JP2001044850A - Data compression method, data decoding method and information processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data compression method that can freely decode only the required characters at a high compression rate and fast decoding processing. SOLUTION: In this data compression method, data are compressed by using a dictionary registering registration data streams in relation to a registration number to replace two combinations or more of data streams with the registration number. In steps A1, A2, a dictionary is generated and updated, and in a step A3, whether an optimum dictionary is obtained is discriminated. The update of the dictionary is conducted by an incremental separation method or an increased separation method or the like. As shown in steps A4, A5, when the optimum dictionary is generated, the dictionary is outputted for a final decoding static dictionary, data are compressed by using a static dictionary, and compressed data streams are outputted as final decoding compressed data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression method performed by using a dictionary in which a registration data string is registered in association with a registration number, a data restoration method for restoring compressed data, and an information processing apparatus.

[0002]

2. Description of the Related Art Recently, there has been a demand for information processing devices such as printers to increase the added value by supplying bitmap fonts and outline fonts of various print sizes, and to maintain high print quality at any print size. Is growing. Therefore, recently in the field of printers and the like,
Attention has been paid to a data compression technique for efficiently storing a large amount of information composed of these font data.

[0003] As a conventional technique of data compression for storing or transferring a large amount of information with a capacity as small as possible, a technique of converting fixed-length bit data into a variable-length bit code such as Huffman code, and a so-called Lempel- Z
Compression is performed by using the coincidence between a data sequence appearing in the past and a data sequence to be compressed, as in the iv patent (US Pat. No. 4,446,650) and the LZW patent (US Pat. No. 4,583,902). Techniques are known.

However, these conventional techniques are data compression methods performed using a so-called dynamic dictionary. That is, this method is a method of analyzing target data to be compressed, registering the data in the form of a dictionary structure while checking the appearance frequency and the like, and simultaneously compressing the data using this dictionary. In this case, the dictionary is characterized in that it changes rapidly in real time.
In this data compression using the dynamic dictionary, only the target data subjected to the compression processing remains as a product, but when restoring the data, the characteristics of the history data at the time of compression are examined, and the next dictionary is created while recreating the dictionary. You have to restore the data and go. Therefore, there is a problem that the processing must be performed in order from the beginning of the compressed target data.

[0005] In the compression of font data in a printer or the like, the compressed font data is stored in a storage device of the printer or the host computer. Then, at the time of printing, necessary compressed font data is taken out of the storage device, and the compressed font data is restored to normal data to form print data. Therefore, when data compression is used under such conditions, what is needed is how to quickly restore the data stored in the storage device and form print data.

Another important point when compressing font data in a printer or the like is to have the flexibility of outputting any print data (characters) in any order. That is, it is required that a printer or the like can randomly access stored data and form print data at random.

[0007]

SUMMARY OF THE INVENTION Lempel-Ziv or L
Data compression methods such as ZW are expected to become the mainstream of compression technology in the future. However, on the other hand, since these compression techniques are data compression using dynamic dictionaries, when restoring data, it is necessary to check past data characteristics and update, and it takes time to restore. There's a problem.

[0008] In data compression methods such as Lempel-Ziv and LZW, when decompression, data must be decompressed sequentially from the beginning of the compressed data, and only necessary print data can be freely extracted when necessary. There is a problem that can not be.

Further, in a data compression method called an incremental decomposition method used in LZW or the like, the number of data strings that can be registered in the dictionary cannot be increased by only one. Has a problem that the data compression ratio cannot be increased.

The present invention has been made in order to solve the above-mentioned problems, and the object thereof is to solve the problem described in Le.
A data compression method and compression that maintains a high compression ratio like the data compression method of mpel-Ziv and LZW, but still requires a relatively short time for the decompression process, and allows only necessary data strings to be freely decompressed. It is an object of the present invention to provide a data restoration method and an information processing device capable of restoring a data string that has been set.

Another object of the present invention is to create a method having a higher compression ratio than the incremental decomposition method, so that a compressed data sequence and a dictionary generated at that time are stored. It is an object of the present invention to provide a data compression method capable of reducing a storage capacity, a data restoration method capable of restoring a compressed data string, and an information processing apparatus.

[0012]

In order to solve the above problems, the present invention uses a dictionary capable of registering a registration data string in association with a registration number, and replaces two or more combinations of data strings with the registration number. In the data compression method of performing data compression in, the dictionary is updated until a dictionary optimal for data compression of a data string to be compressed is generated, and when the optimal dictionary is generated, the dictionary is finalized. Outputting as a static dictionary for decompression, performing data compression of a data string to be compressed by the static dictionary, and outputting the compressed data string as final compressed data for decompression. .

According to the present invention, a registration data string is registered in the dictionary in association with the registration number. Then, data compression is performed by replacing two or more combinations of data strings with the registration number. If data compression is performed in this manner, the original data string can be restored by reading out the registered data string using the registration number at the time of restoration. In this case, the dictionary is updated until an optimal dictionary for data compression of the data string to be compressed is generated.
That is, the dictionary is updated until, for example, the data amount of the compressed data, the data amount of the dictionary, and the like are optimized. Then, when the optimum dictionary is generated, the dictionary is output as a final static dictionary for restoration. Further, data compression of a data string to be compressed is performed by the static dictionary,
The compressed data string is output as final decompressed compressed data. And the final static dictionary output,
The compressed data is stored in, for example, a storage device, a storage medium, or the like, and is restored by an information processing device such as a printer or a computer, so that the original data string is restored. As described above, according to the present invention, the dictionary is updated until the optimum dictionary is generated, and the data compression is performed by using the optimum dictionary as a static dictionary. Therefore, it is possible to optimize the amount of output static dictionary and compressed data. Furthermore, since the dictionary until output is not required to be a static dictionary, it is possible to update the dictionary and perform data compression by a data compression algorithm using a dynamic dictionary having a very high data compression ratio, for example. Become. This makes it possible to greatly increase the data compression rate of the final compressed data. On the other hand, since the output dictionary is a static dictionary, a necessary data string can be freely restored using this static dictionary.

Further, according to the present invention, the updating of the dictionary until the generation of the optimal dictionary is performed in such a manner that a combination of data strings having a large number of combinations is registered preferentially. It is characterized in that registration of a registration data string is performed by deleting until the number of registrations in the dictionary reaches a predetermined number.

According to the present invention, a dictionary is generated by preferentially registering combinations of data strings having a large number of combinations. As a method for generating such a dictionary, for example, a method called a slide dictionary can be used. When this slide dictionary technique is used, the longest match data string between the past data string and the target data string is searched for by the slide dictionary technique, and the longest match data string is registered in the dictionary. A dictionary will be generated. This makes it possible to generate a dictionary in which combinations of data strings having a large number of combinations are registered with priority. By using such a dictionary, a combination of data strings having a large number of combinations is preferentially replaced with a registration number of the dictionary, so that the data compression ratio can be optimized. On the other hand, the dictionary generated in this way may have a very large number of registrations. Therefore, if the dictionary is updated by deleting the registration of the infrequently used registration data string from the generated dictionary, and the update is completed when the number of registered dictionary has reached the predetermined number, the compression rate is good. ,
In other words, it is possible to generate an optimal dictionary with a small amount of data.

Further, according to the present invention, the updating of the dictionary until the generation of the optimum dictionary is performed by first registering a combination of data strings having a high appearance probability from a dictionary generated in a low frequency. It is characterized in that registration of a registration data string is performed by deleting until the number of registrations in the dictionary reaches a predetermined number.

According to the present invention, a dictionary is generated by preferentially registering a combination of data strings having a high appearance probability. The appearance probability of this combination of data strings can be determined, for example, by examining the appearance probabilities of all data strings to be compressed. Then, by using the dictionary generated in this manner, a combination of data strings having a high appearance probability is preferentially replaced with a registration number, so that the data compression ratio can be optimized. Then, the dictionary is updated by deleting the registration of the infrequently used registration data string from the generated dictionary. If the update is completed when the number of dictionary registrations reaches a predetermined number, the optimal data with a small data amount is obtained. It is possible to generate a simple dictionary.

Further, according to the present invention, the updating of the dictionary until the generation of the optimum dictionary becomes a compression target while updating the dictionary by a data compression algorithm that dynamically changes during data compression. Performs data compression processing on all data strings, performs data compression processing on all data strings to be compressed while updating the dictionary again using the data compression algorithm using the dictionary updated by the processing, and performs data compression. The processing is performed by repeating the above processing until the compression ratio becomes optimal.

According to the present invention, data compression processing for all data strings to be compressed while updating the dictionary with a data compression algorithm that dynamically changes the dictionary during data compression, for example, an incremental decomposition algorithm, an incremental decomposition algorithm, or the like. Is performed. Then, data compression processing is performed on all the data strings to be compressed while updating the dictionary again by the data compression algorithm using the dictionary updated by this processing. Then, by repeating this process until the data compression becomes optimal, an optimal dictionary is generated. According to the present invention, data compression is performed by a data compression algorithm using a dynamic dictionary having a high data compression rate, and updating of the dictionary is completed when the data compression rate is optimal. It becomes possible. On the other hand, since the output dictionary is a static dictionary, a necessary data string can be freely restored using this static dictionary.

Further, the present invention is characterized in that it is determined whether or not an optimal dictionary has been generated based on the number of times of output of compressed data which increases when compressed data is to be output by the data compression algorithm.

The present invention also relates to a data compression method for compressing data by using a dictionary capable of registering a registration data string in association with a registration number and replacing two or more combinations of data strings with the registration number. A) a step of extracting a predetermined number of data strings from a data string to be compressed and storing the data strings in a work area having a predetermined number of buffers; and (B) a combination of data strings stored in adjacent buffers in the work area. It analyzes whether or not it is registered in the dictionary, and if it is registered in the dictionary, replaces the combination of the data string with the registration number in the dictionary and works on the data string so as to fill the empty buffer generated by the replacement. Shifts the data in the area, fetches the subsequent data string into the empty buffer generated at the end of the work area, and returns A step of analyzing whether a combination of data strings stored in the buffer is registered in the dictionary; and (C) a combination of data strings stored in adjacent buffers in the work area by the analysis of the step (B). Are not registered in the dictionary, the combination of the data strings stored in the first and second buffers from the head in the work area is registered in the dictionary and the first data string is registered. Erasing the data sequence in the work area so as to fill the empty buffer generated by the erasure, and capturing the subsequent data sequence into the empty buffer generated at the end of the work area as a result. The steps (B) and (C) are repeated until all the data strings are stored in the work area.

According to the present invention, it is analyzed whether or not a combination of data strings stored in adjacent buffers in the work area is registered in the dictionary, and if so, replaced with the registration number of the dictionary. Can be Then, the subsequent data string is taken into the resulting free buffer, and it is analyzed again whether or not the combination of the data strings is registered in the dictionary. If it is determined that the combination of the data strings is not registered in the dictionary, the first and second data string combinations from the top are registered, and the subsequent data string is taken into the resulting empty buffer. And
Data compression is performed by repeating these processes until all data strings are stored in the work area. As described above, according to the present invention, a work area having a predetermined capacity is provided, and when the same data string as the data string of interest exists in the work area, replacement processing using a dictionary is performed. I have. Therefore, especially when the same data string is subjected to compression processing, the number of dictionary registrations can be greatly reduced as compared with the conventional incremental decomposition method, and compression is performed by replacing with the registration number of the dictionary. The compressed data itself can have a much smaller data amount than the conventional incremental decomposition method.

Further, the present invention provides a method in which the repetition of the steps (B) and (C) until processing is performed on all data strings to be compressed is defined as one pass.
A data compression method for compressing data in a current pass using a dictionary updated in a previous pass, wherein the number of repetitions of the steps (B) and (C) in the current pass is equal to or less than the number of repetitions in the previous pass In the case of (1), the process proceeds to the next pass, and the steps (B) and (C) in the current pass are performed.
If the number of repetitions is larger than the number of repetitions in the previous pass, the dictionary updated in the previous pass is output as a final static dictionary for restoration, and the data of one pass is output by the static dictionary. It is characterized in that compression is performed, and a compressed data string is output as compressed data for final decompression.

According to the present invention, it is determined whether or not the data compression ratio has been optimized by comparing the number of repetitions of processing in the current pass with the number of repetitions of processing in the previous pass. Then, at the stage when the data compression ratio becomes optimal, the dictionary is set as a final static dictionary, and the static dictionary and data compressed by the static dictionary are output. Therefore, the data compression becomes possible by the additive decomposition algorithm having a very high compression ratio, and the data compression ratio is greatly increased. In addition, since the output dictionary is a static dictionary, a necessary data string is used by using this static dictionary. It is also possible to restore freely.

Further, according to the present invention, the dictionary includes a registration number,
The use frequency information is stored together with the registered data string, and the method includes a step of sequentially deleting the registered data string with a low use frequency.

According to the present invention, there is included a step of sequentially deleting the registration of the registration data string that is used less frequently. Therefore, when the number of dictionaries that can be registered is limited, the data amount of the dictionary can be set to an optimal size.

According to the present invention, the deletion of the registered data string having a low frequency of use sequentially reduces the frequency of use of the registered data string registered in the dictionary, and the frequency of use is initially reduced to a predetermined number or less. It is characterized in that it is performed by preferentially deleting the registered data sequence that has become lost.

According to the present invention, the frequency of use of the registered data sequence registered in the dictionary is sequentially reduced, and the registered data sequence whose frequency of use becomes equal to or less than a predetermined number is first deleted with priority. . This makes it possible to set the data size of the dictionary to an optimal size when the number of dictionaries that can be registered is limited. In addition, since the registered data strings whose use frequency has become equal to or less than the predetermined number at the beginning are preferentially deleted, the registered data strings frequently used can be left in the dictionary, and an optimal dictionary can be generated.

Further, the present invention is characterized in that the data string to be compressed is font data necessary for printing characters.

According to the present invention, font data necessary for character printing is to be compressed. Examples of such font data include bitmap font data, outline font data, and the like. In the case of compressing bitmap font data, for example, dot data of a predetermined number of units (for example, 1 byte unit and 1 word unit) arranged in the vertical and horizontal directions can be compressed. When compressing outline font data,
For example, attribute information of each point constituting a character outline, information for controlling vector coordinates of each point, and the like can be compression targets.

Further, the present invention is characterized in that only a part of the font data becomes the data string to be compressed, and the other part is data compressed by another data compression method.

According to the present invention, part of data is compressed by a data compression method using a static dictionary, an incremental decomposition algorithm, an incremental decomposition algorithm, or the like in accordance with the characteristics of data constituting font data. Some are compressed by other compression methods, such as Huffman coding. As described above, by changing the applied compression method according to the characteristics of the data, it is possible to further increase the data compression ratio.

Further, according to the present invention, the information representing the attribute of each point of the outline font and the information representing the characteristics of the character become the data string to be compressed, and the information representing the vector coordinates of the launch direction at each point. A data compression method characterized in that the data is compressed by another data compression method.

Further, the present invention is characterized in that data compression is performed on the characters having a common font using the common dictionary.

According to the present invention, data compression is performed on characters having a common font using a common dictionary. For example, for all Mincho fonts, the dictionary is updated and data compressed using a dictionary dedicated to Mincho font, and final static dictionaries and compressed data are obtained. For Gothic character strings, a dictionary dedicated to Gothic is used to update the dictionary and compress data to obtain final static dictionaries and compressed data. By sharing a dictionary for each of the fonts, data can be efficiently compressed.

Further, the present invention is characterized in that the data string to be compressed is a character string.

According to the present invention, character string data is to be compressed. As a result, for example, the capacity required for storing characters can be saved.

Further, the present invention provides a restoration-dedicated registration data string obtained by converting the registration data string into a restoration-dedicated data format from the registration number and the information of the registration data string contained in the finally generated dictionary. It is characterized in that a dictionary exclusively for restoration including the data length of the registration data string exclusively for restoration and information on the start address of the registration data string exclusively for restoration is generated.

According to the present invention, a restoration-only dictionary including a restoration-dedicated registration data sequence, a data length of the restoration-dedicated registration data sequence, and information on a start address of the restoration-dedicated registration data sequence is generated. Then, at the time of restoration, data restoration is performed using this restoration-dedicated dictionary. That is, data is read from the dictionary by reading the restoration-dedicated registration data string having the length designated by the data length from the position designated by the start address, and restoration processing is performed.
In this case, the restoration-dedicated registration data string is converted into a restoration-only data format. Therefore, the restoration process can be performed much faster than when a normal dictionary is used.

Further, the present invention uses a compressed data generated by any one of the above-described data compression methods and a final dictionary to execute a data string compressed by a decompression process according to the data compression method. Is restored.

According to the present invention, the original data string is restored using the compressed data generated by the data compression method and the final dictionary. As a result, predetermined processing, for example, processing such as printing of characters, can be performed using the restored data string.

Further, according to the present invention, using a compressed data generated by any of the above data compression methods and a final dictionary, a data string compressed by a decompression process according to the data compression method is provided. Is included.

According to the present invention, the original data string is restored by the restoration means using the compressed data generated by the data compression method and the final dictionary. The restoring means can be incorporated in an information processing device such as a computer and a printer.

[0044]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The preferred embodiments of the present invention will be described below. In the following first and second embodiments,
In order to simplify the description, a description will be given of a case where a character string is mainly compressed as a data string as an example. However, the data string in the present invention includes not only such a character string but also any kind of data string such as a byte string and a word string for constituting font data.

1. First Embodiment FIG. 1 is a flowchart for explaining a data compression method according to the present embodiment. In the data compression method of the present embodiment, a dictionary that can register a registration data string in association with a registration number is used. Then, data compression is performed by replacing two or more combinations of the data strings with this registration number. First, a dictionary is generated from all data strings (for example, character strings) to be compressed (step A).
1) The update of the dictionary is repeated until a dictionary optimal for data compression is generated (steps A2 and A3). And
When the optimal dictionary is generated, this dictionary is used as the final static dictionary for restoration, and the data string to be compressed is compressed by the static dictionary (step A4). Then, the compressed data for final decompression and the static dictionary for final decompression are output, and the mask R
It is stored in a storage device such as an OM or an EEPROM, or a storage medium. Then, using the static dictionary and the compressed data stored in the storage device or the like, data restoration processing is performed in an information processing device such as a printer or a host computer.

As a method of generating and updating the dictionary and determining at which stage the dictionary is optimal for data compression, various methods can be considered as described later.

FIG. 2 shows an example of the configuration of the data compression apparatus 12 using the data compression method of the present embodiment.
The entire data string 11 to be compressed is first input to the dictionary generating / updating means 13, which generates and updates the dictionary. Then, when a dictionary optimal for data compression is generated, the dictionary is held as a static dictionary 14 in the static dictionary holding unit 15. The stored optimal static dictionary 14 is output by the static dictionary output unit 16 to an external storage unit (not shown), while the data compression unit 1
7 is used to compress the entire data string 11. This data compression is performed by replacing a combination of two or more data strings with a registration number of a dictionary. Then, the compressed data 19 obtained as a result is output to the external storage means by the compressed data output means 18.

Next, the configuration of the dictionary in this embodiment will be described. For example, "static_string_dictionar
Consider a case in which a data string (character string) called “y” is compressed. In this case, the dictionary stores “st”, “at”, “ic”, “_”, “st”, “
Assume that a combination of eight data strings is registered, such as “ring_d”, “ic”, and “tionary.” In this case, the combination of these data strings is associated with a registration number as shown below, for example. 0: st 1: at 2: ic 3: _ 4: ring_d 5: tionary Then, “static_string_diction” is registered.
The data string “nary” can be replaced with 0, 1, 2, 3, 0, 4, 2, 5, and so on by these registration numbers, thereby compressing the data.

Next, various methods for generating an optimal dictionary will be described. In this embodiment, finally, two of the static dictionary and the compressed data are output, and these are stored in the storage device. Therefore, in order to save the used capacity of the storage device, it is necessary to reduce the data amount of the final static dictionary and the compressed data, and a dictionary capable of reducing the data amount can be called an optimal dictionary. That is, in order to obtain an optimal dictionary, it is desirable that the data amount of the dictionary itself can be reduced, or that the data amount of the compressed data can be reduced. For this purpose, this embodiment uses, for example, the following first to fourth methods. In the following, a character string will be described as an example of a data string.

(A) First Method for Obtaining an Optimal Dictionary In this method, a dictionary called a slide dictionary is generated by preferentially registering combinations of data strings having a large number of combinations. . Then, an optimal dictionary is obtained by sequentially deleting the generated registration number of the dictionary based on the use frequency information.

First, a technique called a slide dictionary will be described with reference to FIGS. In this method, all the character strings 41 to be compressed are sequentially sorted from the beginning to the target character strings 4.
As No. 3, it is stored in a memory space which is a work area. Then, the past character string 42 (the target character string 4
Before 3), it is checked whether there is a character string identical to the target character string 43. If the same character string has not been found in the past, the first character of the target character string 43 is
To the previous character string 42 (slide). On the other hand, if the same character string is found in the past, it is checked whether the next character string also matches, and by repeating this, the longest matching character string between the past character string 42 and the target character string 43 (character string (The one with the largest number of combinations).

For example, if the character string to be compressed is ABCA
Consider the case of BCDEF. In this case, first
Is the target character string 43, but since there is no same character string in the past, A is moved to the slide dictionary, which is the past character string 42 (see FIG. 3B). Then, B and C are moved to the slide dictionary because there is no same character string in the past (FIG. 3).
(C)). Then, A becomes the target character string 43. In this case, since A is in the slide dictionary, it is checked whether or not the next B matches (see FIG. 3D).
Then, in this case, since they match, it is checked whether or not C matches (see FIG. 3E). And then D
Are checked for a match, but D is not in the slide dictionary. Therefore, in this case, ABC is the longest matching character string.

Now, in a method called a slide dictionary,
By finding the longest matching character string ABC in this way, for example, the character string ABCCADEF is changed to "A",
"B", "C", "Same as 3 characters before 3", "D",
Compressed to “E”, “F” .Specifically, if there is no match, the character string is represented by “No match flag = 1 and its character code”. Data compression is performed by expressing the longest matching character string as “no match flag = 0 and matching location and matching length”.

However, in this embodiment, the slide dictionary technique is not used for actual data compression,
It is used only to find the longest matching character string in the string to be compressed. In this embodiment, when the longest matching character string is found, the longest matching character string is registered in the dictionary in association with the registration number. In the above example, ABC
Is registered in the dictionary. In this way, the longest matching character that appears twice or more in all the compression target character strings is found, and these are registered in the dictionary.
As a result, a character string having a large number of combinations is preferentially registered in the dictionary.

When using the slide dictionary method, it is necessary to store past character strings in a memory. However, as shown in FIG. 3A, the capacity of the memory is finite and the range that can be stored in the memory is finite. Therefore, in this case, only the past character strings 42 within the range that can be stored in the memory become the slide dictionary.

FIG. 4 is a flowchart for explaining a technique for obtaining an optimal dictionary by using the slide dictionary technique. As shown in FIG. 4, first, the longest matching character that appears two or more times in the character string to be compressed is found using the slide dictionary method, and a dictionary is generated by registering these characters in the dictionary (step B1). ). Next, using the generated dictionary, find a combination of the character string that matches the dictionary registered character string and the longest from all the compression target character strings,
When found, the frequency of use of the registered character string is increased by one (step B2). For example, if the character string to be compressed is A
In BCDEF, when the registered character strings are AB and ABC, the use frequency of the registered character string ABC is increased by one.

After the frequency of use is calculated in this way, the registered character strings with the least frequency of use, for example, 100
About registrations are deleted (step B3). in this case,
For example, the number that can be registered in the dictionary is 4096-256 = 38
If the number is 40, the registration is deleted so that the total number of registrations does not become less than 3840. Specifically, for example, if the total number of entries in the dictionary before deletion is 3900, only 60 are deleted.

Next, the total number of registrations is 3840 (predetermined number)
It is determined whether or not the number is below (step B4). If the number is more than 3840, the process returns to step B2, the usage frequency is calculated again, and about 100 registrations are deleted in step B3. Steps B2 to B4 are repeated in this way, and the number of registered dictionaries is gradually reduced. When the number of registrations reaches 3840, the process proceeds to step B5. As described above, for example, 3840 entries can be registered in the dictionary. Therefore, when the number of registrations reaches 3840, the dictionary is determined to be the optimal dictionary.

Finally, the optimal dictionary is used as a static dictionary, all the character strings to be compressed are compressed by the static dictionary, and the final static dictionary and compressed data are output (step B5, B6).

(B) Second Method for Obtaining an Optimal Dictionary In this method, a dictionary is generated by preferentially registering a combination of character strings having a high appearance probability. Then, an optimal dictionary is obtained by sequentially deleting the generated registration number of the dictionary based on the use frequency information.

First, the entire character string to be compressed is, for example, 1
The appearance probability of each character string is calculated by analyzing the number of times. Then, based on the appearance probability, the appearance probability of the combination of character strings is obtained.

For example, by the above analysis and calculation, a,
Assume that the appearance probabilities of b, c, and d are determined as follows: a; 50% b; 20% c; 10% d; Then, the appearance probability of the character string combination aa or the like is as follows: aa; 25% aaa; 12.5% ab; 10% ba; 10% ac; 5% ca; 5% aab; 5% aba; 5% baa; % Ad; calculated as 2.5% ‥‥‥. However, this is an expected value when it is assumed that there is no correlation between the appearance probabilities between the character strings.

After the appearance probability of a combination of character strings is obtained in this way, a dictionary is generated by preferentially registering the combination of character strings having a high appearance probability. Next, for example, by using the same method as in steps B2 to B4 in FIG. 4, the registration of a character string that is used less frequently is deleted until the number of dictionary registrations reaches a predetermined number, for example, 3840, and an optimal dictionary is generated. Then, by the same method as in steps B5 and B6, the final static dictionary and the compressed data compressed thereby are output. According to the above method, there is an advantage that the first analysis can be completed in one scan.

(C) Third Method for Obtaining an Optimal Dictionary In this method, an optimal dictionary is obtained using an incremental decomposition algorithm. For more information on the incremental decomposition algorithm, see
This will be described in comparison with an additive decomposition algorithm in a second embodiment described later.

FIG. 5 is a flowchart for explaining a technique for obtaining an optimal dictionary by using the incremental decomposition algorithm. In this method, first, a dictionary is initialized (step C1). Thus, the dictionary registration numbers 0 to 0
A registered character string is registered only in 255. Specifically, ASCII code characters are stored in 0 to 255. Next, NUM0 indicating the number of times of output of the compressed data is NU
M0 = 0 is set (step C2). Thereafter, data compression is performed on all the character strings to be compressed while updating the dictionary using the incremental decomposition algorithm, and the value of NUM0 is increased by one when compressed data is to be output (step C3). At this time, the compressed data itself is not output to the outside. This is because the process of step C3 is a process for updating the dictionary.

Next, NUM1 is set to NUM1 = 0 (step C4). Thereafter, the same processing as in step C3 is performed using the dictionary finally obtained in step C3. That is, data compression is performed on all character strings to be compressed while updating the dictionary using the incremental decomposition algorithm, and the value of NUM1 is incremented by one when compressed data is to be output (step C5). At this time, the compressed data itself is not output to the outside.

Next, it is determined whether or not NUM1> NUM0 (step C6). Thus, it is determined whether or not the data compression ratio has been optimized by the processing of step C5, that is, whether or not the dictionary has been optimized. And NUM1 ≦
If it is NUM0, it is determined that it is not yet the optimal dictionary, and NU
M0 = NUM1 (step C7) and step C
4. The processing of C5 is repeated. NUM0 and NUM1 represent the number of times of output of the compressed data, and a small number means that the data amount of the compressed data is also small. Therefore, NUM1 ≦ NUM0 means that the data compression ratio has been improved in the process of step C5. Therefore, in this case, the processes of steps C4 and C5 are further repeated. And step C
If NUM1> NUM0 at 6, it is determined that the dictionary is optimal for data compression.

Next, this optimal dictionary is defined as a static dictionary.
All character strings to be compressed are compressed by this static dictionary (step C8). However, the dictionary is not updated during this data compression. Finally, the final static dictionary and the compressed data compressed thereby are output (step C9).

FIG. 6 shows a visual representation of the above processing. The entire character string 21 to be compressed is first subjected to the first analysis, whereby a provisional dictionary 22 is created (this dictionary is a dynamic dictionary). Next, the entire character string 21 to be compressed is again subjected to the second analysis, and the dictionary 23 of the updated version 1 is updated.
Is created. Further, the dictionary 24 of the updated version 2 is created in the same manner. In this way, the dictionary is repeatedly updated, the compression ratio of the data is determined, and when an optimal dictionary is obtained, this is set as the static dictionary 25 of the definitive version. Then, the compression of all the compression target character strings 21 is performed again by the static dictionary 25.
However, at this time, the static dictionary 25 remains static,
Do not update the dictionary. Then, the static dictionary 25 and the compressed data are output to the outside and stored in the storage device.

The main differences between the methods shown in FIGS. 5 and 6 and LZW are as follows. That is, in LZW, after data compression, only compressed data is finally output as a product, and the dictionary is a dynamic dictionary and is not output. LZ
W is a data compression technique used for data communication through a telephone line. The transmitting side outputs compressed data by LZW, and the receiving side restores the compressed data. At the time of the restoration, it is necessary to analyze the characteristics of the compressed data and restore the data while re-creating the dynamic dictionary from the compressed data. Therefore, processing must be performed in order from the beginning of the compressed data, the restoration speed is slow, and it is not possible to randomly access and restore a necessary data string. Therefore, it is difficult to use data compressed by LZW as font data for a printer or the like.

On the other hand, in the methods shown in FIGS. 5 and 6, unlike LZW, two types of data, that is, compressed data and a static dictionary are finally output. Therefore, even when data is restored, there is no need to create a dictionary again, so that the restoration speed is high. Further, since a static dictionary is used, a desired data string in the compressed data can be accessed at random. The advantages that the restoration speed is high and that the data string can be accessed randomly are obtained not only in the third method, but also in the first and second methods described above and the fourth method described later. Is an advantage.

(D) Fourth Method for Obtaining an Optimal Dictionary In this method, an optimal dictionary is obtained using an additive decomposition algorithm. The additive decomposition algorithm will be described in detail in the second embodiment. However, the flow of the processing itself is substantially the same as that of FIG. 5 except that the data compression in steps C3 and C5 is performed by the additive decomposition algorithm.

In the third method, high-compression data can be obtained because data is compressed using the incremental decomposition algorithm. However, since there is a step of registering one character at a time, the processing speed is high. It is not very fast, and the compression ratio is not high compared to the additive decomposition algorithm. Therefore, when increasing the compression ratio, it is desirable to adopt the fourth method using this additive decomposition algorithm.

In the first to fourth methods,
The compressed data is output as a fixed bit length code.
This code is, for example, 12 bits or one word of 16 bits. And 16 bit 1
In the case of using words, there is an advantage that the processing speed is improved, but on the other hand, there is a disadvantage that the compression ratio is lower than in the case of 12 bits.

As described above, according to this embodiment, in order to create an optimal static dictionary, it is necessary to delete the dictionary registration based on the frequency of use or scan the data a plurality of times during compression. Processing time is longer. However, this is not a problem for the user at all. In other words, the compression process is necessary when data is written to, for example, a non-rewritable storage device or storage medium (ROM or the like), and this has no effect on the processing time at the time of decompression after compression. It is not. Specifically, when a user performs printing using a printer equipped with a storage medium storing data compressed by the compression method, it takes much time for the manufacturer to create the compressed data, This is because the speed of the restoration processing performed when the user wants to perform the processing does not become particularly slow.

2. Second Embodiment (A) Incremental Decomposition Method FIG. 7 is a diagram schematically illustrating a data compression method (hereinafter, referred to as an incremental decomposition method or an additive decomposition algorithm) used in the second embodiment. Is shown.

The character string data 1 to be compressed is first stored in the work buffer 2 as work storage means from the first part of the data. Next, a comparison is made as to whether or not the character strings registered in the dictionary 3 are present in the character strings in the work buffer 2. If the character strings are present, the information in the dictionary 3 is updated, and the character strings in the work buffer 2 are updated. The character string is replaced with the registration number, and the next target character string is added to the work buffer 2. On the other hand, if there is no registered character string, the first two characters in the working buffer 2 are registered in the dictionary 3,
The first character is output as compressed data 4. And
The next target character string is added to the work buffer 2, and a comparison is made again as to whether a character string registered in the dictionary 3 does not exist in the work buffer 2. Data compression is performed by repeating the above processing.

Next, a case where the following character string data is processed as an object to be compressed using the additive decomposition method to obtain a static dictionary and compressed data will be described in detail. Here, the character string data to be compressed is "ABCEBCHA, BBCCDBCH, ABCABE
HD, ABBCGABK ‥‥ ”. However, here, it is assumed that ASCII code character data is registered in numbers 0 to 255 of the dictionary composed of rewritable storage elements. It is also assumed that a buffer having a capacity of 16 bytes is prepared in the working storage unit.

(1) First, data of the capacity (16 bytes) of the working buffer is fetched from the beginning of the character string data to be compressed. Then, the character string in the working buffer becomes “ABCEBCHA, BBCCDBCH”. Here, each character is numerical data represented by one byte, and follows the ASCII code table as described above.
For example, "A" is 65, "B" is 66, and "C" is 67.

(2) Next, it is checked whether or not a character string (combination of character strings) of two or more characters already registered in the dictionary exists in the work buffer. However, since such a character string is not registered, for example, 25
Sixth, the first two characters "AB" are registered with a length of 2 (for characters) and 65 + 66 (ASCII code of "A" and "B") in the data column. The frequency of use is once. Furthermore, the first character "A" in the working buffer
Is output as compressed data, and is deleted on the work buffer. Then, all of the subsequent character strings are shifted in the leading direction so as to fill the resulting blank byte on the work buffer, and the resulting blank byte at the end of the work buffer is the target character. One character "A" is taken from the continuation of the column data. Then, the character string of the work buffer is “BCEBCHAB, B
CCDBCHA ".

(3) Next, it is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the work buffer. Then, the working buffer 7, 8
Since the character string "AB" exists in the character, replace the 2-byte character string "AB" with 256 characters and update the usage frequency of the "256" character string in the dictionary to twice. I do. As a result of this replacement, a blank space of 1 byte is created in the work buffer, so that all subsequent character strings are shifted toward the beginning so as to fill this space, and the resulting blank 1 byte at the end of the work buffer is filled. One character "B" is taken in from the continuation of the character string data. Then, the character string in the work buffer is “BCEBCH256 B, CCDBCH
AB ".

(4) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, since the character string "AB" exists at the 15th and 16th characters of the working buffer, the character string "AB" for 2 bytes is replaced with 256 characters, and "25" in the dictionary is changed to "25".
Update the frequency of use of the 6th string to 3 times. This replacement creates a 1-byte blank space in the work buffer, so all subsequent strings are shifted to the top to fill this space. One character “C” from the continuation of the target character string data is fetched into the one blank byte at the end of the resulting work buffer, and the character strings in the work buffer are “BCEBCH256 B, CCDBCH256”.
C ".

(5) It is again checked whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string has not been registered, for example, the first two characters "B" at the 257th position in the dictionary
C "is registered with a length of 2 (characters) and 66 + 67 in the data field, and the frequency of use of the number" 257 "is set to once. The registration number "66" of B "is output as compressed data, erased in the work buffer, and all subsequent character strings are shifted to the beginning so as to fill one byte of a blank space in the work buffer. The character string "A" is fetched from the continuation of the target character string data in the blank one byte at the end of the resulting work buffer, and the character string in the work buffer becomes "CEBCH256 BC, CDBCH256 CA". "
become.

(6) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, since the character string of "BC" exists in the third, fourth, seventh and eighth characters, and the eleventh and twelfth characters of the working buffer, the character string of a total of 6 bytes "
Replace "BC" with 257 characters each, and "257" in the dictionary
The use frequency of the number string is updated to four times. As a result of this replacement, a three-byte blank space is created in the work buffer, so that all subsequent character strings are shifted in the leading direction so as to fill the space, and the resulting blank three bytes at the end of the work buffer are not affected. "BEH" for three characters from the continuation of the character string data of Then, the character string in the work buffer is "CE257 H256257CD, 257H256C
ABEH ”.

(7) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, since the character string "AB" exists at the 13th and 14th characters of the working buffer, the character string "AB" of 2 bytes is replaced with 256 characters, and "25" of the dictionary is changed to "25".
Update the frequency of use of the 6th character string to 4. This replacement creates a 1-byte blank space in the working buffer, so all subsequent strings are shifted to the top to fill this space. The resulting blank byte at the end of the work buffer captures one character “D” from the continuation of the target character string data, and the character string in the work buffer is “CE257 H256257CD, 257H256 C25”.
6 EHD ”.

(8) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string is not registered, for example, the first two characters "C" at the 258th position in the dictionary
E "is registered as having a length of 2 (for characters) and 67 + 69 in the data field, and the frequency of use of the number" 258 "is set to once. The registration number "67" of C "is output as compressed data, erased in the work buffer, and all subsequent character strings are shifted in the leading direction so as to fill one byte of a blank space in the work buffer. After the shift, the one byte at the end of the resulting work buffer contains "A" for one character from the continuation of the target character string data, and the character string in the work buffer becomes "E257 H256257CD257, H256 C256". EH
DA ".

(9) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string is not registered, the first two characters "E
257 ", the length is 2 (characters), and 69 + 257 is registered in the data field, and the frequency of use of the number" 259 "is set to once. The registration number "69" of E "is output as compressed data, erased on the work buffer, and all subsequent character strings are shifted in the leading direction so as to fill the resulting one byte of blank space on the work buffer. The character string "B" for one character from the continuation of the target character string data is fetched into the blank one byte at the end of the resulting work buffer, and the character string in the work buffer becomes "257 H256 257 CD257, H256 C256". EHD
AB ".

(10) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, since the character string "AB" exists at the 15th and 16th characters of the working buffer, the character string "AB" of 2 bytes is replaced with 256 characters, and "2" of the dictionary is replaced with "2".
Update the frequency of use of the 56th string to 4 times. This replacement creates a 1-byte blank space in the working buffer, so all subsequent strings are shifted to the top to fill this space. One character “B” from the continuation of the target character string data is taken into the one blank byte at the end of the resulting work buffer, and the character string in the work buffer becomes “257 H256257CD257, H256 C256E”.
HD256 B ".

(11) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string has not been registered, the first two characters "2
57H ", the length is 2 (for characters), and 257 + 72 is registered in the data field, and the frequency of use of this" 260 "is set to once. 257 "is output as compressed data, erased in the work buffer, and all subsequent character strings are shifted to the beginning so as to fill one blank byte generated in the work buffer. One character from the continuation of the target character string data is "C" in the blank 1 byte at the end of the work buffer.
Take in. Then, the character string of the work buffer is "H25
6257CD257H, 256C256, EHD256BC ".

(12) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, the character string of "257H" exists at the 7th and 8th characters of the working buffer, and the character string of "BC" exists at the 15th and 16th characters. 257H "is replaced with 260 and" BC "is replaced with 257, and the usage frequency of" 260 "in the dictionary is updated twice and the usage frequency of" 257 "is updated five times. As a result of this replacement, a 2-byte blank space is created in the work buffer, so that all subsequent character strings are shifted toward the beginning so as to fill this space, and the resulting blank 2 bytes at the end of the work buffer are replaced. "GA" for two characters from the continuation of the character string data of. Then, the character string in the work buffer is "H256257
CD260 256C, 256 EHD256257GA ".

(13) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string has not been registered, for example, the first two characters at the 261st position in the dictionary "
H256 "is registered as having a length of 3 (for characters) and 72 + 256 in the data column, and the frequency of use of the number" 261 "is set to once. The registration number "72" of "H" is output as compressed data, erased in the work buffer, and all subsequent character strings are shifted in the leading direction so as to fill one byte of a blank space in the work buffer generated by this. The character string "B" is fetched from the continuation of the target character string data into the blank one byte at the end of the work buffer resulting from the shift.
B ".

(14) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. However, since such a character string has not been registered, the first two characters "2
56257 ", the length is 4 (for characters), and the data column is 256 + 257
And the frequency of use of "262" is once.
Furthermore, the first character "256" in the working buffer is output as compressed data, deleted on the working buffer, and the subsequent characters are filled so as to fill the resulting blank one byte on the working buffer. All columns are shifted toward the beginning, and the resulting one-byte blank space at the end of the working buffer contains one character "K" from the continuation of the target string data.
Take in. Then, the character string of the work buffer is "257
CD260256C256E, HD256257GABK ".

(15) It is checked again whether or not a character string of two or more characters already registered in the dictionary exists in the working buffer. Then, since the character string of "256257" exists in the 11th and 12th characters of the working buffer, the "256257"
To 262 and update the usage of "262" in the dictionary to twice. As a result of this replacement, a blank space of 1 byte is created in the work buffer, so that all subsequent character strings are shifted toward the beginning so as to fill this space, and the resulting blank 1 byte at the end of the work buffer is filled. One character is taken from the continuation of the character string data. Then, the character string in the working buffer is “257 CD260 256C256 E, HD
262 GABK. "

As described above, according to the additive decomposition method of this embodiment, data is sequentially fetched from the character string data to be compressed so that the length of the character string in the working buffer is always kept constant. Then, if a combination of adjacent character strings from the head in the work buffer is registered in the dictionary, the combination is replaced with the registration number of the dictionary to perform compression. Further, if this combination of character strings is included in the working buffer, this is also replaced with the registration number of the dictionary to further perform compression. Therefore,
The data amount of the compressed data is reduced as compared with the conventional incremental decomposition method, and as a result, the compressed data can be stored with a small storage capacity. If the combination of adjacent character strings from the head in the working buffer is not registered in the dictionary, the compressed data is output after the combination of the character strings is registered in the dictionary.

FIG. 8A schematically shows the above processing using a tree structure. In FIG. 8A, numbers are assigned according to (1) to (15) used in the above description. However, (1), (10), (12), (15)
Is not shown in FIG.

For example, the process (2) for registering “AB” in a dictionary and outputting “A” is expressed as follows in FIG. 8 (A). In other words, in FIG. 8A, "AB" is set at the end of the upward arrow to indicate that the dictionary is registered. In addition, the letter “A” at the end of the down arrow indicates that the character is output as compressed data. Further, by assigning a process number “2” near the up arrow and the down arrow, it indicates that the process is the process of the process (2).

When the character "AB" registered in the dictionary is found in the work buffer as in the step (3) and replaced with the registration number, the character string to be replaced is wrapped up. , And the number “3” of the step in which the process is performed.

In the present embodiment, compressed output data is as follows. "A, B, C, E, 257, H, 256, 257, C, ...
・ ”In addition, the following character strings are eventually registered in the dictionary. 0 to 255 ASCII code 256 AB 257 BC 258 CE 259 E257 260 257 H261 H256 262 256 257 ‥ ‥‥‥

(B) Incremental Decomposition Method Next, a case will be described in which the same character string data as in the above example is processed by a method using a conventional incremental decomposition method to obtain a dictionary and compressed data. However, here, it is assumed that ASCII code characters are registered in the dictionary from 0 to 255 from the beginning. "ABCECHA, BBCCDBCH,
ABCABEHD, ABBCGABK ‥‥ "

(1) The sent character string is not once stored in the working buffer by a fixed number of bytes at a time as in the case of the above-described additive decomposition method. Only the work buffer is loaded. Therefore, it is checked whether the first character "A" in the working buffer is registered in the dictionary. Then, since "A" is registered in the 65th place in the dictionary, the use frequency count number is set to one.

(2) Next, the second character "B" is added, and it is checked whether the character "AB" in the working buffer is registered in the dictionary. However, a character string already registered in the dictionary with two or more characters cannot be found. Therefore, the first two characters "AB" are registered as the 256th in the number, the length is 2 (for characters), and 65 + 66 is registered in the data column, and the use frequency of "256" in the dictionary is described as once. Next, the head "A" in the working buffer is output as compressed data, and the second character "B" of the target character string data is focused on.

(3) Check whether the new first character “B” is registered in the dictionary. Then, since "B" is registered in the 66th place in the dictionary, the count number of the frequency of use is set to one.

(4) Next, a new second character "C" is added, and it is checked whether the character "BC" in the working buffer is registered in the dictionary. However, a character string already registered in the dictionary with two or more characters cannot be found. Therefore, the first two characters "BC" are registered as the 257th number, the length is 2 (characters), and 66 + 67 is registered in the data column, and the usage frequency of "257" in the dictionary is described as once. next,
The head "B" in the working buffer is output as compressed data, and the next focus is on the third character "C" of the target character string data.

(5) The same operation is continued from the beginning to the fourth character E, and after “EB” is registered, is the new first character “B” in the working buffer registered in the dictionary? Find out if. However, since "B" is registered at the 66th position in the dictionary, its use frequency count number is set to 2 this time.

(6) Next, a new second character "C" is added, and it is checked whether or not the character string "BC" in the working buffer is registered in the dictionary. Then, this time, the character "BC" is registered at the "257" th position in the dictionary.
The frequency of use is described as twice, and it is checked whether the character string "BCH" to which the new third character is added is registered in the dictionary.

(7) Then, since this is not registered, this two-character string "BCH" is registered as the 260th of the number, the length is 3 (for character), and 257 + 72 in the data column. The frequency of using "260" in the dictionary is once. Next, the leading "B"
The registration number "257" of C "is output as compressed data, and attention is paid to the seventh character" H "of the target character string data.

FIG. 9 is a flowchart for explaining the operation of the above-mentioned incremental decomposition method.

The incremental decomposition method is performed in this manner, but the major difference from the incremental decomposition method of the present embodiment is that the character string data to be processed is fetched one character at a time from the beginning, and matching with the dictionary is performed. Is to go.

The above processing is schematically shown using a tree structure as in the example of FIG. 8A, as shown in FIG. 8B. However, the numbers 1 to 8 in the figure do not follow the numbers used in the description of the above steps. Here, for example, "
In a process of registering “AB” as a dictionary and outputting “A”,
A dictionary registration is represented by “AB” at the top of the up arrow, and a character is output as compressed data by “A” at the bottom of the down arrow.

Now, as understood from FIG. 8B,
In the conventional incremental decomposition method, the character "AB" registered in the dictionary is found in the middle of the working buffer other than the head as in the step (3) in FIG. No process is possible. The only possibility is to create a new character string by adding a new character to the right side of the character string registered in the dictionary and enclosing it on the right, as in step 5 in FIG. 8B. .

Incidentally, the output data compressed by the incremental decomposition method is as follows. "A, B, C, E, 257, H, 256, 257, C, D, 25
7, H, 256,... "In addition, the following character strings are eventually registered in the dictionary. 0 to 255 ASCII code 256 AB 257 BC 258 CE 259 EB 260 257 H 261 HA 262 256 B 263 257 C 264 CD 265 DB 266 260 A ‥ ‥‥‥

When FIG. 8A and FIG. 8B are compared,
In the range shown in the example, the output compressed data is the same, but the character strings registered in the dictionary are definitely different, and the difference increases as the character string of interest comes after the character string data to be compressed. Come.

As described above, the conventional incremental decomposition method can increase only one character at a time.
With respect to character string data in which the same character string continues as shown in (B), there is a great difference from the incremental decomposition method of the present embodiment, and the incremental decomposition method can store data with a smaller storage capacity. .

FIG. 10A shows a case where a continuous character string is compressed by the additive decomposition method of this embodiment.
(B) shows a case where a continuous character string is compressed by the incremental decomposition method. In the case of the incremental decomposition method shown in FIG. 10B, the dictionary contains A, AA, AAA, AAAAA, AAAAA, and AAA.
Because the registration is performed in the order of AAAA, ..., the letter of A
In order to compress 16 bytes, 16 outputs and 17 dictionary entries must be performed. On the other hand, in the case of the additive decomposition method of this embodiment in FIG. 10A, the dictionary contains A, AA, and AAA.
A, AAAAAAAAA, AAAAAAAAAAAAAAA
Since the registration is performed in the order of AA,..., It is only necessary to output the dictionary five times and perform the dictionary registration six times to compress the 16 bytes of the character A.

As described above, in the case of character string data in which the same character is continuous, the effect of the incremental decomposition method of this embodiment is clear, and the compression ratio is much higher than that of the conventional incremental decomposition method. This means, in other words, that a smaller number of output times is required to process the same amount of character string data. Since the character string at the time of output is always replaced with the fixed-length byte of the registration number of the dictionary and output, a small number of times of output indicates a small number of bytes, meaning that the compressed data is small.

(C) Generation of Optimal Dictionary The processing by the additive decomposition method described above is performed a plurality of times for all character strings to be compressed, as in FIG. An optimal dictionary that can be compressed well is generated. Hereinafter, the dictionary optimization processing will be described with reference to FIGS.

The flowchart of FIG. 11 shows the processing of one pass performed by the additive decomposition method to optimize the dictionary. Here, one pass means that all character strings are stored in the work buffer and processing is performed on all character strings. In FIG. 11, first, NUM = 0 is set (step E1). Next, it is checked whether or not a combination (that is, AB) of character strings stored in buf1 and buf2 (first and second in the work buffer) is registered in the dictionary (step E2). If it is determined that the character string has been registered, the combination of the character strings stored in the buf1 and buf2 is replaced with the registration number of the dictionary. It is stored (step E3). Thereafter, the process returns to step E2 to check again whether or not the combination of the character strings buf1 and buf2 is registered in the dictionary.

On the other hand, in step E2, buf1, buf1
If it is determined that the combination of the character strings 2 is not registered in the dictionary, it is checked whether or not the combination of the character strings buf2 and buf3 is registered in the dictionary (step E4). If it has been registered, the process proceeds to step E5, and thereafter returns to step E2. If not registered, move to the next step.

In this way, the combination of character strings buf15 and buf16 is checked (step E8).
If the combination of the character strings is not registered, it is checked whether or not the current pass is the last pass by the final pass flag (step E10). If it is not the last pass, the process proceeds to step E11, where buf1, buf1,
The combination of the character strings 2 is newly registered in the dictionary. Then, the character string stored in buf1 is erased, and the character string in the work buffer is shifted to the left, and the next character string is stored in buf16 so as to fill the resulting space. Further, NUM = NUM + 1 is set, and the value of NUM is increased by one, and thereafter, the process returns to step E2. On the other hand, if it is determined in step E10 that the current path is the last pass, buf
1 is output to the outside as compressed data, and buf1
Is erased and shifted one character to the left.
The next character string is put in the step (step E12).

In FIG. 11, for example, step E8,
After E9, the process returns to step E2. However, the present invention does not necessarily return to step E2. That is, in this case, the replaced character strings are buf15 and buf16.
, It is sufficient to return to step E6 where this affects the determination. This is the same in other steps such as step E6.

When the above processing is repeated and all the character strings to be compressed are stored in the work buffer and the first pass is completed, NUM2 = NUM is set as shown in FIG. 12 (step F2). Here, NUM corresponds to step E1 in FIG.
1 is incremented by one for each pass, and represents the number of times compressed data is output (actually, step E1
1, compressed data is not output to the outside until the final pass is reached).

Next, as shown in step F3, a second pass is performed. In the second pass, the dictionary updated in the first pass is used, and the dictionary is matured. The second pass is performed on all the compression target character strings by the processing shown in FIG. 11, as in the first pass. Next, in step F4, NUM> NUM2
Is determined. If NUM ≦ NUM2, it means that the number NUM of compressed data outputs in step F3 is smaller than the number NUM2 of compressed data outputs in step F1, which means that the data compression ratio has been improved. Therefore, in this case, the process returns to step F2,
The process is repeated until an optimal dictionary is obtained. When NUM> NUM2, it is determined that the optimal dictionary has been generated, the final path flag is set to 1 (step F5), and the final path compression operation is performed (step F6).

In the compression operation of the last pass, since the last pass flag = 1, the determination in step E10 in FIG. 11 is always in the YES direction, and the process always proceeds to step E12. That is, in this case, step E12
, The character string stored in buf1 is output to the outside as compressed data. In step E12, a new character string is not registered in the dictionary as in step E11, and the dictionary remains unchanged from the dictionary in the previous pass and is not updated. The dictionary that is not updated is output to the outside together with the compressed data as a static dictionary, and is stored in a storage device or the like.

(D) Frequency of Use As described above, in the process of creating an optimal dictionary, the number of dictionary registrations has reached the limit, and a new frequently used character string is to be registered in the dictionary. Also cannot be registered. In this case, the registered character string that is used less frequently is erased, and the character string to be registered is written in a vacant portion created by the erase.

FIG. 13A shows the structure of a dictionary including such usage frequency information, and FIG. 13B shows a flowchart of a registration deletion process based on this usage frequency.

Here, P represents the registration number of the dictionary to be newly registered, and the initial value of P is 256. Then, as shown in FIG. 13A, registration numbers 256 to 4095 are assigned to the registration locations of the dictionary.
The location of each registration number includes a location for entering the first character number, a location for entering the second character number, and a location for entering the frequency of use. The initial value of the usage frequency is all “0”,
Usage frequency = 0 indicates that no registered character string is registered at that location.

When newly registering a character string composed of a combination of the first character and the second character, as shown in step G1 of FIG. 13B, the location of P in the dictionary (FIG.
In (A), the character string of the first character and the second character is written in the 258th place. Next, P = P + 1 (step G
2) Look at the next place (259th place). in this case,
When P = 4096, P = 256 is set (steps G3 and G4). Next, the use frequency of the location P (location 259) is checked (step G5), and if the use frequency> 1, the use frequency of the location P is reduced by 1 (step G7). Then, returning to step G2, P = P +
Look at the next place (place 260) as 1.

For example, consider the case where there is no space at 256 to 4095 in the dictionary. In this case, steps G2 to G2 are performed until a location where the use frequency = 0 is generated.
Step 5 is repeated. And first, the use frequency = 0
Is deleted (step G6). When a new dictionary is registered, the dictionary is registered at the deleted location.

With the above processing, the registration numbers 256 to 40
95, it is possible to delete the least frequently used registered character string, and it is possible to newly register another character string at the location where this registration has been deleted. This makes it possible to generate a dictionary in which registered character strings that are frequently used are registered with priority.

(E) Restoration Processing The dictionary generated according to this embodiment has a tree structure as follows. (Dictionary) 256: 30 + 56 257: 80 + 16 258: 256 + 257 259: 256 + 257 260: 259 + 257 Therefore, this dictionary allows, for example, registration numbers 256 to 260.
Restoring the registered character string registered in (Character string restoration) 256 = 30 + 56 257 = 80 + 16 258 = 256 + 40 = 30 + 56 + 40 (from 256 = 30 + 56) 259 = 256 + 257 = 30 + 56 + 257 (from 256 = 30 + 56) = 30 + 26 + 80 + 16 (from 257 = 80 + 16) 260 = 259 + 257 + (259 = 257) = 256 + 257 + 257 (from 256 = 30 + 56) = 30 + 56 + 257 + 257 (from 257 = 80 + 16 = 30 + 56 + 80 + 16 + 257 = 30 + 56 + 80 + 16 + 80 + 16 (from 257 = 80 + 16)

In order to restore the character string compressed according to the present embodiment as described above, it is necessary to perform a decomposition process until all the character strings constituting the registered character string become smaller than 256. However, when the above-described decomposition processing is performed during restoration, there is a problem that the restoration speed of a character string is extremely slow. Therefore, in order to solve this problem, it is necessary to generate a dictionary dedicated for restoration as described below from the generated dictionary, store the dictionary dedicated for restoration in a storage medium such as a ROM, and use the dictionary for restoration. Is desirable.

FIG. 14A shows the structure of this restoration-only dictionary. A character string length (data length) and information on a character string start address are stored in the head portion of the restoration-dedicated dictionary in the order of registration numbers 256 to 4095.
Then, after that, the core part of the character string (a list of character strings)
Is stored. As described above, the core portion of this character string is generated by performing a decomposition process until all the character strings constituting the registered character string are smaller than 256, and these character strings are registered only for restoration. It becomes a character string.

When the restoration process is performed, as shown in FIG. 14B, the corresponding start address in the core portion of the character string is determined by the character string start address stored in the position of the registration numbers 256 to 4095. Is specified. Then, data is restored by extracting a character string (restore-dedicated registered character string) by the length specified by the character string length from the start address. In this case, since the core part of the character string has been decomposed in advance as described above, the speed of the restoration processing can be made much faster than that of a tree-structured dictionary.

FIG. 15 shows a flowchart of the restoration processing when the dictionary dedicated for restoration is used. First,
At step H2, a 12-bit code is read, and if the code is an end symbol, the process ends (step H2).
4) If it is not the end symbol, the process proceeds to step H5. Then, information on the character string start address and character string length to be restored from the dictionary is obtained according to the code number. Then, a character string corresponding to the character string length from the obtained character string start address is written in the restoration buffer (step H6). By repeating the above processing until an end symbol is detected, data can be restored.

[0135] 3. Third Embodiment A third embodiment is an embodiment in which so-called font data is compressed by the data compression method described in the first and second embodiments.

(A) Bitmap Font Data First, the case of compressing bitmap font data will be described. Bitmap font data is usually arranged in the raster direction. However, in the case of compressing bitmap font data according to the present embodiment, the compression ratio becomes considerably better when compressed in the vertical direction. Therefore, FIG.
As shown in (A), data is compressed in the vertical direction. However, of course, the data may be compressed in the horizontal direction.

As shown in FIG. 16A, the bitmap font data (40 dots × 40 dots) is configured by arranging five data blocks having a fixed length of 40 bytes. In this embodiment, the first and second data blocks are
Compression by the compression method described in the embodiment. Normally, data compression is performed in such a manner. However, when priority is given to the restoration speed over the compression ratio, the data compression is not performed in byte string units as shown in FIG. 16A but as shown in FIG. 2 side by side
It is also possible to perform data compression on a word string basis, with a byte as one word. As a result, the restoration speed can be increased, but on the other hand, there is a problem that the number of character strings that can be registered in the dictionary needs to be about 32768.

In FIG. 17, for example, the characters “Ni”
What is shown in the bitmap image is shown. FIG.
The part where "1" is written becomes black. Then, "0" is written in other portions, and this portion becomes white. As is clear from FIG.
The byte strings (data strings) of Nos. 2, 13, 30 to 39 are 00
h (hex display). Also, 10, 11, 14-1
The byte strings of Nos. 6, 27-29 are 01h. Also, 8
The byte sequence No. is 02h, and the byte sequences Nos. 9 and 17 to 26 are 03h. As described above, when the bitmap font data is compressed in the vertical direction, byte strings having the same value continue. Therefore, it is understood that the data is compressed with high efficiency by the data compression methods of the first and second embodiments.

(B) Outline Font Data An outline font is intended to represent the outline of a character by several points and straight lines and curves connecting the points. Normally sized characters (for example, 10 points) are printed using the above-described bitmap font, but some users may want to print at a large size such as 32 points. In such a case, it is difficult to prepare bitmap font data of each size in advance in relation to the data capacity. Therefore, in such a case, a basic outline font is provided for each character, the outline of the character is represented by the outline font, and the outline is enlarged by scaling. Convert to image data. This makes it possible to print characters of a desired size.

For example, as shown in FIG. 18, the outline of the character "2" in the outline font is represented by points A, B, C, D, E, F, G,. Can be represented. In the present embodiment, the outline font data is described in a data format called an NSI format.

In this embodiment, the outline font data described in this NS1 format is
As shown in FIG. 9 (A), three parts, namely, FLG part, D
It is compressed into an AT part and a VCT part.

Here, the VCT portion is information representing the vector coordinates in the launch direction at each point, and the X, Y
Contains vector coordinate information. Note that these X and Y vector coordinates do not represent absolute values of the coordinates, but represent relative values from the immediately preceding point.

The FLG portion is information indicating the attribute of each point, for example, "point A is a starting point", "point A and B
"Points are connected by a straight line", "Points B and D are connected by a curve (Bézé curve) with point C as an intermediate point", "Points D and E are connected by a straight line", "Points E and G The points are connected by a curve (Bézé curve) having the point F as an intermediate point ”. Furthermore, the DAT portion is information indicating the characteristics of the character, and includes information indicating the size of the character, information indicating the line width of the character, and the like. In this embodiment, the VCT part is compressed by the Huffman coding method. On the other hand, the FLG portion and the DAT portion have a certain degree of regularity such that the same data continues, and so are compressed by the data compression method described in the first and second embodiments.
By separating and compressing the components constituting the outline font data for each component in this manner, it is possible to greatly improve the compression ratio. Then, as shown in FIG. 19B, the structure of the compressed data is as follows: first, only the compressed data of the FLG part is arranged, then only the compressed data of the DAT part is arranged, and then the compressed data of the VCT part is arranged. Only the lines are arranged. In addition, FLG part, DAT
A dedicated restoration dictionary corresponding to each of the VCT portion and the VCT portion is prepared.

When decompressing this compressed data, FLG
Part, the DAT part, and then V
The CT part will be restored. FIG. 19C shows a flowchart of this restoration processing. First, in step I2, the FLG part is restored with a FLG static dictionary, and this is temporarily written in a buffer provided for the FLG. Next, in step I3, the DAT portion is restored with a DAT static dictionary, and this is temporarily written in a buffer provided for DAT. In this state, the pointer for reading data is a compressed V as shown in FIG.
This indicates the head of the CT part.

Next, as shown in step I4, DAT
After reading and outputting the portion and the FLG portion from the buffer according to the rules of the NSI format, the VCT portion is Huffman-coded according to bits 0 to 3 of the read FLG portion (the details of these bits will be described later). To restore. This makes it possible to obtain the original outline font data returned in the order of the NSI format.

In this embodiment, in order to further improve the data compression ratio, the FLG portion and the VCT portion are changed to a format that can be represented by the following coordinate point control unit and coordinate point definition unit, and then the compression is performed. It is carried out.

Here, bits 4 to 7 of the coordinate point control unit are the same as bits 4 to 7 of the original FLG part, and these bits 4 to 7 specify the attribute of each point.
In other words, the user designates a start point, an intermediate point, or an end point, or any of movement interpolation, linear interpolation, and curve interpolation. Bit 0 to bit 3 of the coordinate point control unit indicate that the value of the vector coordinates defined in the coordinate point definition unit has the following meaning. That is, bit 0 indicates that the X vector coordinate is a non-zero value and exists, and bit 1 indicates that the Y vector coordinate is a non-zero value and exists. Bit 2 indicates that the X vector coordinate is a negative value. Further, bit 3 indicates that the Y vector coordinate is a negative value.

A 10-bit code (0
To 1023). Then, by specifying these codes and bits 0 to 3 of the coordinate point control unit, vector coordinate values in the range of −1024 ≦ X vector coordinate ≦ 1024 −1024 ≦ Y vector coordinate ≦ 1024 can be represented.

For example, when bit 0 = 1, it indicates that the coordinate point definition section has X-1 data. Also, when bit 1 = 1, it indicates that the coordinate point definition unit has Y-1 data. If there are vertical bars, horizontal bars, etc. in the character, the X vector coordinates may be 0 or the Y vector coordinates may be 0. In this case, X
It is not necessary to output vector coordinates and Y vector coordinates as compressed data. Therefore, in this case, if the presence / absence of X vector coordinates and Y vector coordinates in the coordinate point definition unit can be designated by controlling bits 0 and 1 of the coordinate point control unit, the data compression rate can be further improved. Become. Also, the positive and negative designations of the X vector coordinates and the Y vector coordinates are designated by the bits 2 and 3 of the coordinate point control unit as described above, whereby the data compression ratio can be further improved.

As described above, according to this embodiment, bitmap font data and outline font data can be effectively compressed. At the time of this compression, the static dictionary described in the first and second embodiments is used. In this case, it is desirable that the static dictionary be a common static dictionary for characters having a common font. For example, for all Mincho fonts, the dictionary is updated and data compressed using a dictionary dedicated to Mincho font, and final static dictionaries and compressed data are obtained. In addition, for Gothic characters, the dictionary is updated and data compressed using a Gothic dedicated dictionary, and the final static dictionary,
Get compressed data. By sharing a dictionary for each of the fonts, data can be efficiently compressed. This is because, when the font is common, for example, the way of changing the outline of the character (such as how a line jumps up) and the thickness of the outline are common among the characters, the characteristic information of the character itself, This is because the attribute information of the points and the like become the same, and the data is easily compressed by the common dictionary. In this case, the output static dictionary and compressed data are classified for each font and stored in the storage device.

FIG. 20 shows a use mode of the data compression method of the present embodiment described above. The data string (font data) to be compressed is input to the data compression device 31,
Data is compressed by the data compression method of the present embodiment. Then, the generated static dictionary and compressed data are stored in a storage device (ROM or the like) in the data decompression means 32. The data restoring means 32 is built in an information processing device, for example, a printer 33 or a computer 34 (the restoring means may be divided and built in two devices).

In a conventional printer or the like, font data and the like are stored in a storage device without data compression. By managing storage addresses of individual compressed data, necessary font data is output at any time and characters are printed. I was
On the other hand, according to the present embodiment, the static dictionary and the compressed data are stored in the storage device, and the necessary addresses are output as needed by managing the storage addresses of the individual compressed data and the static character string dictionary. Becomes possible. Thus, the amount of data stored in the storage device can be reduced, and the cost of the device can be reduced. Such an advantage is particularly effective for kanji characters and Hangul characters that require a large storage capacity, but of course the effect is also great for characters in other languages.

The present invention is not limited to the above embodiment, and various modifications can be made within the scope of the present invention.

For example, in the present embodiment, processing of character string data has been mainly described, but application to other symbol data is also possible.

The data string compressed according to the present invention is not limited to a character string, a byte string, and a word string, but includes various data strings, such as a data string for forming a two-dimensional or three-dimensional figure. Can also be included.

The registration number in the present invention also includes information that performs substantially the same function as the registration number.

[0157]

According to the present invention, since the optimal dictionary is used as a static dictionary and data compression is performed by the static dictionary, the output static dictionary and the amount of compressed data can be optimized. can do. As a result, the capacity of the storage device and the storage medium in which the static dictionary and the compressed data are stored can be reduced, and the cost can be reduced. Further, for example, a data compression algorithm using a dynamic dictionary having a high data compression rate can be used, and the data compression rate can be greatly increased. In addition, it is possible to freely restore necessary data strings using this static dictionary. For example, when restoring this compressed data string by an information processing device or the like, it is possible to arbitrarily extract a desired data string. Becomes possible. Further, according to the present invention, there is a possibility that the time required for optimizing the dictionary and obtaining the final dictionary and compressed data may be long, but there is also an effect that the restoration processing speed itself does not become particularly slow.

Further, according to the present invention, a combination of data strings having a large number of combinations is preferentially replaced with a registration number in the dictionary, so that the data compression ratio can be optimized. On the other hand, the registration of the registered data string that is used less frequently is deleted, and the update ends when the number of registered data reaches the predetermined number. Therefore, it is possible to generate an optimal dictionary with a small data amount. There is also an advantage that a dictionary can be easily generated by a technique called a slide dictionary. Note that the predetermined number in this case can be a number such that when the number of bits of the output data is, for example, 12 bits, the number of registered dictionaries falls within the range of 12 bits.

Further, according to the present invention, a combination of data strings having a high appearance probability is preferentially replaced with a registration number, so that the data compression ratio can be optimized. On the other hand, the registration of the registered data string that is used less frequently is deleted, and the update ends when the number of registered data reaches the predetermined number. Therefore, it is possible to generate an optimal dictionary with a small data amount. Since the first dictionary can be generated by one analysis, there is an advantage that the processing can be greatly simplified.

According to the present invention, data compression is performed by a data compression algorithm using a dynamic dictionary having a high data compression rate, and updating of the dictionary is completed when the data compression rate is optimal. The rate can be greatly increased. On the other hand, since the output dictionary is a static dictionary, a necessary data string can be freely restored using this static dictionary.

Further, according to the present invention, especially when the same data string is subjected to compression processing, the number of dictionary registrations can be greatly reduced as compared with the conventional incremental decomposition method,
The compressed data itself that has been replaced with the registration number of the dictionary and has been compressed can have a much smaller data amount than the conventional incremental decomposition method.

Further, according to the present invention, the data compression rate can be greatly increased, and a required data string can be freely restored using this static dictionary. Another advantage is that it is possible to determine whether or not the optimum compression ratio has been reached only by the number of repetitions of the processing.

Further, according to the present invention, when the number of dictionaries that can be registered is limited, the data amount of the dictionary can be set to an optimum size by a very simple method.

Further, according to the present invention, the data amount of the dictionary can be set to an optimum size, and a registered data string frequently used can be left in the dictionary, so that an optimum dictionary can be generated.

Further, according to the present invention, bitmap font data and outline font data can be compressed, and the capacity of a storage device and a storage medium for storing these font data can be reduced. As a result, it is possible to reduce the cost of the storage device, the printer, the computer, and the like including the storage medium.

Further, according to the present invention, the compression method to be applied can be changed according to the characteristics of the data, and the data compression ratio can be further increased.

Further, according to the present invention, a dictionary can be shared for each of fonts such as Mincho and Gothic. For example, the manner of changing the outline of a character, the thickness of the outline, and the like are often the same for the same font, and therefore, according to the present invention, data can be more efficiently compressed.

Further, according to the present invention, the capacity required for storing characters can be saved.

Further, according to the present invention, the restoration-dedicated registration data string is converted into a restoration-dedicated data format. For example, in the incremental decomposition method and the incremental decomposition method, the registration data string is decomposed until the tree structure is eliminated. ing. Therefore, the restoration process can be performed much faster than when a normal dictionary is used.

Further, according to the present invention, predetermined processing, for example, processing such as printing of characters, can be performed using the restored data string.

According to the present invention, the original data string can be restored by the restoring means, and this restoring means can be built in an information processing apparatus such as a computer or a printer.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a data compression method according to a first embodiment.

FIG. 2 is a block diagram illustrating an example of a configuration of a data compression device 12 using the data compression method of the embodiment.

FIGS. 3A to 3E are schematic explanatory diagrams for explaining a technique of a slide dictionary.

FIG. 4 is a flowchart for explaining a method of obtaining an optimal dictionary using a slide dictionary method.

FIG. 5 is a flowchart for explaining a method of obtaining an optimal dictionary by using an incremental decomposition algorithm.

FIG. 6 is a diagram visually illustrating a process of creating a static dictionary and compressed data.

FIG. 7 is a schematic explanatory diagram for schematically explaining a data compression method used in the second embodiment.

FIGS. 8A and 8B are diagrams schematically showing a difference in processing between an incremental decomposition method and an additive decomposition method.

FIG. 9 is a flowchart for explaining the operation of the incremental decomposition method.

FIGS. 10A and 10B are diagrams schematically showing a difference in processing between an incremental decomposition method and an additive decomposition method.

FIG. 11 is a flowchart for explaining processing of one pass by an incremental decomposition method.

FIG. 12 is a flowchart illustrating a process of a plurality of passes for optimizing a dictionary.

FIG. 13A is a diagram illustrating a structure of a dictionary including usage frequency information, and FIG. 13B is a flowchart illustrating registration deletion processing based on usage frequency.

14A is a diagram illustrating a structure of a restoration-dedicated dictionary, and FIG. 14B is a diagram illustrating designation of a core portion of a character string by a character start address and a character string length. .

FIG. 15 is a flowchart of a restoration process when a dictionary dedicated for restoration is used.

FIGS. 16A and 16B are schematic explanatory diagrams for describing compression of bitmap data. FIG.

FIG. 17 is a schematic explanatory diagram for explaining a bitmap image of a character.

FIG. 18 is a schematic explanatory diagram for explaining an outline font.

FIGS. 19A and 19B are schematic explanatory diagrams for describing compression of outline font data, and FIG. 19C is a flowchart of a process for restoring the compressed data.

FIG. 20 illustrates a use mode of the data compression method according to the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Character string data to be compressed 2 Work buffer 3 Dictionary 4 Compressed data 5 Completed dictionary 11 All data strings to be compressed 12 Data compression device 13 Dictionary generating means 14 Static dictionary 15 Static dictionary holding means 16 Static Dictionary output means 17 Data compression means 18 Compressed data output means 19 Compressed data

Claims (14)

[Claims] 1. A data compression method for compressing data by using a dictionary capable of registering a registration data string in association with a registration number and replacing two or more combinations of data strings with the registration number. The dictionary is updated until an optimal dictionary for data compression of the data string is generated, and at the stage when the optimal dictionary is generated, the dictionary is output as a final static dictionary for restoration. A data compression method, comprising compressing a data string to be compressed and outputting the compressed data string as final decompressed compressed data. 2. The method according to claim 1, wherein the updating of the dictionary until the optimal dictionary is generated includes:
The method is characterized in that registration of infrequently used registration data strings is deleted from a dictionary generated by preferentially registering combinations of data strings having a large number of combinations until the number of registered dictionary entries reaches a predetermined number. Data compression method. 3. The method according to claim 1, wherein the updating of the dictionary until the optimal dictionary is generated includes:
It is characterized in that registration of infrequently used registered data strings is deleted from the dictionary generated by preferentially registering a combination of data strings with high appearance probabilities until the number of registered dictionaries reaches a predetermined number. Data compression method. 4. The method according to claim 1, wherein updating the dictionary until the optimal dictionary is generated includes:
At the time of data compression, the dictionary is updated by a data compression algorithm that changes dynamically, data compression processing is performed on all data strings to be compressed while the dictionary is updated, and the data is again updated using the dictionary updated by the processing. A data compression method comprising performing data compression processing on all data strings to be compressed while updating a dictionary by a compression algorithm, and repeating the processing until the data compression ratio is optimized. 5. The data according to claim 4, wherein it is determined whether or not an optimum dictionary has been generated based on the number of times of output of compressed data that increases when compressed data is to be output by the data compression algorithm. Compression method. 6. The method according to claim 1, further comprising the step of storing usage frequency information together with a registration number and a registration data string in the dictionary, and sequentially deleting the registration data strings having a low usage frequency. Data compression method. 7. The method according to claim 6, wherein the deletion of the registered data string having a low frequency of use sequentially reduces the frequency of use of the registered data string registered in the dictionary. A data compression method, which is performed by preferentially deleting a registered data string that has been changed. 8. The data compression method according to claim 1, wherein the data string to be compressed is font data required for printing characters. 9. The method according to claim 8, wherein only a part of the font data is a data string to be compressed, and another part is data compressed by another data compression method. Data compression method. 10. The outline font according to claim 9, wherein the information representing the attribute of each point of the outline font and the information representing the characteristics of the character are data strings to be compressed, and represent vector coordinates in the launch direction at each point. A data compression method, wherein information is compressed by another data compression method. 11. The data compression method according to claim 8, wherein data compression is performed on the characters having a common font using a common dictionary. 12. The data compression method according to claim 1, wherein the data string to be compressed is a character string. 13. A data string to be compressed by a decompression process according to the data compression method, using the compressed data generated by the data compression method according to claim 1 and a final dictionary. A data restoration method characterized by restoring data. 14. A data string to be compressed by a decompression process corresponding to the data compression method, using the compressed data generated by the data compression method according to claim 1 and a final dictionary. An information processing apparatus comprising means for restoring the information.