim 1, further comprising an LZW encoding unit having a function of outputting the data. 3. LZW coding unit, the data compression method of claim 2 having a function to encode該被compressed data dictionary selected flagged to indicate whether data to be compressed in advance which areas. 4. A code string input buffer for storing compressed data, a decoding dictionary including a learning dictionary and one or more specialized dictionaries, a character string output buffer corresponding to each dictionary, and a dictionary selection buffer. And a function of decoding the compressed data from the input buffer using the respective dictionaries and outputting the decoded data to the output buffer. The registered characters and the registered characters learned in the respective dictionaries are A function of selecting a dictionary most suitable for decoding compressed data from the learning dictionary and the dedicated dictionary that stores data indicating how many times the compressed data appears, and outputting a selection signal to the dictionary selector. 2. An LZW decoding unit having:
2. The data compression method according to 1 . (5) Character string input buffer that stores compressed data
And learning dictionary, communication-specific dictionary suitable for the communication field, sound
Voice-only dictionary and image field suitable for voice field Picture suitable for
An encoding dictionary with a dedicated image dictionary, and
Learning dictionary code string output buffer, communication-specific dictionary code string
Output buffer, voice-only dictionary code string output buffer and
Image-only dictionary code string output buffer, dictionary selector, and
From the input buffer by the LZW method using each dictionary
The data to be compressed is encoded, and the encoded data is output as described above.
Function to output to the output buffer
The registered characters and the registered characters learned in
For learning, which stores data indicating how many times it has appeared
When encoding data to be compressed from a dictionary and a dedicated dictionary
Select the most suitable dictionary and output a selection signal to the dictionary selector
Data compression comprising an LZW encoder having a function of performing
method. 6. Prefix indicating whether the data to be compressed in advance which areas
Encodes the data to be compressed with a document selection flag
The data according to claim 5, further comprising an LZW encoding unit having a function.
Data compression method.