JPH06195063A - Character pattern data compressing device - Google Patents

Abstract

PURPOSE:To compress character pattern data with the use of a compressing method which suppresses redundancy by making the appearance frequency of data partial while making the combination pattern of one data constituting the character pattern data and last data correspond to fixed data for reproduc tion when the combination pattern appears frequently. CONSTITUTION:An uncompressed font pattern data storage part 101 aims at respective data constituting a character pattern and detects the frequency of the combination of the aimed data and its last data and a key word selection part 105 generates a key word table when the combination is high in frequency. A data appearance frequency counting part 107 makes this key word table correspond to command codes for reproduction to allow the appearance frequency of the data to be deviated. A compressed code table generation part 109 generates a compressed code string having its redundant part cut corresponding to the appearance frequency of the data. A compressive encoding part 112 substitutes the compressed code string for the character pattern data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character pattern data compression device for compressing character pattern data incorporated in a document creation device such as a word processor.

[0002]

2. Description of the Related Art Generally, in a word processor, input and created document data is stored in a code format in a text memory, and when a document print command is given, the document data in the text memory is referred to by referring to a character pattern memory. Is converted into a dot pattern character by character and printed out. Here, the character pattern memory is composed of a semiconductor read-only memory (ROM) or the like, and a document is printed with a character pattern of a typeface provided as standard in the character pattern memory.

[0003]

By the way, when the character pattern data is, for example, a 48-dot font, the number of dots contained in one character is 48 × 48 = 2304 dots, and one character requires 288 bytes. This is, for example, 7,000 characters or more for the JIS kanji and JIS kana symbols as a whole of JIS, and the capacity is very huge, 2 Mbytes. For this reason, if a plurality of types of character patterns having different typefaces are provided as standard equipment in the character pattern memory in advance, the character pattern memory becomes huge and the cost becomes high. Further, even in the case of supplying a character pattern from an external storage device such as a floppy disk, a plurality of external storage devices are required for each typeface, resulting in high cost. Therefore, conventionally, it is known that the character pattern data is compressed and stored, and the compressed pattern is restored to the original character pattern when the document is printed. In this type, for example, when the bit data of “0” forming the character pattern data is continuous, the character pattern is compressed by replacing the continuous number with the bit string data of “0” and storing it. However, with such a compression method, the compression rate is low, and the actual situation is that the effectiveness is not so high. The subject of this invention is
Redundancy is suppressed by associating this with fixed data for reproduction and biasing the appearance frequency of data when a combination pattern of certain data that constitutes character pattern data and the data before it frequently appears. That is, the character pattern data can be compressed by using the compression method.

[0004]

The means of the present invention are as follows. In a character pattern data compression device that compresses each character pattern data of the bit map system and incorporates it in the document creation device, the combination frequency detection means sequentially focuses on each data forming the character pattern data, and The operation of detecting the combination frequency is performed for all characters according to the combination with the data. When a combination having a high frequency is detected by the combination frequency detecting means, the table creating means creates a candidate table for storing, for each combination, a data candidate that the data of interest is this if the previous data is this combination. . The compression code generation means biases the appearance frequency of the data by associating the candidate table with the fixed data for reproduction, and generates the compression code string in which the redundant portion is suppressed according to the appearance frequency of the data. The compressing means compresses the character pattern data by replacing each character pattern data with the compression code string generated by the compression code generating means. It should be noted that, for example, when the combination frequency detecting means detects a combination having a high frequency, the table creating means creates a plurality of candidate tables having different focused data for one combination of the previous data, and compresses The code generating means may associate fixed data with each of the plurality of candidate tables. Further, for example, the unit of each data forming the character pattern data may be a bit group forming one word occupying one address when the data is held by the bitmap method.

[0005]

The operation of the means of the present invention is as follows. First, attention is sequentially paid to each data forming the character pattern data, and an operation of detecting the combination frequency is performed for all the characters in accordance with the combination of the focused data and the previous data. When a combination with a high frequency is detected in this way, a candidate table is created for storing, for each combination, a data candidate that the data of interest is this if the previous data is this combination. Then, by making this candidate table correspond to the fixed data for reproduction, the appearance frequency of the data is biased, and the compression code string in which the redundant part is suppressed is generated according to the appearance frequency of the data to compress each character pattern data. Character pattern data is compressed by replacing it with a code string. Therefore, when a combination pattern of a certain data forming the character pattern data and the data before it frequently appears, it is associated with the fixed data for reproduction and the frequency of appearance of the data is biased to achieve redundancy. Character pattern data can be compressed using a compression method that suppresses

[0006]

[First Embodiment] A first embodiment will be described below with reference to FIGS. FIG. 1 schematically shows a process of compressing / reproducing character pattern data. The character pattern data compressing device 100, when the product of the document creating device 200 is completed, converts the character pattern data of, for example, 13000 characters into a peculiar embodiment. Data is compressed according to the rules, and various data (compressed font data, index table, inverse conversion table, keyword table described later) obtained by the compression are incorporated in the document creating apparatus 200. That is, the character pattern data compression device 100 and the document creation device 200 are separate and independent devices. The character pattern data compression device 100 is a device for storing compressed character data in the document creation device 200. The compressed character data is written when the product is completed.

The document creating apparatus 200 has a character pattern reproducing apparatus 201 peculiar to this embodiment and a document creating machine main apparatus 202 which is usually provided. The character pattern reproducing apparatus 201 is provided with a compressed character data memory 203. Has been. The compressed character data memory 203 includes a compressed font storage unit 204, an index table storage unit 205, an inverse conversion table storage unit 206, and a keyword table storage unit 207.
The compressed font data, the index table, the inverse conversion table, and the keyword table generated and output from 00 correspond to the corresponding storage unit 204 in the compressed character data memory 203.
To 207. At the time of printing the document, the character pattern reproducing device 201 reversely converts the compressed character pattern according to the rules peculiar to the present embodiment, restores and reproduces the same uncompressed character pattern as usual, and gives it to the document creator main unit 202. As a result, the same printing result as usual is obtained.

The above-mentioned "compressed font data",
The “index table”, the “inverse conversion table”, and the “keyword table” will be described in detail later, and therefore will be briefly described here. "Compressed font data" biases the appearance frequency of data by associating frequently appearing data patterns of the data forming the character pattern data with a specific code (fixed value), and makes them redundant according to the appearance frequency of data. Data obtained by compression using a compression method having a characteristic of cutting a part, and normal character pattern data is compressed for each character according to the compression method described above, and stored in the compressed font in the document creation device 200. Compressed font data for all characters is written in the section 204. The "index table" is a table indicating which character is included in which byte of the "compressed font data". The "inverse conversion table" and the "keyword table" are tables used when the compressed font data is restored to normal character pattern data and reproduced.

FIG. 2 shows a character pattern data compression device 100.
It is a block configuration diagram of. The uncompressed font pattern data storage unit 101 stores all the characters (for example, 13000 characters) to be installed in the document creating apparatus 200.
FIG. 3 shows the data structure of character patterns stored in the uncompressed font pattern data storage unit 101. The character pattern consists of 48 x 48 dots per character, 8
Arrange the bit-structured memory vertically and set it horizontally 48
The data is stored in a dot-map manner in a memory having a structure in which bytes are arranged and further vertically stacked in six stages. Therefore,
A character pattern for one character has a structure of 288 bytes consisting of vertical 8 bits × 6 rows = 48 dots and horizontal 48 dots. In this case, the youngest address is the address corresponding to the upper left corner of the character pattern, the youngest is the address next to the right, and the last is the address corresponding to the lower right corner. Data corresponding to the upper left corner of the next character is stored in the address area following the address of the lower right corner. Also,
The bit arrangement is arranged so that the MSB is at the top in the figure.

The source buffer 102 temporarily stores the character pattern data for one character read from the uncompressed font pattern data storage unit 101, and the content of the uncompressed font pattern data storage unit 101 is one character. Are stored in the source buffer 102 one by one. In this way, the character pattern data to be processed is taken out character by character from the uncompressed font pattern data storage unit 101 and temporarily stored in the source buffer 102 to prevent the processing speed from decreasing. The source buffer 102 stores the character pattern data with the data structure shown in FIG.

The keyword selection counting control unit 103 has two functions, one of which is for calculating a counter address for the keyword selection counter array 104 based on the contents of the source buffer 102, and the other is for the other. It is a function of counting up the value of the counter designated by the calculated address.

FIG. 4 is a count control unit 103 for keyword selection.
3 is a diagram for explaining the first function of FIG. The keyword selection counting control unit 103 extracts continuous 3 bytes of data from the source buffer 102, that is, adjacent 8 bits × 3 data, and the currently focused data is “DR”,
Assuming that the immediately preceding data (on the left) is “DM” and the further previous data is “DL”, the keyword selection counting control unit 103 first determines the upper 4 bits of the immediately preceding data “DL”. And the lower 4 bits of the exclusive OR XOR (indicated by the symbol “∀”) are used as the upper 4 bits of the first element address for accessing the keyword selection counter array 104. In the case of the present embodiment, this first element address is a 12-bit address, and the lower 8 bits are the previous data "DM" itself. That is, the first element address is a combination of the operation result of the exclusive OR XOR of the data "DL" two before and the data "DM" one before. Here, in the illustrated example, the data “DL” two data before is “70h” (h is a symbol indicating hexadecimal notation), the operation result of the exclusive OR XOR is “7h”, and the data before one is “ "DM" is "3c
Since it is "h", the first element address is 12 of "73ch".
It becomes a bit address. On the other hand, the keyword selection counting control unit 103 sets the focused data “DR” itself as the second element address (8-bit address) of the keyword selection counter array 104. Therefore, in the example of FIG. 4, the address “73ch, 1Eh” is calculated by the address calculation function. Such address calculation processing is performed while paying attention to each byte from the third byte (excluding the first and second bytes) to the last 288th byte in the source buffer 102. The reason for starting from the 3rd byte is that the data is referred to from 2 bytes before, and moreover, the first 2 bytes of the character pattern data are often blank, which is not a problem.

Second of count control unit 103 for keyword selection
The function is to access the keyword selection counter array 104 at the address obtained by the address calculation function,
"1" is added to the value of the corresponding counter, and the content is incremented each time the corresponding counter is accessed.

The keyword selection counter array 104 is a group of counters for counting the combination frequency of consecutive 3-byte data. If the previous two data are this combination, the next data is this definition table (keyword table). Used when creating a. Here, for all characters in the uncompressed font pattern data storage unit 101,
The frequency of data combination is examined. That is, if all the characters are 13000, the keyword selection counting control unit 103 sequentially looks at the data from the 3rd byte to the 288th byte for each character, and checks the combination frequency of consecutive 3 bytes of data. The operation is repeated for 13,000 characters.

FIG. 5 shows a keyword selection counter array 1.
In the configuration of No. 04, the keyword selection counter array 104 has a two-dimensional spread, and the first element address (1
One counter is designated by 2 bits) and the second element address (8 bits), and the value is incremented. The count array 104 is 2 ^{8 in the} column direction.
= 256 columns and 2 ¹² = 4096 rows in the row direction. Further, "0" is set as the initial value of each counter in this embodiment.

When the keyword selection counter array 104 counts the combination frequency of the data for all characters in the uncompressed font pattern data storage unit 101, the keyword selection counter array 104 outputs the keyword selection counter array 104.
The second element address (value of the focused data “DR”) having the maximum count value is checked for each of the first element addresses of, and it is registered as the first candidate keyword in the keyword table storage unit 106. FIG. 6 shows an example of the configuration of the keyword table. In the illustrated example, the counter "C73C1E (the leading" C "is a counter identification symbol)" addressed by the address "73Ch, 1Eh" from the keyword selection counting control unit 103. Has the largest value in the counter group of the first element address “73Ch”. In this case, the focused data DR “1Eh” is registered in the item “73Ch” of the first candidate keyword, the operation result of the exclusive OR XOR two bytes before is “7h”, and the data one byte before is “3Ch”. If so, the data of interest is "1E
It means that there is a high possibility that it is “h”. 2 more
The value of the second element address having the second largest value is the second candidate keyword, and the values of the second element address having the third and fourth largest count values are the third and fourth candidates. Registered as a keyword. As described above, in this embodiment, four types of data candidates are registered according to the combination of data.

The keyword table stored in the keyword table storage unit 106 is a definition table in which when the previous two data are this combination, the next data of interest is this definition table. Since it is a table for deriving, the address originally has 16 bits for the previous two. However, such a table has a capacity of 64 Kbytes. Further, it is unlikely that the character pattern has all the combinations that fill up such a table, and the table itself becomes redundant. Therefore, in this embodiment, a method of reducing the table capacity to 1/16 of the normal size is used.

That is, the keyword selection counter array 1
The first element address 04 is a combination of the previous data "DM" and the previous two data "DL" with respect to the data of interest "DR", but the previous two data "DL" is Exclusive or X of upper 4 bits and lower 4 bits
Since the original 16-bit address becomes a 12-bit address because it is reduced to 4-bit data obtained by the OR operation, the capacity of the keyword table is 2 ¹²
/ 2 ¹⁶ = 1/16. Here, the previous data “D
L is reduced because it has less influence on the focused data “DR” than the previous data “DM”. Also, the exclusive OR XOR operation is selected because of the influence of all bits. This is because the remaining 4 bits remain. Further, the reason that the upper 4 bits and the lower 4 bits are calculated is because the continuity of the pattern data is taken into consideration. According to this method, there are 16 kinds of combinations having the same address, but it has been experimentally confirmed that most of them can be covered by 4 kinds of combinations which have a high frequency. Two types are sufficient, but in this embodiment, four
I try to select a combination of types.

The data appearance frequency counting unit 107 operates after the keyword table is created, and not all the data hits the keyword only with the keyword table. Therefore, the keyword selection counting control unit 103
The number of appearances of data is counted again in the same manner as. Here, the number of appearances of the data that did not hit is counted. When the data appearance number counting unit 107 starts its operation, all the characters in the uncompressed font pattern data storage unit 101 are stored in the source buffer 102 one by one, and the data appearance number counting unit 107 sets the number of characters to 1.
It is processed character by character.

That is, the data appearance frequency counting unit 107 uses the method described in the keyword selection counting control unit 103 to select the data from the previous two data "DL" and the previous data "DM" to the target data "DR". The bit address is obtained, and the keyword table storage unit 10 is calculated using this 12-bit address.
6 is accessed to check whether the focused data “DR” matches the first to fourth candidate keywords. Here, when the focused data “DR” matches any of the first to fourth candidate keywords,
The value of the counter in the count counter array 108 corresponding thereto is incremented by "+1". If none of the keywords matches, the value of the counter in the number-of-times count array 108 corresponding to the data of interest is incremented by "+1".

FIG. 7 shows the structure of the frequency count array 108. In the frequency count array 108, when the data of interest "DR" does not match the keyword, the data of interest "00" to "FF" are displayed. Counters "C000h" and "C001" that count the number of times data appears
h "..." C0FFh "and counters" C100h "," C101h ", and" C10 "that count the number of appearances of data in the case of matching with any of the first to fourth candidate keywords.
2h ”and“ C103h ”are provided. Here, each counter has a specific code “000h” to “0FFh”,
"100h", "101h", "102h", "103"
Each code is associated with “h” and defines a reproduction command for restoring and reproducing the compressed character data to the original character pattern data.

FIG. 8 shows the contents of each command code and its execution command. Command code "000h"
~ "0FFh" means "store the contents (focused data) of the lower 8 bits of the command code", and the command code "100h" refers to the first candidate keyword and stores the contents (focused data). Means. Furthermore, the command code "101h" means "store the contents by referring to the second candidate keyword", and the command code "102h""stores the contents by referring to the third candidate keyword". The command code “103h” means “store the contents by referring to the fourth candidate keyword”. As described above, as for the content of the command, there is one in which a part of the command data is directly reproduced as data, and the other in which the content is reproduced as data by referring to the keyword table.

As described above, the data appearance frequency counting unit 107 counts up the value of the corresponding counter when the focused data “DR” matches the keywords of the first to fourth candidates, but a considerable amount of data is the first. Since the keywords match the candidates to the fourth candidate, the count values in the count array 108 are command codes “100h” to “103”.
The counter value corresponding to “h” becomes extremely large, and a bias occurs. Since the occurrence frequency of data is biased in this way, effective data compression can be realized.

The compression code table generation unit 109 generates a compression code using the count value in the number count array 108 when the data appearance frequency counting unit 107 finishes processing for all characters. In this case, as long as the code has a characteristic of cutting redundancy such as a Huffman code generation algorithm, its content does not matter. Note that the Huffman coding method and the like are widely known techniques, and therefore description thereof will be omitted here. The code table generated by the compression code table generating unit 109 is written in the compression code table storage unit 110.

FIG. 9 shows a compression code table. The compression code table is composed of a compression code string and the number of bits thereof corresponding to a command code. This compression code string is variable length data, and according to the Huffman coding method. A data having a high appearance frequency is represented by a compression code string having a small number of bits. It should be noted that the contents of the compression code table are arranged in order from the one having the smallest number of bits in order to make it easy to see in the figure, but it is not actually necessary to arrange them in this way. This compression code table has two uses. One is when converting a non-compressed font pattern into a compressed command string, and the other is when creating an inverse conversion table used when decompressing a compressed command string.

The compression encoding unit 112 performs substantially the same operation as the data appearance number counting unit 107, but instead of counting up the counter of the number counting array 108 in the data appearance number counting unit 107, the keyword table is used. Referring to the contents of the storage unit 106, the bit sequence (compression code sequence) stored in the compression code table storage unit 110 is stored in the compressed font storage unit 1 via the data writing unit 113.
The operation of exhaling to 14 is performed.

That is, all the characters in the uncompressed font pattern data storage unit 101 are stored character by character in the source buffer 102 and processed by the compression encoding unit 112 character by character. In this case, the compression encoding unit 112
Obtains a 12-bit address from the previous two data "DL" and "DM" for the data of interest "DR" by the method described above, accesses the keyword table storage unit 106 with this 12-bit address, and obtains the data of interest "DR". Checks whether any of the first to fourth candidate keywords matches. Here, when the focused data “DR” matches any of the first candidate keyword to the fourth candidate keyword, the compression code table is referred to, the command code corresponding to the candidate is designated, and the compression code string is selected. read out.
This compressed code string is given to the data writing unit 113 and written in the compressed font storage unit 114. On the other hand, if they do not match, the command code corresponding to the data of interest is designated and the compression code string is read. This compressed code string is given to the data writing unit 113, and the compressed font storage unit 1
14 is written. The compression code string for one character written in the compressed font storage unit 114 in this manner becomes variable length data. Then, when a compression code string for one character is written in the compressed font storage unit 114, the data writing unit 113 uses an index as an address for indicating from which byte of the compressed font storage unit 114 the character is started to be written. It is set in the table storage unit 115.

The inverse conversion table generating section 116 rearranges the compression code sequences forming the compression code table into numerical values and rearranges them in ascending order of numerical values. At this time, the inverse conversion table generating unit 11
6 is regarded as a numerical value by adding "0" to the right side according to the code having the maximum bit length. For example, the code string of the command code “101h” is “10”, but if the maximum bit length is 16 bits, “0” is set to 1 to the right of this.
Add four and handle as "8000h". When the contents of the compression code table storage unit 110 are rearranged in the numerical order in this manner, the arrangement is as shown in FIG.

Here, when a certain bit string is given when the contents of the compressed font storage unit 114 are restored and reproduced, in order to check what the command code is, one bit is taken out from the beginning of the bit string and checked. At the end, if the analysis ends when the code string of the command code matches, it is extremely convenient in terms of processing speed. Therefore, looking at FIG. 10, it is understood that if the leading bit is “0”, it can be concluded that it is the command code “100h”. However, if it is a "1", then the next bit must be seen. Now, when looking at the next bit, if it is "0", the command code is "101h",
If it is "1", then the next bit must be seen.
Here, FIG. 11 shows the compression code table of FIG. 10 in a tree list structure.

In this binary tree list structure, "ni
The branch point labeled "l" indicates where the command code is fixed, and the branch point labeled "dir" indicates that the next bit needs to be seen. Therefore, if the "dir" portion is represented by a structure as shown in FIG. 12, it is possible to obtain a very efficient inverse conversion table. Here, as for “dir”, “behavior” is divided into two cases, where the target bit is “0” and “1”. Therefore, one set of "dir" is represented by two sets of 16-bit data, that is, four bytes. "Address + 0 (16 bits)" represents the behavior when "0", and "Address + 2 (16 bits)" represents the behavior when "1". The table address is "00" from the root of the list.
It is assigned to 0 ”,“ 004 ”,“ 008 ”.... FIG. 13 shows the structure of the behavior data in the inverse conversion table. The "behavior" may be "nil" there, and the data may be fixed or may be newly "dir". Therefore, as shown in FIG. 13, an identifier of "dir" or "nil" is placed in the MSB of 16-bit data,
The remaining 15 bits represent the address on the inverse conversion table in which the next "dir" is described, or the value of the command code. When the MSB identifier is "1", it is "nil", and the following 15 bits represent a command code, and when the MSB identifier is "0", it is "dir".
Then, the following 15 bits represent the next "dir" address. FIG. 14 specifically shows the contents of the inverse conversion table. The inverse conversion table generator 116 creates an inverse conversion table having a structure as shown in FIG. 14 according to the compression code table shown in FIG. Conversion table storage unit 117
To store.

Next, the configuration of the document creating apparatus 200 is shown in FIG.
5, and will be described with reference to FIG. FIG. 15 is a block diagram showing the overall configuration of the document creation device 200. The main control unit 210 controls the overall operation of the document creating apparatus 200 according to various programs, and the key operation unit 211.
The document data input from the main control unit 210 is fetched, displayed and output from the display device 213 via the display control unit 212, or stored in the document memory 214. The document memory 214 stores the document data in a code format, the content of which is given to the external storage control device 215 in response to a registration command from the key operation unit 211, registered and stored in the external storage device 216, The data is read in response to a print command from the key operation unit 211, converted into corresponding character pattern data by the character pattern reproducing device 201, and then printed out from the printing device 218 via the print control unit 217.

The character pattern reproducing device 201 is configured to have a decompression control unit 219 and a character pattern buffer 220 in addition to the above-mentioned compressed character data memory 203. In the compressed character data memory 203, as shown in FIG. A compressed font storage unit 204, an index table storage unit 205, an inverse conversion table storage unit 206, and a keyword table storage unit 207, which store various data supplied from the data compression apparatus 100, are provided. The compressed character data is restored to normal character pattern data on the basis of the above data and reproduced and written in the character pattern buffer 220.

FIG. 16 is a block diagram showing in detail the character pattern reproducing device 201. The expansion control unit 219 has a character code register 221, an index table access unit 222, a compressed data access unit 223, a compression pattern buffer 224, a reproduction command generation unit 225, and a reproduction command execution unit 226. When the character code to be printed is set in the character code register 221, the index table access unit 222 searches the index table storage unit 205 to access where in the compressed font storage unit 204 the character is written. In this case, since the contents of the index table storage unit 205 follow the order of JIS, the index table access unit 222 can retrieve the corresponding address from the index table storage unit 205 based on the character code.
The address thus retrieved is sent to the compressed data access unit 223. The compressed data access unit 223 accesses the compressed font storage unit 204 according to this address, reads the compressed character data for one character, and transfers it to the compressed pattern buffer 224.

The reproduction command generation section 225 has a pointer (not shown) for the inverse conversion table, and the compressed character data (compression code string) in the compression pattern buffer 224 is converted into a command code according to the contents of the inverse conversion table storage section 206. The command code thus converted is sent to the reproduction command execution section 226 for each command.

The reproduction command execution unit 226 sequentially executes the store processing according to the contents of the execution command shown in FIG. 8 in accordance with the command code from the reproduction command generation unit 225. Data is extracted as necessary, and the data is written in the character pattern buffer 220. Here, when the command code is “100h” to “103h”, that is, the content is “store with reference to first candidate keyword to fourth candidate keyword”, it is written in the character pattern buffer 220. The data two bytes before is read from the character pattern buffer 220, the address of the keyword table is generated from this data by the method described above, and the table contents designated by this address are written in the character pattern buffer 220 as reproduction data. As a result, the restored and reproduced character pattern data is stored in the character pattern buffer 220. Here, the document creator main unit 202 uses the character pattern buffer 22 as usual.
The character pattern data is read from 0 and printed out.

Next, the operation of this embodiment will be described. First,
A pattern compression operation by the character pattern data compression device 100 will be described. Various character pattern data stored in the uncompressed font pattern data storage unit 101 by the bitmap method is transferred to the source buffer 102 one by one from the first character. Then, the keyword selection counting control unit 103 first determines the source buffer 1
The data of the 1st to 3rd bytes is cut out from 02, and the data of the 3rd byte is focused.
The 2nd byte data immediately before “R” is “DR”,
The data of the immediately preceding first byte is “DL”. Here, the keyword selection counting control unit 103 takes an exclusive OR XOR of the upper 4 bits and the lower 4 bits of the data “DL” two before, and uses this to access the keyword selection counter array 104. The most significant 4 bits of the first element address. Now, in the example of FIG. 4, the second previous data, that is, the data of the first byte is "011100".
00 (70h) ”, the result of the XOR is“ 0111 ”.
(7h) ”. Then, since the data “DM” of the immediately preceding second byte is “00111100 (3Ch)”, the first element address is “011100111100”.
(73 Ch) ”is a 12-bit address. On the other hand, the keyword selection counting control unit 103 sets the focused third byte data “DR” itself to the second element address “0001111” of the keyword selection counter array 104.
0 (1 Eh) ". Therefore, in the example of FIG.
The address “3Ch, 1Eh” is calculated.

Next, the keyword selection counting control unit 103
Pays attention to the 4th byte of the source buffer 102, and calculates the address of the keyword selection counter array 104 based on the 2nd previous data of the 2nd byte and the 1st previous data of the 3rd byte. Such address calculation processing is 1
It is repeatedly executed byte by byte up to the last 288th byte. At the same time, the keyword selection counting control unit 103 accesses the keyword selection counter array 104 according to the calculated address and counts up the corresponding counter value. As a result, the combination frequency of the data is determined by the keyword selection counter array 104.
Will be counted.

After counting the combination frequencies of all the characters in the uncompressed font pattern data storage unit 101 in this way, the keyword selection unit 105 analyzes the contents of the keyword selection counter array 104 and opens the keyword table. create. That is, the keyword selection unit 10
5 is the second element address having the largest counter value for each first element address of the keyword selection counter array 104, that is, the focused data having the highest combination frequency is the first
Register as a candidate keyword, register the second element address of the next largest counter value as the second candidate keyword,
Similarly, the second element address having the third and fourth largest counter values is registered as the third candidate keyword and the fourth candidate keyword. As described above, four types of candidates that the following data of interest is this are extracted based on the combination frequency of the data and stored in the keyword table storage unit 106.

Now, when the creation of the keyword table is completed, the data appearance frequency counting section 107 starts its operation while referring to it. Also in this case, the character pattern data is sequentially stored in the source buffer 102 from the uncompressed font pattern data storage unit 101 one character at a time. Here, the data appearance number counting unit 107 counts the number of times the data appears as in the keyword selection counting control unit 103 as described above, but this time, the keyword table storage unit 106.
First to fourth candidates in the case of matching with the keyword in
The number of appearances for each candidate and the number of appearances for each data when the keywords do not match are counted.

That is, the data appearance number counting unit 107 accesses the keyword table storage unit 106 with the 12-bit address obtained from the previous two data with respect to the data of interest, and the data of interest is one of the first to fourth candidates. If it matches the keyword, the counter value in the number-of-times count array 108 corresponding to the keyword is incremented. For example, the 12-bit address is "73Ch", and the first candidate keyword "73Ch" has an item "1" as shown in FIG.
When “Eh” is registered, the data of interest is “1Eh”
If this is the case, this focused data matches the first candidate keyword, so the counter “C
The value of “100h” is incremented by “+1”. On the other hand, if none of the keywords matches, the counter value in the number-of-times count array 108 corresponding to the focused data is counted up. For example, if the data of interest is "01h", the counter "C00 in the count counter array 108
The value of "1h" is incremented by "+1". Such an operation is repeatedly executed for every character until the end of one character. As a result, since a considerable amount of data matches the keyword, the values of the counters "C100h" to "C103h" become extremely large, and the optimum bias for data compression occurs. Here, since each counter in the count counter array 108 is associated with a reproduction command code (see FIG. 8) used when reproducing the compressed character data into the original character pattern data, a considerable amount of data is reproduced. It is possible to rewrite the commands "100h" to "103h".

Next, the compression code table generator 109 uses the count data of the number count array 108 to generate a compression code by a method such as the Huffman coding method that suppresses the redundant portion according to the appearance frequency of the data. , Compression code table storage unit 110
To store. In the compression code table generated in this way, a code string having a high data appearance frequency is represented by a code string having a small number of bits as shown in FIG.

When the compression code table is completed in this way,
The compression encoding unit 112 starts operating. Also in this case, the character pattern data in the uncompressed font pattern data storage unit 101 is sequentially stored character by character in the source buffer 102. Here, when the compression encoding unit 112 accesses the keyword table storage unit 106 with the 12-bit address obtained from the previous two data for the data of interest, and the data of interest matches the keywords of the first to fourth candidates. , The compression code table is referred to, the command code corresponding to the candidate is designated, and the compression code string is read. On the other hand, if they do not match, the command code corresponding to the data of interest is designated and the compression code string is read.

Now, for example, if the upper left data string of the character pattern data is represented by a command code, "100
h + 100h + 0FFh + 102h + 101h + ... ”
It is assumed that they are arranged. In this case, referring to the compression code table and rewriting the command code with the corresponding compression code string, the above-mentioned command string becomes "0 + 0 + 110".
10000 + 1100 + 10 + ... ”. This 8
When expressed by dividing each bit, "00110100 + 0"
"0110010 + ...", and when expressed in hexadecimal, it becomes "34h + 32h + ...". In the present embodiment, attention is focused on the third byte of the character pattern data, but the first two bytes are the upper left 16 dots of the character, and are often "0" (blank). Therefore, as a special case, when this area is “0”, it is rewritten to a command (“100h” in the example of FIG. 8) that refers to the first candidate keyword. In this way, when the playback command is executed, if the command code "100h" comes in the first 2 bytes, if "0" is promised to be written, no problem will occur.

In this way, the compression encoding unit 112 sequentially processes the character pattern data from the upper left, and one character 288 is processed.
When the bytes are processed, they are separated once, and "0" is inserted until the last bit string of one character is 8 bits. This is because the start bits of the data are aligned for each character. If this is not done, the contents of the "index table" for each character will be a lot of data, such as "from the xth bit of the address XXXXX". This is because the storage capacity must be increased and the table capacity increases, resulting in not only adversely affecting the overall compression ratio, but also inconvenient for the processing speed during expansion.

The compression code string thus obtained is supplied to the data writing unit 113 for each character and written in the compressed font storage unit 114. At this time, the data writing unit 113 creates an index table and writes it in the index table storage unit 115. The index table stores the first address of each character stored in the compressed font storage unit 114. On the other hand, the inverse conversion table generating unit 116 rearranges the contents of the compression code table storage unit 110 in numerical weight order and converts the contents into a compression code table as shown in FIG. 10, and also converts this compression code table into a binary tree list structure. 14 is stored in the reverse conversion table storage unit 117.

As a result, the character pattern data compression device 1
When the compressed font storage unit 114, the index table storage unit 115, the inverse conversion table storage unit 117, and the keyword table storage unit 106 in 00 all have data, the operation of the character pattern data compression device 100 ends. The contents of the compressed font storage unit 114, the index table storage unit 115, the inverse conversion table storage unit 117, and the keyword table storage unit 106 are the contents of the document creation device 20.
When the product of No. 0 is completed, the corresponding compressed font storage unit 204, index table storage unit 205, inverse conversion table storage unit 206, keyword table storage unit in the document creation apparatus 200 are replaced with the normal bitmap character pattern data. It is installed in 207.

With the compressed font data, index table, inverse conversion table, and keyword table incorporated as described above, the document creation device 200
Operates as follows. When printing a document, the code to be printed is read from the character document memory 214 and given to the main control unit 210 via the main control unit 210.
Then, this character code corresponds to the character pattern reproducing device 201.
Since it is set in the character code register 221, the index table access unit 222 retrieves the corresponding address from the index table storage unit 205 based on this character code, and the compressed data access unit 223 stores one character corresponding to this address. The compressed character data (compressed code string) is read and transferred to the compressed pattern buffer 224.

Here, the reproduction command generator 225 refers to the contents of the inverse conversion table storage 206 and converts the compression code string in the compression pattern buffer 224 into a corresponding command code. In this case, the reproduction command generator 225 is used. 225
A pointer (not shown) for the inverse conversion table is provided inside, and first, "di" of Root of the tree list structure is provided.
The “r” address “0” is set as the initial value of the pointer. In this state, 1-bit data is read from the head of the compression code string in the compression pattern buffer 224, and it is checked whether it is "0" or "1". If the first bit is "0", "pointer value + 0"
The contents of the inverse conversion table addressed by are read, and it is checked whether the MSB of the 16-bit data is "0". Here, if the MSB is “1”, the “behavior” becomes “nil”, the data is fixed there, and the data of the lower 15 bits becomes a command code, which is given to the reproduction command execution unit 226. On the other hand, if the MSB is "0", the "behavior" is "dir" and the next bit needs to be seen. Therefore, the data of the lower 15 bits becomes the next table address, which is stored in the pointer.

On the other hand, if the first bit of the compression code is "1", the contents of the inverse conversion table addressed by "pointer value + 2" are read and it is checked whether the MSB of the 16-bit data is "0". To do. Even in this case, if the MSB is “1”, the “behavior” becomes “nil”, and the lower 15 bits of the data become a command code, which is given to the reproduction command execution unit 226.
If it is “0”, the “behavior” becomes “dir”, and the lower 15 bits of data become the next table address,
This is stored in the pointer. Such an operation is repeatedly performed on the compression code string bit by bit.

Now, for example, the compression code string is "00110".
100 + 00110010 + ... ”, and the contents of the inverse conversion table storage unit 206 have the data structure shown in FIG. In this case, the first bit of the compression code string is "0", the table content addressed by "pointer value + 0" is "8100h", and therefore the lower 15-bit data "100h" of this table content is transmitted as a command code. To be done. Next, since the code of the second bit is also “0”, the command code “100
h "is sent. Next, the code of the third bit is "1", and the content of the table addressed first is "0".
004h ”, that is, the lower 15-bit data of the table contents is set in the pointer as the next table address“ 0004 ”. Here, since the code of the next 4th bit is also "1", the table content addressed by "pointer value (0004) +2" is read out.
In this case, the table content is "0008h", and therefore the lower 15 bits of data of this table content are set in the pointer as the next table address "0008". Next, the data of the 5th bit is “0”,
The table content “000Ch” addressed by “pointer value (0008) +0” is read, and the lower 15 bits of data are read to the next table address “000”.
"C" is set in the pointer.

Similarly, the code of the next 6th bit is "1", the table address jumps from "000C" to the address "0010", and the code of the 7th to 9th bits is "0". Therefore, the table address jumps to “address 0014”, “address 0018”, and “address 001C” in order. The code of the 10th bit is also "0", but in this case, the table content is "80F.
Eh ", and the lower 15 bits of data" 0FE "
h ”is sent as a command code. In this way, the data of the third bit to the 10th bit of the compression code string is converted into the command code “0FEh”. Similarly, the 11th to 14th bits of the compression code string (1100)
Table address is read as "00".
"00" to "0004", "0008",
Jump to "000C address" in sequence, and go to "000C address"
When the table content "8102h" of the above is read, the lower 15-bit data "102h" is sent out as a command code. In this way, the reproduction command generator 2
25 sequentially converts the contents of the compression pattern buffer 224 into command codes by referring to the inverse conversion table.

Then, the reproduction command execution section 226 executes the processing according to the command code while referring to the contents of the keyword table storage section 207. That is, the reproduction command execution unit 226 outputs the reproduction command generation unit 225
The processing according to the contents of the execution command shown in FIG. 8 is executed according to the command code sent for each command. Here, when the command code "100h" is generated for the first two bytes of one character as described above, the fixed value "0 (blank)" is written in the character pattern buffer 220. Now, if the command code is "000h" to "0FFh", the contents of the lower 8 bits of this command code become the reproduction data as they are,
It is stored in the character pattern buffer 220. On the other hand, the command codes are "100h", "101h", "102".
If h ”and“ 103h ”, the reproduction command execution unit 22
6 reads the previous two data from the character pattern buffer 220 and calculates the address of the keyword table based on this data. Then, the keyword table storage unit 207 is accessed, and the command code “100h”,
The first to fourth candidate keywords corresponding to "101h", "102h", and "103h" are designated, and data corresponding to the calculated address is extracted from the designated keywords. This data becomes the reproduction data for the command codes “100h” to “103h” and is stored in the character pattern buffer 220. By repeating such processing for one character of 288 bytes, one character of pattern data is reproduced in the character pattern buffer 220.

As described above, the character pattern data compression apparatus 100 creates the keyword table of the first candidate to the fourth candidate for each combination of frequently used data, and converts each keyword table into command codes "100h" to "103h". By making them correspond to each other, the data appearance frequency of the command is extremely biased, and it is possible to obtain a high compression rate by compressing it by the Huffman coding method having a characteristic of cutting redundancy. That is, a considerable amount of data is replaced with a compression code string having a small number of bits corresponding to the command codes “100h” to “103h”, so that the compression rate can be made extremely high. In this case, in the present embodiment, since the keyword tables of the first to fourth candidates are created, the probability of hitting is higher than in the case where the number of keyword tables is smaller, and high compression is possible. For example, when the capacity of the uncompressed font pattern data storage unit 101 is about 7 Mbytes (13000 characters × characters × 2 typefaces), the compressed font storage unit 204 has 2M bytes.
Bytes, the index table storage unit 205 is 3 bytes x the number of characters, and the inverse conversion table storage unit 206 is 1 Kbyte /
Typeface, that is, 2K bytes, keyword table storage 2
07 is 16 Kbytes / typeface, that is, 32 Kbytes, which has a memory capacity of about 1/3 as a whole and enables extremely high data compression. Here, since the address of the keyword table is reduced from the 16-bit address to the 12-bit address, the capacity of the keyword table is 1/16 of that of the 16-bit address, and the table capacity can be significantly reduced, resulting in the entire compression. It contributes to the improvement of the rate. Further, when reproducing the compressed font data compressed in this way into the original character pattern data, the document creation apparatus 200 converts the compression code into a command code, and then simply executes this command to easily restore and reproduce. It is possible. In this case, since the compression code is converted into the command code according to the binary tree list, high speed reproduction is possible.

[0054]

Second Embodiment Next, a second embodiment will be described with reference to FIGS. 17 and 18. Here, in this embodiment, one set of typeface data generated by the character pattern data compression apparatus 100, that is, compressed font data, an index table, an inverse conversion table, and a keyword table should be incorporated in the document creation apparatus 200 in advance. However, in the present embodiment, the present invention is applied to the document creating apparatus of the multi-face type system, and two sets of typeface data are incorporated in the document creating apparatus 200 in advance, and other typeface data may be added as necessary. Is supplied from an external storage device such as a floppy disk. In this case, the feature of the present embodiment is that only the compressed font data and its index table are supplied from the external storage device.

FIG. 17 shows a document creating apparatus 30 according to this embodiment.
It is a block block diagram of 0. The document creation device 300 includes a document creation device main device and a character pattern reproduction device 301, and ROMs 302 and 303 forming a compressed character pattern memory.
And a RAM 304 forming a compression pattern buffer,
It is configured to have an external storage device 305 such as a floppy disk. The ROM 302 stores two sets of compressed font data A-1, A-2 and index tables B-1, B-2 corresponding to two types of fonts, and the ROM 303 supports four types of fonts. Then, four sets of keyword tables C-1, C-2, C-3, C-4 and inverse conversion tables D-1, D-2, D-3, D-4 are stored. Here, the compressed font data A-1, the index table B-1, the keyword table C-1, and the inverse conversion table D-1 correspond to the first typeface data, and the compressed font data A-2 and the index table B-. 2,
The keyword table C-2 and the inverse conversion table D-2 correspond to the second typeface data, and when the first typeface or the second typeface is selected, the same as in the first embodiment. Restoration reproduction processing is performed by referring to the corresponding table contents.

In the present embodiment, the third and subsequent typefaces are supplied from the external storage device 305, and when any character is designated, the index table B-3 in the external storage device 305 is referred to. Compressed font data A
Data for one character is read from -3 and set in the RAM 304 which constitutes the compression pattern buffer. Here, the character pattern reproducing apparatus is the same as the above-mentioned first embodiment.
The data in the AM 304 is converted into a command code by referring to the inverse conversion table, and the command is executed by referring to the keyword table. At that time, which inverse conversion table or keyword table is referred to is 4 All types can be referenced, and the designation is performed by referring to 1 byte (table designation information) of a fixed address in the external storage device 305. When dealing with many typefaces supplied from the external storage device 305,
This is because it is possible to select the most suitable table (the one that can be highly compressed) for the supplied typeface from the various tables built in the document creation apparatus 300, whereby the compressed font data is stored in the external storage device 305. It is sufficient to store only A-3 and the index table B-3.

The compressed font data A-1, A-2, the index tables B-1, B-2, the keyword tables C-1, C- built in the document creating apparatus 300 are included.
2, C-3, C-4, inverse conversion tables D-1, D-2,
D-3 and D-4 are transferred and incorporated from the character pattern compression device as in the first embodiment, but the compressed font data and index table corresponding to the third and fourth typefaces are the documents. It is deleted without being made resident in the creating apparatus 300.

Further, in the first embodiment described above, "0" is set as the initial value of the number-of-times count array 108 which constitutes the character pattern data compression apparatus 100. However, many subtly different typefaces are externally stored. Only the keyword table and the inverse conversion table supplied from the device 305 are RO
When it is shared with M, the count array 1
“1” is set as the initial value of 08. This is because it is assumed that data exists at least once in any data, and if the initial value of the counter count value is set to "0" as in the first embodiment, the compression code string for the command is This is to avoid the inconvenience caused by the compression code generation method having the characteristic of cutting the redundancy that it does not exist.
As a result, the compression code string is generated even for the data in which the data does not appear, so that the character pattern data of a slightly different typeface can be dealt with even if supplied externally.

FIG. 18 shows a process of writing compressed font data and an index table in the external storage device 305. First, uncompressed font pattern data P-1, P-2 corresponding to the first to fourth typefaces,
When P-3 and P-4 are compressed using the character pattern data compression apparatus 100 in the first embodiment, compressed font data, index table, inverse conversion table, and keyword table are generated for each typeface, and A compression code table is generated in the generation process. this house,
Keyword tables X-1, X-2, X- for each typeface
3, X-4 and compression code tables Y-1, Y-2, Y-3, Y-
Take 4 out.

Next, when compressing the typeface (uncompressed font pattern data P-n) supplied to the document creating apparatus 300 by the external storage device 305 by the character pattern data compressing apparatus 100 of the first embodiment, Typeface-4th
Among the keyword tables X-1 to X-4 and the compression code tables Y-1 to Y-4 that have been extracted in advance corresponding to the typeface, the one most suitable for the typeface supplied by the external storage device 305 (that is, The compression device having the highest compression rate when actually compressed for each typeface is selected by the selection device 400. Then, the selected keyword table and compression code table are used as the character pattern data compression apparatus 1 of the first embodiment.
00 is stored in the keyword table storage unit 106 and the compression code table storage unit 110. In this state, the compression coding unit 112 starts the operation. The compressed font data n and the index table n thus obtained are written in the external storage device 305.

In this embodiment having the above-mentioned structure, the same effects as those of the first embodiment can be obtained. In addition, only the compressed font data and the index table need be supplied from the external storage device 305, and the external storage device 305 can be used. Since the keyword table and the inverse conversion table are not required for this, the capacity of the external storage device 305 can be effectively used. Further, since “1” is set as the initial value of the number-of-times count array 108, the compression code can be generated even for data that does not appear when the appearance frequency of the data is checked.

Although the first to fourth candidate keywords are prepared in the above embodiment, only two kinds of keywords may be used and the number thereof is arbitrary. Also,
In the above embodiment, the combination frequency of the data of interest with the previous two data is checked, but the combination frequency with the previous data or the combination frequency with the previous three data may be checked. . Further, in the above-described embodiment, as a method of reducing the first element address of the keyword table, a method of taking an exclusive OR XOR of the upper 4 bits and the lower 4 bits of the data "DL" two before is used. You may make it reduce by calculation. In this case, for example, a method of alternately performing an OR operation bit by bit can be considered. Further, the present invention can also be configured by software controlled by a program, which makes it possible to provide a high-speed and low-cost character pattern data compression device 100.

[0063]

According to the present invention, when a combination pattern of a certain data forming the character pattern data and the preceding data frequently appears, this is associated with the fixed data for reproduction and the appearance frequency of the data. Since the character pattern data can be compressed by using a compression method that suppresses redundancy by biasing the, the high compression becomes possible, and when the compressed font data is incorporated in the ROM of the document creation device, It is possible to effectively incorporate compressed font data into a memory with less memory.

[Brief description of drawings]

FIG. 1 is a diagram schematically showing a compression / reproduction process of character pattern data in the first embodiment.

FIG. 2 is a block configuration diagram of a character pattern data compression device 100.

FIG. 3 is a diagram showing a data structure of a character pattern stored in an uncompressed font pattern data storage unit 101.

FIG. 4 is a diagram for explaining a first function of a keyword selection counting control unit 103.

FIG. 5 is a diagram showing a configuration of a keyword selection counter array 104.

FIG. 6 is a diagram showing a configuration of a keyword table.

FIG. 7 is a diagram showing a configuration of a frequency counting array 108.

FIG. 8 is a diagram showing the contents of each command code and its execution command.

FIG. 9 is a diagram showing a compression code table.

FIG. 10 is a diagram showing a compression code table sorted in numerical weight order.

11 is a diagram showing the compression code table of FIG. 10 in a tree list structure.

FIG. 12 is a configuration diagram of an inverse conversion table.

FIG. 13 is a diagram showing the structure of “behavior” data in the inverse conversion table.

FIG. 14 is a diagram specifically showing the contents of an inverse conversion table.

FIG. 15 is a diagram showing the overall configuration of a document creation device 200.

16 is a block configuration diagram of a character pattern reproducing device 201. FIG.

FIG. 17 is a block configuration diagram of the document creation device 300 in the second embodiment.

FIG. 18 is a diagram showing a process of writing compressed font data and an index table in the external storage device 305.

[Explanation of symbols]

 100 Character Pattern Data Compressor 101 Uncompressed Font Pattern Data Storage Unit 103 Keyword Selection Counting Control Unit 104 Keyword Selection Counter Array 105 Keyword Selection Unit 106 Keyword Table Storage Unit 107 Data Appearance Counting Unit 108 Number Counting Array 109 Compression Code Table Generation unit 110 Compression code table storage unit 112 Compression coding unit 113 Data writing unit 114 Compressed font storage unit 115 Index table storage unit 116 Inverse conversion table generation unit 117 Inverse conversion table storage unit 200 Document creation device

Claims (3)

[Claims] 1. A character pattern data compression apparatus for compressing each character pattern data of a bit map system and incorporating it in a document creation apparatus, focusing on each data constituting the character pattern data sequentially,
A combination frequency detecting means for detecting the combination frequency of all the characters in accordance with the combination of the data of interest and the preceding data, and a case where a high frequency combination is detected by the combination frequency detecting means. If the previous data is this combination, the data of interest is this table creation means for creating a candidate table that stores data candidates for each combination, and by making the candidate table correspond to the fixed data for reproduction. A compression code generation means for generating a compression code string in which the appearance frequency of the data is biased and a redundant portion is suppressed according to the appearance frequency of the data, and a compression code string generated by the compression code generation means for each character pattern data. A character pattern comprising: a compression unit that compresses the character pattern data by replacing Data compression device. 2. When a combination having a high frequency is detected by the combination frequency detecting means, the table creating means creates a plurality of candidate tables having different data of interest for one combination of previous data, 4. The character pattern data compression apparatus according to claim 1, wherein the compression code generation means is adapted to associate fixed data with each of the plurality of candidate tables. 3. The unit of each data forming the character pattern data is a bit group forming one word occupying one address when the data is held by the bit map method. Character pattern data compression device described.