JPH05127866A - Image data compression system - Google Patents

Abstract

PURPOSE:To improve the compressibility of the image data by performing the approximate retrieval of the registered character strings to a dictionary including the patterns and run lengths similar and approximate to each other, coding these retrieved character strings, and registering approximately the character strings obtained by adding the input characters to those registered character strings. CONSTITUTION:The character string inputted from a pre-processing means 10 is retrieved out of a dictionary 12 by a dictionary retrieving meens 14. If the dictionary 12 does not include the character string that is coincident with the relevant character string including the input characters, the characters approximate to the input characters are read out by an approximate dictionary retrieving means 16. Then the means 16 retrieves a character string that is coincident in the largest length with the character string including the approximate characters. A coding means 18 codes the character string retrieved by the means 14 or 16 in a reference number of the dictionary 12. Then the character string obtained by adding the input characters to the reference number of the character string that is coded immediately before detection of a fact that the character string coincident with the character string including the input characters cannot be retrieved out of the dictionary 12 and no approximate character is obtained is registered into the dictionary 12 as a new reference number.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression system which applies universal coding after converting an image into a pattern of two-dimensional information and a run length. In recent years, office automation has been developed, and documents are handled as binary image information in a facsimile or an optical disk file system. When the document information is used as digital data, the data amount of the image information is much larger than that of the character image and is ten to several ten times as large.

Recently, in order to improve the image quality, in a facsimile, about 200 dp of a conventional G3 machine is used.
The resolution will increase from i to 300 dpi or 400 dpi on the next G4 machine, and the amount of data will increase. Therefore, in order to handle image information efficiently in storage, transmission, etc., it is essential to reduce the data amount by adding efficient data compression.

[0003]

2. Description of the Related Art The inventors of the present application have already proposed a method of universal encoding after adding preprocessing for capturing a two-dimensional property to a black and white binary image. According to this method, a black and white binary image is obtained. Conventional two known as compression method
Compression rates comparable to two schemes have been obtained, namely the MMR scheme and the predictive coding scheme.

This method does not depend on the type of data and can be efficiently compressed by one method, that is, by a simple circuit. FIG. 8 shows one of the preprocessings for capturing the two-dimensional property of a black and white binary image. FIG. 8A shows an original image, and the number of consecutive patterns of the same type as the black-and-white pattern for each 4-line pixel for this original image,
That is, the data is converted into run length information to obtain the data shown in FIG. The converted data is, for example, 8-bit data in which the upper 4 bits indicate the type of pattern and the lower 4 bits indicate the run length.

The universal coding method is used to optimize the code while learning the statistical properties such as continuity of the conversion data consisting of the pattern and the run length, and efficient compression is performed on images of various properties. It can be performed. FIG. 9 shows a tree structure of a dictionary registered by universal encoding performed on the converted data of FIG. 8B.

Originally, the universal code is an information storage type data compression method, and since it does not assume the statistical property of the information source in advance at the time of data compression, it is applied to various types of data such as character codes and object codes. can do. In a document image, there are similarities in character outlines and character spacing. Further, the halftone images have similar halftone dot periodicity, halftone dot shape identity, and the like. The redundancy with this similarity can be reduced by universal coding, and effective compression can be performed.

[0007] Here, the LZW code which is one of the universal coding is adopted (TA Welch, "A Techniqu.
e for High-Performance Data Compression ”, Compute
r, June 1984). In the LZW code, the next symbol is incorporated in the next subsequence so that only the index (dictionary reference number) can be used for encoding.

Initially register 256 characters in FIG.
FIG. 11 shows a tree structure of a dictionary created when encoding is performed, and FIG. 11 shows data (character strings) and LZW in the LZW code.
An index (reference number of dictionary) as a code word obtained by encoding is shown. FIG. 12 is a flowchart showing a detailed algorithm of conventional LZW coding.

This LZW encoding has a rewritable dictionary, divides the input character string into different character strings, assigns numbers to the character strings in the order in which they appear, and registers them in the dictionary. The present character string is represented by the number of the longest matching character string registered in the dictionary and is encoded. Figure 1
In the LZW encoding process of No. 2, first, in step S1, a character string consisting of one character for all characters is registered in advance as an initial value, and then encoding is started. In step S2, the input first character K is used as a reference number ω for dictionary search, and this is used as a prefix string. Next, in step S3, the next character K of the input data is read, and in step S4, whether or not (ωK) obtained by adding the character K read in step S3 to the initial character string ω obtained in step S2 is in the current dictionary. Or search.

If the character string (ωK) is found in the dictionary in step S4, the character string (ωK) is replaced with the reference number ω in step S5, and it is determined in step S6 whether or not the input data has ended. Returning to this, the search for the maximum matching length is continued until the character string (ωK) cannot be searched from the dictionary.
Next, in step S4, if the character string (ωK) is not in the dictionary,
Proceeding to step S7, the reference number ω of the character string obtained in step S2 is output as a code word code (ω), and a new reference number is added to the character string (ωK) and registered in the dictionary. The input character K is replaced with the reference number ω, the dictionary address N is incremented, and after the check in step S5 is received, the process returns to step S2 to read the next character K.

Next, the LZW encoding will be specifically described with reference to FIGS. 14 and 15. Note that, in FIGS. 14 and 15, the case of compressing data consisting of a combination of three characters of abc is taken for simplification of description. In addition, 3 of a, b, c
The characters are initially registered in advance. First, the input data of FIG. 14 is read from left to right. When the first character a is entered, there is no matching character string other than the character a in the dictionary, so
The output code (reference number ω) is output as a code word. Then, the reference character 4 is attached to the expanded character string ab and registered in the dictionary. The actual registration is in the form of a character string (1b).

Then, the second b becomes the beginning of the character string.
Since there is no matching character string other than b in the dictionary, the reference number 2 is output as a codeword, and the expanded character string ba is actually added to the dictionary with the reference number 5 and registered in the dictionary. The third a is the beginning of the next character string. Hereinafter, this processing is similarly continued. Next, 13 LZW encoding will be described.

This decoding is the inverse operation of the encoding of FIG. First, similarly to the encoding, in step S1, a character string consisting of one character for all characters is registered in the dictionary in advance as an initial value, and then decoding is started. Next, in step S2, the first code (reference number) is read and the current CODE is OLD code.
Since the first code corresponds to any one-character reference number already registered in the dictionary, the character code (K) matching the input code CODE is searched for and the character K is output. The output character (K) is ch for exception processing later.
Set it in ar.

Next, in step S3, the next code is read and set in CODE as NEW code. Next, in step S4, the code CO input in step S3 is input.
It is checked whether DE is defined (registered) in the dictionary. Normally, the input codeword is registered in the dictionary by the processing up to the previous time, so the processing proceeds to step S5 and the code CODE is entered.
The character string code (ωK) corresponding to is read from the dictionary, the character string K is temporarily stacked in step S6, and the reference number
The code (ω) is set as a new CODE and returned to step S5 again, and the procedure of steps S5 and S6 is recursively repeated until the reference number ω reaches one character, and finally, the process proceeds to step S7 and the characters stacked in step S6. Is popped up and output in the LILO (Last In Fast Out) format.

At the same time, in step S7, a new reference number is added to the character string in which the code ω used last time and the first character K of the character string restored this time are represented as a set (ω, K) and registered in the dictionary. .. If the code is not registered in step S4 (which occurs when the immediately preceding reference number is referred to in encoding), the OLD is determined in step S9.
After returning code to CODE and returning code (OLDcode, char) to NEWcode, the process proceeds to step S5.

The decoding process will be described in detail with reference to FIG. Note that FIG. 16 shows the case where data consisting of a combination of three characters abc is compressed, and the three characters a, b, and c are pre-registered in advance in order to simplify the explanation. In FIG. 16, the first input code is 1, and the characters a, b, and c are already registered in the dictionary as reference numbers 1, 2, and 3 as shown in FIG. It is replaced with the character string a of the reference number that matches 1 and output. Next code 2
Is also replaced with the character b and output. At this time, the previously processed code and the first character b decoded this time
A new reference number 4 is added to (1b) which is a combination of and and is registered in the dictionary.

The third code 4 replaces 1b with ab by searching the dictionary and outputs the character string ab. At the same time, the code 2 processed last time and the first character a of the character string decoded this time
The character string 2a (= ba) in combination with
Is added and registered in the dictionary. Similarly, this process is repeated thereafter. However, the decryption of FIG. 16 has the following exception processing. This exception processing occurs in the decoding of the sixth input code 8. Code 8 is not defined in the dictionary at the time of decoding and cannot be decoded. In this case, the character string 5b obtained by adding the first character b of the previously decoded character string ba to the previously processed code 5
Is calculated and further replaced with 2ab and bab and output. Then, the reference number 8 is added to the character string 5b obtained by adding the character b of the character string decoded this time to the previous code word 5 to the output word of the character string and registered in the dictionary.

This exception processing is performed through the processing of steps S4 and S9 in the decoding processing flow of FIG.
Finally, in step S7, the character string is output and the new character string is added to the reference number and registered in the dictionary. Incidentally, FIG.
2. The encoding and decoding processes of FIG. 13 are performed while creating the same dictionary.

[0019]

However, black and white 2
In the conventional image data compression method in which LZW encoding is performed after performing preprocessing on the value image to convert it into data including a set of a pattern and a length, the image information is completely saved. The pre-processed data of very similar images shown in FIG. 8 is subjected to universal encoding as completely different data, which has been an obstacle to improving the compression rate.

The present invention has been made in view of the above-mentioned conventional problems, and provides an image data compression method for improving the compression rate by providing a closeness to the dictionary search of universal encoding. The purpose is to do.

[0021]

FIG. 1 illustrates the principle of the present invention. First, the present invention is directed to an image data compression method that applies universal encoding to two-dimensional information of an image. In the present invention which employs such an image data compression method, the preprocessing means 10 for converting the image data into characters composed of a set of patterns and run lengths, and the preprocessing means 1.
A dictionary search unit 14 that searches the dictionary 12 for a character string whose maximum length matches a character string including a character input from 0, and a character that matches the character string ωK including the input character K in the dictionary search by the dictionary search unit 14. When the string is not in the dictionary 12, the approximate character K ′ that can be approximated to the input character K is read, the character string having the maximum length matching the character string ωK ′ including the approximate character K ′ is searched again, and the same character string is searched. If there is not, the dictionary search is stopped, and if there is the same character string, the dictionary search means 16 that continues the dictionary search using the approximate character string ωK ′ and the dictionary search means 14 or the approximate dictionary search means 16 are searched. Encoding means 18 that encodes the character string ω with the reference number in the dictionary 12, and the character string ωK including the input character K from the dictionary 10.
When it is not possible to search for a character string that matches with and there is no approximate character K ′, a new reference number is added to the character string ωK obtained by adding the input character K to the reference number of the character string ω encoded immediately before. The dictionary registration means 20 for registering in the dictionary 12 is provided.

Here, the approximate dictionary search means 16 has a table 22 in which an approximate character K'having a set of patterns and lengths that approximates the input character K composed of a set of patterns and lengths is set in advance. 22 is referred to only by the input character K and the corresponding approximate character K ′ is read out. Further, the approximate dictionary search means 16 has a table 22 in which an approximate character K ′ is predetermined from the input character K ₁ and the input characters K ₀ and K ₂ before and after the input character K ₁ , and the table 22 is used as the input character K ₁ and before and after it. It is characterized in that an approximate character K ′ of the input character K currently being processed is read out by referring to the input characters K ₀ and K ₂ .

Further, the approximate dictionary retrieval means 16 reads characters having the same run length and similar patterns as approximate characters, or reading character strings having the same pattern and similar run lengths as approximate characters. Furthermore, the approximate dictionary search means 16 approximates the input character K ₁ when reading the approximate character K ′ having the same pattern and similar run length from the input character K ₁ and the input characters K ₀ and K ₂ before and after the input character K _1. The run lengths of the subsequent input character K ₂ are corrected so as to correct the run length mismatch of the character K ′.

[0024]

According to the image data compression method of the present invention having such a configuration, as shown in FIG. 2, preprocessing, universal approximation search, universal approximation encoding, universal approximation registration is applied to monochrome binary image data. , Is to do.

Here, in the universal approximate search for data composed of a set of patterns and run lengths obtained in the preprocessing, that is, an input character string, when the input character is not in the dictionary, the run length is the same and the pattern is If they are similar, the patterns are the same and the run lengths are similar, etc., based on the conditions such as, replace the input character with the approximate character and search the dictionary to find a longer character that matches the input character string. It can be converted into a column dictionary reference number, which further improves the compression rate of universal encoding.

The reading of the approximate character is performed on the input character one to one.
In addition to the method of storing the approximate character corresponding to the above in the table and reading the table, the approximate character table may be created and read from the relationship between the input character and the characters before and after the input character.

[0027]

FIG. 3 is a block diagram of an embodiment showing one embodiment of the present invention. In FIG. 3, 24 is a CP as a control means.
U, the program memory 26 for the CPU 24
And the data memory 30 are connected. Program memory 2
6 is provided with control software 28, preprocessing software 10, dictionary search software 14, approximate dictionary search software 16, encoding software 18, and dictionary registration software 20.

The control software 28 performs preprocessing of image data and overall control of LZW encoding. The pre-processing software 10 uses, for example, black-and-white binary image data as shown in FIG. 8A as a pattern showing the number of repetitions of the same pattern every four lines and run-length image data, that is, an input character in LZW conversion. Convert. The input character converted by this preprocessing is an 8-bit character code in which the pattern is represented by upper 4 bits and the run length RL is represented by lower 4 bits as shown in FIG. 8B.

The dictionary search software 14 searches the registered dictionary 12 for a character string whose maximum length matches the character string including the input character consisting of the pattern and run length obtained by the preprocessing software 10. Further, in the present invention, the approximate dictionary search software 16 is newly provided. The approximate dictionary search software 16 is an approximate character that can approximate the input character at that time when the dictionary registration character string that matches the character string including the input character is not in the dictionary 12 by the dictionary search by the dictionary search software 14. Is searched from the approximate information table 22, and the dictionary 12 is searched for a character string having a maximum length matching the character string including the read approximate character. If the same character string is still not found, the search is stopped and the same character string can be searched. If so, the string search is continued using the approximate character string.

The encoding software 18 performs encoding to output the index (dictionary reference number) in the dictionary 12 of the character string retrieved by the dictionary retrieval software 14 or the approximate dictionary retrieval software 16 as a code code. Further, the dictionary registration software 20 cannot search the character string in the dictionary 12 that matches the character string including the input character by the dictionary search software 14, and the approximate dictionary search software 16 refers to the approximate information table 22 to approximate the character string. When the character cannot be read, a new index is added to the character string obtained by adding the input character to the reference number of the character string encoded immediately before, and the character string is registered in the dictionary 12.

On the other hand, the data memory 30 is provided with a dictionary 12, an approximation information table 22 and a data buffer 24. The dictionary 12 is created by the dictionary registration software 20 while performing LZW encoding. The data buffer 24 stores image data to be preprocessed and character strings to be LZW encoded after preprocessing.

Further, in the approximate information table 22, the input character by the approximate dictionary search software 16 and the approximate character read out according to the relationship of the characters before and after the input character are registered in advance. FIG. 4 shows a first embodiment of the approximation conditions stored in the approximation information table 22 of FIG. 3. In this embodiment, the approximation conditions are preset when the run lengths are the same and the patterns are similar. is doing. There are two conditions for this approximation, that is, the case where the approximation is judged only by the input data in FIG.
There is one.

When the approximation is judged only by the input data of FIG. 4A, if the input character K has a pattern of white = 2 and black = 2 and the run length RL = 2, the input character K is input. Run length RL is an approximate character of RL =
Approximate letter K, which is the same as 2, but with a different pattern
₁ 'and K _2' should define a. Approximate character K ₁ 'is black = 3, which is one more black than character K, and approximate character K _{1
In 2} ', there is one black less than the letter K, that is, black = 1.

By setting the approximation condition shown in FIG. 4 (a) in the approximation information table 22, the approximation character K ₁ 'or K ₂ ' is read out by referring to the input character K and the approximation dictionary search is performed. become. In FIG. 4B, in the case where the run lengths are the same and the patterns are similar to each other, the preceding and following data are also referred to determine the approximation. That is, the current input character is K ₁ , the previous input character is K ₀ ,
Further, if the next input character and was K _2, black pixels before and after the character K _0, K ₂ is the same as the current character is the approximate character one more K'.

By presetting the condition for judging the approximation by also looking at the data before and after shown in FIG. 4B, the input character K ₁ currently being processed and the characters K before and after it are processed. _The approximate information table 22 according to ₀ and K ₂ .
, And K ′ is read as an approximate character of the input character K ₁ . FIG. 5 shows a second embodiment of the approximation conditions stored in the approximation information table 22 of FIG. 3. In this embodiment, approximation conditions with the same pattern but similar run lengths are set. There is.

FIG. 5 (a) shows a case where the approximate character is judged only by the input data when the patterns are the same and the run lengths are similar. That is, assuming that the input character K at the present time has a run length RL = 3 in a pattern of white = 1 and black = 3, a character K ₁ 'of RL = 2 having the same pattern and a run length of ± 1 is obtained. RL = 4 letter K ₂ '
Are the approximate characters of the character K, respectively.

Further, FIG. 5B shows a condition for making an approximate judgment by also looking at the data before and after when the patterns are the same and the run lengths are similar. That is, the current input character is K
_If it is 1, the previous character is K ₀ , and the next character is K ₂ , the pattern is the same as the three characters K ₀ , K ₁ , K ₂ as shown on the right side. There is one more run length corresponding to the input character K ₁ currently being processed RL = 4
Is defined as an approximate character K '.

In the case of FIG. 5B, the character K before approximation
_While the sum of the run lengths of ₀ , K ₁ and K ₂ is 11, the sum of the three characters K ₀ , K ′, K ₂ and the run length in the case of approximation is 12, therefore The run length RL = 7 of the input character K ₂ is reduced by one so that RL = 7-1 = 6 is corrected so that the run length does not change as a whole even if approximation is performed.

In the case of approximating only the input data of FIGS. 4A and 5A, the approximate character K ₁ 'or K is referred to by referring to the approximate information table 22 by the input character K.
_{Although the} case of reading out ₂ 'is taken as an example, since the approximation relations of these three characters are mutually established, for example, when the input character is K ₁ ', K or K ₂ 'as an approximate character. Will be read.

FIG. 7 is a flow chart showing the LZW encoding of the present invention when the approximation judgment is made only with the input data. In FIG. 7, first, in step S1, the dictionary is initialized so as to include the first character, and the start address n of the dictionary is, for example, input data (input character consisting of a set of preprocessed pattern and run length). ) Is 8 bits, n = 256. Then enter the first letter K,
Let this input character be an index (initial character string) ω.

Next, in step S2, the next character K is input. Then, in S3, it is checked whether or not the character string ωK exists in the dictionary. Step S if it exists
4, the character string ωK is used as a new initial character string ω, the data end is checked in step S5, and the process returns to step S2 to repeat the search for the longest matching character string.

If the character string ωK does not exist in the dictionary in step S3, and the search for the longest character string is completed, the process proceeds to step S6, and the approximate search is performed in the processes after step S6. In step S6, an approximate character candidate K ′ similar to the input character K is obtained from the conditions as shown in FIGS. 4 (a) and 5 (a).
Is in the approximate information table created in advance, and the process proceeds to step S11. In step S11, the initial character string ω at this time is output as the code code (ω), the character string ωK obtained by adding the character K to the initial character string ω is registered in the dictionary, and the character K is further added. A new prefix ω
And the dictionary address n is incremented by one.

On the other hand, if there is an approximation candidate K'in the approximation information table in step S6, the approximation candidate K'is read out in step S7, and the approximation character string ωK 'is read in step S8.
Check if exists in the dictionary. If it exists in the dictionary, it can be approximated and the approximate character string ωK ′ is replaced with the initial character string ω in step S9. Then, in step S10, the next input character is read, and the process returns to step S3 to continue the dictionary search. In step S10, when the run length needs to be corrected as shown in FIG. 5B, the run length value of the next input character K is corrected, and then the process returns to step S3. When only the input data in 6 is used, this correction is not necessary.

On the other hand, if the approximate character string ωK 'does not exist in the dictionary in step S8, the process returns to step S6 again.
A new approximation candidate K'is searched for. Finally approximation candidate K '
If there is not, the process proceeds to step S11, and the code code (ω)
Is output and the character string ωK is registered in the dictionary address n. Further, the letter K is substituted into the initial character string ω, the address n is incremented by 1, and the process returns to step S2 after the check in step S5. If it is determined in step S5 that the data is the final data, the process proceeds to step S12 to output the last code code (ω) and end the series of processes.

FIG. 7 is a flow chart showing the LZW encoding of the present invention when the approximation judgment is performed from the input character shown in FIGS. 4B and 5B and the characters before and after the input character. In FIG. 7, first, in step S1, the dictionary is initialized to include the first character, and the head address n of the dictionary is set to n = 256 assuming that the character data is 8 bits. Then, the first character K is input, and the input character K is used as an index (initial character string) ω.

Next, in step S2, the next character K ₁ is input. Succeedingly, in a step S3, it is checked whether or not the character string ωK ₁ exists in the dictionary. If it exists, the process proceeds to step S4, and the character string ωK is replaced with a new initial character string ω, and the character K ₁ is replaced with the input first character K ₀ . Then, after checking the end of data in step S5, the process returns to step S2 to continue the search for the matching longest character string.

On the other hand, in step S3, the character string ωK ₁
Does not exist in the dictionary and the search for the longest character string is completed, the process proceeds to step S6, and an approximate search is performed after step S6. Step S6, reads the next character K _2, in step S7, for example, 4 (b) and 5 characters from the approximate conditions shown in _{(b) K 0, K 1} , inputs watches K ₂ characters K ₁ It is determined whether or not an approximation candidate K'similar to is in the approximate information table created in advance. If it is not in the table, the process proceeds to step S12 and dictionary registration is performed.

If there is an approximate candidate K'in step S7, the process proceeds to step S8 to read the approximate candidate K ', and in step S9 it is checked whether or not the approximate character string ωK' exists in the dictionary. If it exists in the dictionary, it is assumed that it can be approximated, and the approximate character string ωK ′ is set to the initial character string ω and the approximate candidate K ′ is set to the input first character K _{0 in} step S10.

Then, in step S11, when the run length correction is necessary for the immediately preceding approximation as shown in FIG. 5B, the run length value of the character K ₂ is corrected to K ₂ '. After that, the correction character K ₂ 'is set as the input character K _{1 to be} processed next, and the process proceeds to step S3 to continue the dictionary search. Also,
If the approximate character string ωK 'does not exist in the dictionary in step S9, the process returns to step S7 to search for a new approximate candidate K'. Finally, when there is no approximate candidate K ′, the process proceeds to step S12, the code code (ω) is output, and then the character string ω
Register K ₁ in dictionary address n. Further, the character K ₁ is substituted for the initial character string ω, the dictionary address n is incremented by 1, and the process returns to step S2 via step S5. When it is determined in step S5 that the data is the final data, the process proceeds to step S13, and the last code code
(Ω) is output and a series of processing is ended.

The approximation conditions in the approximate dictionary search according to the present invention are not limited to those shown in FIGS. 4 and 5, and appropriate similar conditions can be set. Although 8-bit data is taken as an example of preprocessed data that is a set of a pattern and a run length obtained by preprocessing, the bit butterfly of the preprocessed data may have an arbitrary bit length suitable for the device that performs the process. it can.

[0051]

As described above, according to the present invention, the universal encoding of the data, which is the set of the pattern and the run length obtained by the pre-processing of the image data, takes into account the similarity of the data and the dictionary. By performing the approximate search, it is possible to further improve the compression rate when the image data is encoded by the universal code such as LZW and compressed.

By performing an approximate search on a similar range that does not damage the original image, an image similar to the original image data can be reproduced when the code data is restored.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of the principle of the present invention.

FIG. 2 is an explanatory view of the operation of the present invention.

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

FIG. 4 is an explanatory diagram showing approximation conditions of the present invention in which the run lengths are the same and the patterns are similar.

FIG. 5 is an explanatory diagram showing the approximation conditions of the present invention in which the patterns are the same and the run lengths are similar.

FIG. 6 is an LZW of the present invention in which an approximate judgment is made using only input data.
Flowchart showing encoding

FIG. 7 is an L of the present invention for making an approximate determination including front and back data.
Float chart showing ZW encoding

FIG. 8 is an explanatory diagram of preprocessing and preprocessing data for conventional black and white binary data.

9 is an explanatory diagram showing a tree structure of a dictionary created by universal encoding of the preprocessed data of FIG.

FIG. 10 is a tree structure diagram of a dictionary in conventional LZW encoding.

FIG. 11 is an explanatory diagram showing a relationship between a character string and a code in conventional LZW encoding.

FIG. 12 is a flowchart showing conventional LZW encoding.

FIG. 13 is a flowchart showing conventional LZW decoding.

FIG. 14 is an explanatory diagram showing a specific example of conventional LZW encoding.

FIG. 15 is an explanatory diagram specifically showing a dictionary configuration created by conventional LZW encoding.

FIG. 16 is an explanatory diagram showing a specific example of conventional LZW decoding.

[Explanation of symbols]

 10: Preprocessing means (preprocessing software) 12: Dictionary 14: Dictionary search means (dictionary search software) 16: Approximate dictionary search means (approximate dictionary software) 18: Encoding means (encoding software) 20: Dictionary registration means ( Dictionary registration software 22: Table (approximation information table) 24: CPU 26: Program memory 28: Control software 30: Data memory 32: Data buffer

 ─────────────────────────────────────────────────── --- Continuation of the front page (72) Inventor Hirotaka Chiba 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Fujitsu Limited

Claims (6)

[Claims] 1. A pre-processing means (10) for converting image data into a character composed of a pattern and a run length in an image data compression method which applies universal coding to two-dimensional information of an image. A dictionary search means (14) for searching the dictionary (12) for a character string having the maximum length that matches the character string including the characters input from the preprocessing means (10), and a dictionary search by the dictionary search means (14) When there is no character string matching the character string ωK including the input character K in the dictionary (12), the approximate character K that can be approximated to the input character K
′ Is read and the character string ωK ′ including the approximate character K ′ is read again.
Is searched for a character string having a maximum length match, and if there is no same character string, the search is stopped, and if there is the same character string, an approximate dictionary search means (16) that continues the dictionary search using the approximate character string ωK ′ And the dictionary search means (14) or the approximate dictionary search means (1
4) an encoding means (18) for encoding the character string ω retrieved by the reference number in the dictionary (12); and a character string matching the character string ωK including the input character K from the dictionary (12). When the search is not possible and there is no approximate character K ′, the character string ωK obtained by adding the input character K to the reference number of the character string ω encoded immediately before is added to the dictionary (12) with a new reference number. An image data compression method characterized by comprising a dictionary registration means (20) for performing. 2. The image data compression method according to claim 1, wherein the approximate dictionary search means (16) has a pattern / length set that is close to the input character K composed of a pattern / length set. An image data compression method comprising a table (22) in which the approximate character K'is determined in advance, and the table (22) is referred to only by the input character K to read the corresponding approximate character K '. 3. The image data compression method according to claim 1, wherein the approximate dictionary search means (16) preliminarily obtains an approximate character K'from the input character K ₁ and the input characters K ₀ and K ₂ before and after the input character K ₁ . A predetermined table (22) is provided, and the table (22) is referred to by the input character K ₁ and the input characters K ₀ and K ₂ before and after it, and the approximate character K ′ of the input character K currently being processed is read out. An image data compression method characterized in that 4. The image data compression method according to claim 2 or 3, wherein the approximate dictionary search means (16) reads a character having the same run length and a similar pattern as an approximate character. Image data compression method. 5. The image data compression method according to claim 2 or 3, wherein the approximate dictionary search means (16) reads a character string having the same pattern and a similar run length as an approximate character. Characteristic image data compression method. 6. The image data compression method according to claim 5, wherein the approximate dictionary search means (16) has the same pattern from the input character K ₁ and the input characters K ₀ and K ₂ before and after the input character K ₁ . When reading the similar character K ', which is similar to
An image data compression method, characterized in that the run length of the subsequent input character K ₂ is corrected so as to correct the run length mismatch between the input character K ₁ and the approximate character K ′.