JP2907668B2 - Data processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing apparatus for transmitting data from a data terminal device such as a personal computer (personal computer) to a facsimile apparatus or another personal computer via a general communication line.

[0002]

2. Description of the Related Art A so-called personal computer communication modem is a device for performing data communication by connecting a telephone line to an RS-232C interface of a personal computer. Recently, a device called a facsimile adapter or FAX modem has been developed in which a function of transmitting and receiving data to and from a facsimile device is added to this type of modem.

FIG. 26 shows a schematic configuration of a conventional facsimile adapter. In FIG. 26, DTE (DAT
A TERMINAL EQUIPMENT 1 indicates a data terminal device such as a personal computer, and a DCE (DATA C
IRCUIT-TERMINATING EQUIPM
ENT) 2 indicates a facsimile adapter. The DCE 2 connects a private line 3 such as an RS-232C interface of the DTE 1 to a general communication line 4 such as a telephone line.
In the following description, the CCITT T.D. It is assumed that facsimile communication according to No. 30 is performed.

When data is transmitted from DTE1, DT
The data output from the E1 to the private line 3 is temporarily stored in the buffer 5 of the DCE 2, and the data processing unit 6
The data is then transferred to a buffer 7 and transmitted to a partner device (such as a personal computer or a facsimile device) via a general communication line 4 by a transmission / reception modem 8. When the DTE 1 receives data, the data on the general communication line 4 received by the transmission / reception modem 8 of the DCE 2 is temporarily stored in the buffer 7, processed by the data processing unit 6 as necessary, and transferred to the buffer 5, The DTE 1 receives the data in the buffer 5 via the private line 3.

The operation of the DCE 2 includes a facsimile communication mode for transmitting image information compressed and encoded between the DTE 1 and the facsimile apparatus, and a data communication mode for transmitting code information between the DTE 1 and another personal computer or the like. Are roughly divided into

In the facsimile communication mode, the image data compressed and coded from the DTE 1 is sent to the general communication line 4 as it is, and the image data from the DTE 1 is processed by the DCE 2 before being sent to the general communication line 4. There are cases. For example, if the other facsimile machine has the ability to receive MR encoded data and the image data from the DTE 1 is MH encoded data, the MH encoding is performed in the data processing unit 6 of the DCE 2 to reduce the communication time. The data is converted into MR encoded data and sent to the general communication line 4. For example, if the other party's facsimile machine has only B5 size recording paper and the image data from DTE1 is of A4 size, DCE2
After the size of the image data is converted in the data processing unit 6, the image data is transmitted to the general communication line 4. In addition, DTE1
If the number of data in one line is less than the predetermined minimum number of data of each line of image data from, the data processing unit 6 of the DCE 2 adds 0 fill to the short line data and sets the minimum data number. And sends it out to the general communication line 4.

When the image data from the DTE 1 is processed in the DCE 2 as described above and then transmitted to the general communication line 4, the DCE 2 performs the following preprocessing in the following procedure.

The data serially transmitted from the DTE 1 via the private line 3 is stored in a buffer 5 in byte units, for example.
Is accumulated in The compression-encoded image data includes a set of line data to which a specific code called EOL (line end code) is added for each line (the number of bytes constituting each line data is not constant). The process of converting the MH encoded data to the MR encoded data, the size conversion process, and the zero fill insertion process must be performed on a line-by-line basis. Therefore, the data processing unit 6 searches the data stored in the buffer 5 for the EOL, and checks whether or not the line data to be processed is arranged in the buffer 5 (this is referred to as an EOL search process). If the line data is complete, the data is transferred to the subsequent data processing means (this is referred to as data cutout means).

The data processing unit 6 of the DCE 2 is usually constituted by using a microprocessor, and the above-mentioned EOL search processing, data cutout processing, code conversion processing, size conversion processing, zero fill insertion processing, etc. are realized by software. I have.

On the other hand, in the data communication mode, a dynamic adaptive data compression and decompression method is generally adopted, and information composed of a set of character codes is compressed and transmitted, and decompressed upon reception. As a result, the amount of communication data is reduced, and efficient communication is performed.

[0011]

One of the important roles of the DCE 2 is to effectively absorb the difference in communication speed between the private line 3 and the general communication line 4 and to perform efficient communication. Normally, the speed of data transmitted from the DTE 1 to the DCE 2 via the private line 3 is not constant.
It fluctuates according to the configuration and operation status of the TE1. When image data from DTE 1 is sent to general communication line 4 toward the other party's facsimile machine while being processed by DCE 2, input data from DTE 1 cannot keep up with the transmission speed to general communication line 4 There is. Such a situation is called an underflow. The modem 8 sends out a 0 fill in order to avoid interruption of communication due to underflow. If an underflow occurs, efficient communication cannot be performed. << Regarding the First Solution >> In the case where the image data from the DTE 1 is processed by the data processing unit 6 of the DCE 2 and transmitted to the other party's facsimile machine, the EOL search processing and the pre-processing of the data stored in the buffer 5 are performed. How the data cutout process and the process of processing the aligned line data (the above-described code conversion process and size conversion process) are executed in chronological order by the data processing unit 6 composed of a microprocessor depends on the efficiency. It is very important in achieving good communication.

In the conventional DCE 2, when data corresponding to the maximum data number MAX of one line of coded image data is input to the buffer 5, an EOL search process is executed, an EOL is found from the data, and line data is found. Had been cut out. In this method, when performing the EOL search, since there is always more than the maximum data number MAX of one line, there is always one or more EOLs, and data of at least one line can always be cut out.

However, the maximum data number MAX per line is 1
One EO, which is considerably larger than the line average data number AVE
There are many opportunities to find many EOLs in L search.
That is, a large number of lines of data are cut out by one EOL search and transferred to the next data processing. When a lot of line data is passed to the data processing means at once,
The processing is likely to be delayed. Even if the processing of the passed data is completed, the maximum data number M per line is stored in the buffer 5.
The opportunity to wait until the data is stored in the AX increases. As a result, data input processing, processing processing, and output processing could not be smoothly performed, and efficient communication could not be performed.

In order to solve the above problem, there is a countermeasure that the EOL search is repeatedly executed in a short cycle, and when one line of data is prepared in the buffer 5, the data is immediately cut out and delivered to the processing. However, in this method, the EOL search is often not performed even when the EOL search is performed, and the EOL search processing that is repeatedly executed in a short cycle is burdensome, and the entire processing from data input processing to processing and output processing is efficiently performed. Can't do it.

The present invention solves the above-mentioned problems, and
It is a first object of the present invention to provide a data processing apparatus capable of performing the entire processing smoothly and efficiently by appropriately and variably controlling the execution interval of the EOL search. The second problem of the "second solution for problems" prior art, an output processing relating to 0-fill delivery of the facsimile communication mode.

[0016]

[0017]

[0018]

[0019]

In the conventional DCE 2 (FIG. 26), the modem 8
The buffer 7 on the side stores data in units of blocks of a predetermined size. When transmitting the encoded image data to the general communication line 4, the data processing unit 6 sequentially writes the data to be transmitted to the buffer 7, and the modem 8 transmits the data to the general communication line 4 in units of one block. When the modem 8 finishes transmitting all the completed data blocks in the buffer 7, it waits for the data processing unit 6 to write the next data block to be transmitted to the buffer 7. In the meantime, the modem 8 keeps sending 0 fill to the general communication line 4 in order to avoid interruption of communication. When the communication speed of the private line 3 is lower than the communication speed of the general communication line 4, such underflow is likely to occur, and the communication time becomes unnecessarily long due to unnecessary 0-fill transmission.

Further, since the image data is configured in units of lines, the buffer 7 managed in units of blocks is used.
The break of one block does not always match the break of line data, so that the 0-fill insertion processing may be performed in the middle of one line. In this case, the image information is easily broken by the 0 fill inserted in the middle of the line, and a code error occurs.

The present invention solves the above-mentioned problems, and
It is a second object of the present invention to provide a data processing apparatus capable of reducing the chance of transmitting a 0 fill by appropriately managing an output buffer, preventing the occurrence of a code error due to the transmission of a 0 fill, and enabling efficient communication. And A third problem of the "third solving the issues" prior art, the conventional method of dynamically adaptive data compression, and decompression of data communication mode is that the processing time becomes long.

The dynamic adaptive data compression and decompression method means that the connection between input data such as characters and numerals and compression code data is sequentially changed according to the cumulative number of appearances of the input data. This is a method of assigning a short compression code. FIG. 28 shows a conventional processing procedure of this method.
29 and FIG.

FIG. 28 shows a compression conversion table.
Indicates a decompression conversion table. As shown in FIG. 28, the compression conversion table has input data 3Am (m = 0 to n-
N), the cumulative number of appearances 3Bm of the input data 3Am (m = 0
To n to N), the compression code 3Cm of the input data 3Am (m =
0 to n to N). The input data 3Am is associated with the compression code 3Cm by the cumulative appearance number 3Bm. Only the table position of the compression code 3Cm is fixed,
It is arranged from the top in the short code order. Input data 3 Am
And the cumulative appearance frequency 3Bm are paired and the table order is rearranged, and are arranged from the top in descending order of the cumulative appearance frequency 3Bm. That is, the input data 3Am having the larger cumulative appearance frequency 3Bm is associated with the shorter compression code 3Cm.

As shown in FIG. 29, the decompression conversion table includes a compression code 3Dm (m = 0 to n to N),
Fm (m = 0 to n to N), and 3Em (m = 0 to n to N), the cumulative number of appearances of the decompressed data 3Fm. Only the table position of the compression code 3Dm is fixed, and is arranged from the top in the short code order. The expansion data 3Fm and the cumulative appearance number 3Em are paired and the table order is rearranged, and are arranged from the top in descending order of the cumulative appearance number 3Em. Decompressed data 3 with a large cumulative appearance count 3Em
It is associated with a compression code 3Dm shorter as Fm.

When the input data X is to be compression-encoded, the input data 3Ax is first searched from the compression conversion table, and the corresponding compression code 3Cx is output.
Is incremented by 3Bx.
Furthermore, by rewriting 3Bx, input data 3
The pair of Am and the cumulative appearance frequency 3Bm is rearranged (accumulated appearance frequency 3Bm order).

To decompress the received data, the decompression data 3Fx corresponding to the received compression code 3Dx is retrieved from the decompression conversion table and output, and the corresponding cumulative appearance count 3Ex is incremented. By further rewriting 3Ex, the decompressed data 3Fm and the cumulative
The Em pairs are rearranged.

As described above, in the conventional method, the process of searching for the input data (compressed data on the receiving side) from the compression conversion table (decompression conversion table) is complicated and time-consuming, and the process of rearranging the tables is also complicated. There was a problem that it took time.

The present invention solves the above-mentioned problems, and
A third object is to provide a data processing device capable of performing dynamic adaptive data compression and decompression at high speed.

[0030]

According to a first aspect of the present invention, there is provided a data processing apparatus for temporarily storing a data string sequentially transmitted, and storing a line in the data stored in the buffer. Searching means for searching whether or not a specific code indicating a delimiter is included; and when the specific code is detected by the searching means, data in the buffer is read out in line units delimited by the specific code. A data extraction unit to be passed to the next data processing unit; and a search unit when the data in the buffer not processed by the data extraction unit is equal to or more than a predetermined number, and the specific code is detected by the search. Otherwise, the number of data obtained by sequentially adding an appropriate increment to the predetermined number is used as a search start condition, and the number of data is set until the specific code is detected. Those having an execution control means for retrieval processing started repeatedly search means.

[0031]

The apparatus of the second invention for achieving the second object,
A buffer memory for accumulating code data for each block unit, and transmitting data from a completed data block in the buffer memory to a communication line and completing transmission of all the completed data blocks in the buffer memory. Transmitting means for transmitting a 0 fill until the data block is completed in the buffer memory; and code data in the buffer memory. If the immediately preceding data block is being transmitted by the transmitting means, the write data is written. Data processing means for determining that the data block being written is completed even if the data amount is less than the predetermined block size and making it a transmittable block. Even if the amount of write data exceeds a predetermined block size, if the data block next to the data block being written is sending data by the sending means, the data processing means continues writing as data of the same block. It is provided with. Further, when completing the data block currently being written, the data processing means is provided for completing the data block when the last one line of data is written even if the data size exceeds a predetermined block size. When the data block being written is the next data block being sent, the data block is provided so that the data block being written is not completed until data is sent from the data block being sent.

The apparatus of the third invention for achieving the third object,
First storage means for storing a predetermined code corresponding to the input code data; second storage means for receiving the predetermined code as an address and outputting the corresponding decompressed data; And a data processing means for changing the correspondence between the input code data and the predetermined code in the first storage means at any time.

[0034]

According to the structure of the first aspect of the present invention, E is set at an appropriate interval.
If an OL search process is performed and an EOL is not found in the search, the search process is performed after the input data in the buffer has increased to some extent. The search is performed before the data is accumulated. Therefore, data of not too many lines is transferred to the next data processing means at appropriate intervals.

[0035]

According to the configuration of the second invention, even if the block size exceeds a predetermined block size, the block is completed when one line of code data is written, and if the next block is a sending block, the writing block is completed. Completes the block even if is less than the predetermined block size. Therefore, the operation of transmitting the data of the buffer is performed without delay, and the zero fill is not transmitted in the middle of the line data. Also, since the transmission of the 0 fill is stopped when data of one line is set in the buffer during the transmission of the 0 fill, the transmission time of the 0 fill can be minimized.

According to the configuration of the third aspect, the process of updating the conversion information of the compression conversion table and the expansion conversion table with the increase in the number of cumulative appearances is performed by rewriting the index, and the table search process is also simple. become.

[0038]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS << Device Configuration in Facsimile Communication Mode >> FIG. 1 is a functional block diagram in a facsimile communication mode of a facsimile adapter (data processing device) according to an embodiment of the present invention. In FIG. 1, DTE1 is a data terminal device such as a personal computer, and DCE2 is equivalent to a facsimile adapter according to the present invention. DCE2 is D
A private line 3 such as an RS-232C interface of the TE 1 is connected to a general communication line 4 such as a telephone line. In the following description, the CCITT T.D.
It is assumed that facsimile communication according to No. 30 is performed.

In the DCE 2, the input buffer 5 has a DT
The data transmitted from E1 is temporarily stored. The data cutout unit 6 searches the data in the buffer 5 for an EOL, and if there is an EOL, passes the line data delimited by the EOL to the decoding unit 7. The decoding unit 7 processes the encoded image data line by line, decodes the image data once, and provides the image data for processing such as size conversion. The line data buffer 8 is a buffer for executing decoding, encoding, and size conversion processing. The size converter 9 uses the line data buffer 8 to perform size conversion of image data. The encoding unit 10 encodes the data once decoded using the line data buffer 8 again. The data buffer 11 is a buffer for temporarily storing data processed by the encoding unit 10. The output buffer control unit 12 controls the data buffer 1
Control of writing the data of 1 into the output buffer 13 is performed.

The output buffer 13 is a buffer memory for storing data to be transmitted in a ring format. The output buffer 13 stores data in block units. The output buffer control unit 12 variably controls the block size. The modem 14 modulates the data in the output buffer 13 and sends it out to the general communication line 4, and also demodulates the data from the line 4 and stores it in the input buffer 16. The NCU 15 (network control unit) controls connection of the communication line 4. The input buffer 16 temporarily stores data received by the modem 14. The code check unit 17 sequentially reads and checks the data in the input buffer 16 and transfers the data to the buffer 18. The received data is transmitted to DTE1 via this buffer 18.

In the above configuration, the data cutout unit 6, the decoding unit 7, the size conversion unit 9, the encoding unit 10, the output buffer control unit 12, and the code check unit 17 are implemented in software by the data processing unit 19 composed of a microprocessor. Has been realized. << Embodiment of First Invention >> Assuming the configuration of FIG.
An embodiment of the first invention will be described with reference to FIG. DTE1
The encoded image data is sequentially transmitted from the
, And the data cutout unit 6 delivers the data in the buffer 5 to the decoding unit 7 line by line according to the procedure shown in the flowchart of FIG.

In FIG. 2, the data cutout unit 6 first obtains the number Dx of unprocessed (not delivered to the decoding unit 7) data stored in the buffer 5 (step A1). Next, it is determined whether or not the data number Dx is equal to or more than (Do + Δd × N) (step A2). Here, Do is a value obtained by multiplying the average value AVE of the number of data of one line by m times the average deviation σ.
Δd is a value obtained by multiplying the average deviation by n (m and n are values of about 1 to 3, respectively). Also, N
Is the count value of the counter described below, and N is 0 in the initial state.

When the number Dx of data in the buffer 5 is (Do + Δd)
× N) or more, EOL search processing (step A3)
Is executed, and if the number of data is insufficient, the process jumps to step A9. When the EOL is found by the EOL search, the process proceeds to steps A4 → A5 → A6, the counter N is cleared to 0, and the line data in the buffer 5 delimited by the EOL is marked and the decoding unit 7 processes the line. The data is delivered as data, and the process proceeds to step A9. Step A9
Is a decoding process of the decoding unit 7, a size conversion process of the size conversion unit 9, an encoding process of the encoding unit 10, an output buffer control unit 12
Is another process (data processing of the next stage) including the control process of (1).

Assuming that N = 0, when the number of data Dx in the buffer 5 becomes equal to or more than Do, an EOL search process is executed.
A7 → A8, counter N is incremented by 1
Then, it is confirmed that (Do + Δd × N) does not exceed a predetermined maximum data number MAX per line (if it does, error processing is performed), and the process proceeds to the next data processing. Then, the next EOL search is performed when the number Dx of data in the buffer 5 exceeds (Do + Δd). If this search does not find EOL,
The counter N is increased to 2 and the next EOL search is executed when the data number Dx of the buffer 5 exceeds (Do + Δd × 2). As described above, when the EOL is not detected, the reference data number for executing the search is gradually increased from the initial value Do. The increment of the reference data number is 2 for Δd.
Can be set as appropriate, such as a value obtained by multiplying by N. In step A2, the number of data D in the buffer 5
If x is less than (Do + Δd × N), the processing immediately jumps to the next data processing without performing the EOL search.

As described above, when the processing as the data cutout unit 6 (from immediately before steps A1 to A9) is performed, when the EOL search is executed with the reference data number being Do, Do is equal to the average number of data per line. Since AVE and its average deviation σ are set, line data can be cut out with a 95% probability, for example. If EOL is not detected in this search, the reference data number is set to (Do + Δd)
When EOL is executed with the number of lines increased, line data can be cut out with a 99% probability. If EOL is not detected in this search, the reference data number is changed to (D
o + Δd × 2), when the EOL search is executed, the line data can be cut out with a probability of 99.9999%. In addition, the number of lines cut out by one process is one line.

That is, the irrationality of repeating the EOL search process at an interval shorter than necessary is eliminated, and the EOL search process is executed at a longer interval.
This eliminates the inconvenience of performing OL search, cutting out data for many lines at a time, and collectively transferring the data to the next data processing. The first invention can be applied to a general facsimile machine, and has the same operation and effect as described above. << Embodiment of Second Invention >> Assuming the configuration of FIG.
According to the flowchart of FIG. 3 and the state transition diagram of FIG.
An embodiment of the invention will be described in detail. As shown in FIG. 1, the data processing unit 19 has three line buffers B0, B1,
B2 is set, and these three line buffers B
The decoding, resizing, and encoding of the input data are performed as follows using 0, B1, and B2.

In the following description, the input data of the n-th line transmitted from the DTE 1 to the buffer 5 of the DCE 2 is Da
(N), Db (n) is data obtained by decoding the data Da (n), and Dc (n) is data obtained by converting the size of the data Db (n). Also, three modulo-3 counters i, j, and k are used as pointers for specifying the three line buffers B0, B1, and B2. As is well known, the count value of the modulo 3 counter periodically changes from 0 → 1 → 2 → 0 → 1 → 2 → 0 every time it counts up. The three counters i, j, and k are set to i = 0, j = 1, and k = 2 in the initial processing. When i = 0, the notation “line buffer Bi” indicates BO. If i = 1, Bi = B1, i =
If 2, Bi = B2. Similarly, the buffer Bj is specified by the contents of the counter j, and the buffer Bk is specified by the contents of the counter k.

The processing contents of the flowchart of FIG. 3 will be described with reference to FIG.

When the image data to be subjected to decoding, size conversion and re-encoding is transferred from the DTE 1 to the buffer 5 of the DCE 2, the processing shown in FIG. 3 is started. First, in the initial processing of step 301, the line buffer B
While setting dummy reference data of 0 and B2,
From the three modulo 3, the counters are set to i = 0, j = 1, k =
Preset to 2.

Next, in step 302, the data Db (n-1) one line before the line buffer Bi = B0 is referred to (dummy data in the initial state), and the input data Da (n) on the nth line is referred to. And decrypted data Db (n)
Is set to the line buffer Bj = B1. If the n-th line is not a target of thinning by size conversion, the process proceeds from step 303 to step 304.

In step 304, the line buffer Bj
= B1 to convert the decoded data Db (n) into data Dc (n), and set the data Dc (n) in the line buffer Bi = B0. Note that the line buffer B
The reference data D b (n−1) that has entered 0 is Da
This is unnecessary since the decoding of (n) has already been completed.

It is to be noted that the data stored in the line buffer B1 is
The signal data Db (n) is directly input to the next input data Da.
(N + 1) is used as reference data for decoding.

In the next step 305 , the data Dc (n) in the line buffer Bk = B2 is referred to (dummy data in the initial state), and the data Dc (n) in the line buffer Bi = B0 is converted. Encode and output.

In the next step 306 , each modulo 3 counter i, j, k is counted up (i = 1, j = 2, k
= 0), and returns to step 302.

Then, in step 302, the decoded data Db (n) of the line buffer Bi = B1 is referred to, and (n +
1) The input data Da (n + 1) on the line is decoded, and the decoded data Db (n + 1) is set in the line buffer Bj = B2. The reference data Dc (n-1) stored in the line buffer B2 is unnecessary because the encoded output of the data Dc has already been completed.

Next, when the process proceeds from step 303 to step 304, the size of the decoded data Db (n + 1) of the line buffer Bi = B2 is converted to produce data Dc (n + 1), and the data Dc (n + 1) is converted to the line buffer Bi = B1. Set to. Data Da contained in line buffer B1
(N) is already unnecessary.

[0057]

In the next step 305 , the data Dc (n + 1) of the line buffer Bi = B1 is encoded and output with reference to the data Dc (n) of the line buffer Bk = B0.

In the next step 306 , each modulo 3 counter is counted up, i = 2, j = 0, k = 1, and the flow returns to step 302.

Therefore, the decoded data Db of Bi = B2
The input data Da (n + 2) on the (n + 2) th line is decoded with reference to (n + 1), and the decoded data Db (n +
2) is set to Bj = B0 (step 302), and Bj
= B0 and Dc (n + 2)
2), and set Bi = B2 (step 30).
4) Bi = B with reference to Dc (n + 1) of Bk = B1
2 is encoded and output (Step 3).
05) .

At step 306 , when each modulo 3 counter is counted up, i = 0, j = 1, k = 2
Returns to the initial state, and repeats the above processing. If it is determined in step 303 that the data Db (n) is a thinning target, the process proceeds to step 309, where the input data Da
(N) is taken into the line buffer Bi, and the processing for the input data Da (n + 1) of the next line is performed in step 3
Return to step 02.

As described above, according to the present invention, the three line buffers B0, B1, and B2 are reused in accordance with the progress of the processing, thereby decoding the input data, converting the size of the decoded data, The converted data can be re-encoded. In other words, it requires only one less line buffer than the conventional processing method. << Embodiment of Third Invention >> Assuming the configuration of FIG.
An embodiment of the third invention will be described in detail with reference to FIGS. In the output buffer 13 of FIG.
Numeral 4 indicates a data block currently being read, n and n + 1 indicate nth and n + 1th data blocks, respectively, and NW indicates an incomplete data block currently being written by the data processing unit 19. This output buffer 13 has at least three data blocks.

FIG. 5 shows the output buffer 1 of the data processing unit 19.
3 shows a control algorithm. As described above, the data whose size has been converted and encoded as necessary is temporarily stored in the data buffer 11 in line units (step D1). In step D2, the number of data (number of bytes) Lsiz of the line data in the buffer 11 is obtained, and compared with a predetermined minimum data number Lmin to determine whether or not Lsiz is equal to or greater than Lmin. The number of data is set to Lmin by inserting 0 fills for the shortage (step D3). Here, the minimum data number Lmin is a value determined by the following equation.

Lmim (b / 1) = transmission speed of general communication line 4 (bps) × minimum recording (transmission) time (s / 1) Further, when Lsiz is not less than Lmim, 0 fill is added as it is. Proceed to step D4. In step D4, the free space Eb of the entire output buffer 13 is calculated. If there is a capacity to receive the Lsiz code data (step D5), the output buffer control unit 12 writes the code data into the data block NW of the buffer 7 (step D7). On the other hand, if the buffer 7 does not have enough capacity to receive the Lsiz code data (step D5), it waits for a certain period of time (step D6) to check the free space of the output buffer 13 again (step D4). Until there is free space for Lsiz in the output buffer 13, the processing from step D6 to step D5 is repeated.
If free space 3 (Step D5) data processing unit 19 writes the code data in the data block N _W (step D7).

The output buffer control unit 12 controls the output buffer 1
When the code data is written in the D data block N _W of No. 3, it is determined whether or not the number of data F (N _W ) in the block N _W has reached the standard size Q of the data block (step D).
8). If the data number F (N _W ) is less than Q, the immediately preceding data block N _W ′ is looked at (step D 9), and the data block N _W ′ is the block N _R currently transmitting the code data. It is determined whether or not (Step D10). In step D9, N _W -1 = N _W + (N-1) m
od N.

[0066] Data If block N _W 'is a block NR, since immediately before the block of the block N _W during writing is a delivery in block N _R, encoded data of this still left to the left when the block N _R is empty , The block N _R is not completed, so that an underflow may occur. Therefore, even if the number of data F (N _W ) is less than the standard size Q of the data block as shown in FIG.
In one line is) is entered with long as shall thereby completing the data block N _W currently writing advancing one data block earlier (Step D13). That is, here, the data block is completed even when the number of code data F (N _W ) is less than the standard size Q. Thus, the output buffer 13 always has the data block N _R ,
Undelivered data blocks, so that at least three data blocks of the data block N _W exists. Also,
The three data blocks may be a data block N _R , a data block N _W , and an unwritten data block, but at least three data blocks always exist in the output buffer 13. In step D13, it means N _W + 1 = N _W +1 mod N.

When new writing of code data is started in the data block advanced by one, the buffer control unit 12 repeats the flow from step D11 to step D8 again. It reaches step D8, where the number of data F (N _W )
Is smaller than Q, as shown in FIG. 7, since the immediately preceding data block N _W ′ is not the transmitting block N _R but a non-transmitting data block (step D10), the output buffer processing unit 12 remains unchanged. continue the writing of code data to the current of the write data block N _W.

On the other hand, in step D8, the number of data F
(N _W) is If it is judged in Q data block N _W immediately after the write data block NW "go to see (step D11). Then, the data block N _W" block N currently encoded data sending It is determined whether it is _R (step D12). If the data block N _W ″ is the data block N _R , there is no next writable empty data block as shown in Fig. 8. In this case, the flow from step D11 to step D5 is repeated again. At step D5, if the total free space Eb of the output buffer 13 has a capacity to receive the code data of Lsiz, as shown in detail in FIGS. further continued writing of code data into data blocks NW in. that is, FIG. 10 (a) continues to write to the data block N _W, the code data from the data block N _R immediately after it represents the state of being transferred are. Next, FIG. 10 (b) in (a) state further proceeds data block N _W in the code data count F (N _W) is located reaching a Q, other de Data block N in the _R code data number F (N _R) is reduced to being transferred. Capacity entire space Eb accepts the code data Lsiz the output buffer 13 in FIG. 10 (c) In Step D5 since it is determined that there is, the output buffer control section 12 continues writing the code data beyond the Q in the space vacated in the output buffer 13 to the data block N _W. further, FIG. 10 (d) the data After the transmission of the code data in the block N _R is completed, the transmission data block is advanced by one, that is, the data block is 1
This means that the data block NW which has been continuously written even after exceeding Q is completed, and the write data block is advanced by one. Be data block all data blocks is completed in this way in the output buffer 13, if the capacity to accept the code data Lsiz in the output buffer 13, data number F in the data block N _W currently writing ( Even if (N _W ) has already reached the standard size Q, the output buffer control unit 12 continues to write the code data to the data block N _W beyond Q. Accordingly, even when there is no empty block in the output buffer 13, the data processing unit 19 does not wait for all the data blocks to be empty, so that the communication is more wasteful than when the data blocks are fixedly managed. It does not lengthen the time.

[0069] In step D12, unless the data block N _R of the data block N _W "immediately following the data block N _W currently writing NOW delivery, beyond the standard size Q as shown in Figure 9 of the last 1 When the data of the line is inserted, the data block _NW is completed and the write data block is advanced by one (step D13), whereby the break of the data block always coincides with the break of one line and one line is always EOL. , No zero fill is sent out in the middle of one line.
Since break of data block N _R during transmission as in the case of fixing the management block size of the data block does not become a middle of the last one line in the block, the next block of data block N _R unfinished Even when transmitting 0 fill in a block, 0 fill is not transmitted in the middle of one line. Therefore, there is no possibility that the image information is destroyed by the 0 fill and a code error is caused.

Next, the control operation of the facsimile communication modem 14 will be described with reference to FIG. First, modem 14 sends a retrieve the encoded data in a predetermined time interval from the transmission data block N _R of the output buffer 13 to the common communication line 4. At this time, it is determined whether or not there is untransmitted code data in the transmission data block N _R (step E1). If there is untransmitted data, the modem 14 continues to transmit the code data as it is (step E2). If there is no untransmitted code data, the data block N _R ′, which is one block ahead of the transmission data block N _R , is checked (step E3). On this occasion,
If the data block N _R ′ advanced by one is the data block N _W currently being written, that is, if it is an unfinished data block (step E4), the code data in this data block N _W is transmitted. 0
The fill is transmitted (step E5). However, since the data block _NW is completed when at least one line of code data is entered, the transmission time of the 0 fill is considerably reduced as compared with the case where the block size is fixedly managed.
On the other hand, if the data block N _R ′ advanced by one is not the data block N _W currently being written (step E
4) Since an untransmitted data block still exists in the output buffer 13, the transmitted data block N _R is advanced by one (step E6). Then, the facsimile communication modem 14 starts to transmit the code data to the general communication line 4 from the data block advanced one step (step E7).

As described above. When the data processing unit 19 completes the data block in the output buffer 13, the data block is completed or exceeds the standard size even if the number of code data F (N _W ) is less than the standard size Q. Even in such a case, by continuing writing to the data block and variably controlling the block size, the time for the modem 14 to transmit the 0 fill can be minimized. Therefore, an increase in communication time due to the zero fill can be minimized. Further, since the break of the data block always matches the break of one line and one line always ends with EOL, there is no possibility that the image information is destroyed by the zero fill and a code error is caused. Therefore, no communication error occurs due to the zero fill, and continuation of communication can be guaranteed.

Although the buffer control of image information has been described in the present embodiment, it is needless to say that the present invention can also be used for audio data used for telephones, audio equipment and the like. Also,
The same applies to control in other fields. << Other Embodiments of Third Invention >> The control for managing data blocks in the output buffer 13 has been described above. The control for managing data in line units as another embodiment of the buffering control will be described below. This will be described with reference to FIGS.

The data processing section 19 performs the following control operation as shown in FIG. The code data after the encoding process or the document size conversion process is temporarily stored in the code buffer 11 in the data processing unit 19. (Step F1).
Subsequently, the output buffer control unit 12 calculates the data amount Liz of the code data by the number of bytes, and determines whether or not the calculation result is equal to or more than a predetermined minimum data number Lmim (step F2). If Lsiz is smaller than Lmim, the buffer control unit 12
Add 0 fill until iz, Lmim to Lsiz
(Step F3). If Lsiz is equal to or larger than Lmim, the process proceeds to step F4 without adding 0 fill. In step F4, the free space Eb of the entire output buffer 13 is calculated. If there is no free space in the output buffer 13 (step F5), the process waits for a predetermined time (step F6) and checks the free space of the buffer 7 again (step F4). Usually, during this waiting time, the data processing unit 19 often performs other processing such as encoding processing or size conversion processing. On the other hand, if there is free space in the output buffer 13 (step F5), then Lsiz
It is determined whether or not there is a capacity for receiving the code data of the minute (step F7). If the output buffer 13 has no capacity to receive the code data for Lsiz, the process proceeds to step F8, and the output buffer control unit 12 moves the code data from the data buffer 11 to the output buffer 13 by the free space Eb, and reduces Lsiz by Eb. (Step F8), and the process proceeds to Step F6. On the other hand, Lsiz
If there is a capacity to receive the code data of
7), the output buffer control unit 12 moves the code data buffer 11 for Lsiz to the output buffer 13 (Step F9). As described above, when transferring the code data, the output buffer control unit 12 has moved to the position of the pointer CW of the output buffer 13. The position of the pointer CW is automatically updated every time the code data is moved as shown in FIGS. Finally, when all the data for one line is transferred to the output buffer 13, the value of the pointer CW is put into the write pointer WT as shown in FIGS. Move on to input processing. As described above, the output buffer control unit 12 does not update the write pointer WT until one line of code data has been written to the output buffer 13.

Next, the control operation of the facsimile communication modem 14 will be described with reference to FIG. First, the modem 14 extracts code data from the pointer RD of the output buffer 13 at regular intervals and sends out the code data to the general communication line 4. At this time, it is determined whether or not there is untransmitted data in the output buffer 13 (step G1). If there is no data that can be transmitted in the output buffer 13 (RD = WT), a 0-fill is transmitted to the general communication line 4 in order to avoid communication interruption (step G2). In this embodiment, if even one line is in the output buffer 13, the transmission of the 0 fill is stopped. Therefore, compared to the case where one block is completed and the transmission of the 0 fill is stopped as in the first embodiment, the transmission of the 0 fill is stopped. The sending time is shorter. Also,
Since the variable control of the block size is not performed, the data processing unit 19 can use the time for the variable control for other processing such as encoding processing or size conversion processing.

On the other hand, if there is code data in the output buffer 13 (step G1), the code data pointed to by the read pointer RD is sent to the general communication line 4 (step G3). When the code data is transmitted, the pointer RD is updated (step G4), and the process ends.

As described above, when the data processing unit 19 writes the code data in the output buffer 13, the load on the data processing unit 19 is reduced by managing the code data on a line basis instead of a block basis. If the code data in the output buffer 13 is for one line, the modem 14 does not transmit the 0 fill but transmits the code data, so that the time for the modem 14 to transmit the 0 fill can be minimized. . Therefore, an increase in communication time due to the zero fill can be minimized. Also in this case, the data break and the one-line break always coincide, and one line always starts with EOL. Therefore, no 0-fill is transmitted in the middle of one line, and image information is destroyed by the 0-fill. There is no risk of causing a code error. Therefore, no communication error occurs due to the zero fill, and continuation of communication can be guaranteed. << Embodiment of Fourth Invention >> FIG. 2 is a functional block diagram in a data communication mode of a facsimile adapter (data processing device) according to an embodiment of the present invention. in this case,
Code data of characters and numerals are transferred from the DTE 1 to the DCE 2 via the private line 3 and are sequentially stored in the buffer 20. The data in the buffer 20 is compression-encoded by the compression unit 21, temporarily stored in the output buffer 22, and
3, transmitted to the general communication line 4 via the NCU 15.
The data received from the communication line 4 is temporarily stored in the input buffer 24 via the modem 23, expanded and restored by the expansion unit 25, and
Is transferred to DTE1.

The compression section 21 and the expansion section 25 are embodied in software by a data processing section 19 comprising a microprocessor. The compression unit 21 and the expansion unit 2
The processing algorithm of No. 5 will be described in detail below.

The compression conversion table comprises an index table (a) and a compression code table (b) shown in FIG. In the index table (a), many types of input data X are arranged in a predetermined order. Here, the data X at the position of the table address m (0, 1, 2,... N) is expressed as Xm. Corresponding to each data Xm, the cumulative occurrence count 1Am of the data Xm and the ascending index 1
Bm, descending index 1Cm, and sign pointer 1D
m are set.

The cumulative appearance count 1 Am indicates the cumulative input count of the data Xm. The ascending order index 1Bm is ascending order information of the order in the table (a) of the cumulative appearance number 1Am.
Specifically, when the cumulative appearance frequency 1Ai of the data Xi is the highest, the ascending index 1Bm is set to 0. If the rank of 1Aj of the data Xj is equal to or lower than the rank of 1Ak of the data Xk, the data Xj is ranked as the ascending index 1Bj of the data Xj.
The address K of k is set.

The ascending order index 1Cm is descending order information of the order in the table (a) of the cumulative appearance number 1Am. Specifically, when the cumulative appearance count 1Ai of the data Xi is the lowest, the descending order index 1Cm is set to 0. Also,
If the rank of 1Aj of the data Xj is equal to the rank of 1Ak of the data Xk or is one rank higher, the address k of the data Xk that is lower by one rank is set as the descending index 1Cj of the data Xj.

The code pointer 1Dm corresponds to FIG.
Of the compressed code table (b). In the compression code table (b), the compression codes 1F are arranged in ascending order of code length, and the code length 1E (i) is arranged corresponding to each compression code 1F (i). That is, the compression code at the position of the table address i is 1F (i), and the code length is 1E (i). The address i of the table (b) is set as the first code pointer 1Dm.

The index table (a) as described above
The compression code 1 corresponding to the input data X is obtained by the detailed procedure described below using the
F is selected and output, and the table is updated.

(1) When data X is input, X is searched from the index table (a). If the data X is located at the table address m (X = Xm),
The corresponding code pointer 1Dm is read, and the compression code 1F at the address (1Dm) of the compression code table (b) is read.
(1Dm) is read and output. That is, the input data Xm is converted into the compression code 1F (1Dm).

(2) Cumulative number of appearances 1 of input data Xm in the previous period
Increment Am. (3) It is checked whether or not the rank of 1A in the table (a) has changed due to the increment of the cumulative appearance count 1Am in the previous period. And rewriting the descending order index 1C.

As described above, the more data X of the cumulative number of appearances
Creates a relationship corresponding to the compressed code 1F having a shorter code length.

Here, an algorithm for checking whether or not the table order changes in response to the increment of the cumulative appearance number 1 Am will be briefly described.

When 1 Am is incremented, the ascending index 1 Bm of the same address m is looked up. 1Bm =
If it is 0, 1Am is the highest order, but if 1Bm = i, it means that 1Ai is the highest order of 1Am. Therefore, the incremented 1Am is compared with 1Ai, and if 1Am exceeds 1Ai, the order is reversed. In this case, the code pointers 1Dm and 1Di
Replace with Also, the ascending index 1A and the descending index 1B are rewritten. Hereinafter, the processing procedure will be described in detail.

The operation will be sequentially described with reference to the flowchart of FIG. First, it is assumed that data Xm is newly input. In this case, the compressed code data 1F (1Dm) is extracted by the code pointer 1Dm of the data Xm. The code length 1E (1Dm), which is the length of the compressed code data, is stored in Bits,
Then, the compression code data itself is input to Code (step I1). Here, Bits indicates the number of bits such as 3 or 4 of the code length 1E (m) shown in FIGS. 19 and 20, and Code indicates 000, 0001 of the compression code 1F (m) shown in FIGS. 19 and 20. , 010 and the like. The input Code is shown in FIG.
(B) is arithmetically shifted right by RCT bits (step I2). Here, RCT indicates the number of bits of the remaining data BYT that has not been byte-aligned last time.

Next, as shown in FIG. 21 (b), the logical sum of the remaining data BYT and the code is defined as Code, and the sum of the number of bits RCT of the remaining data BYT and the number of bits Bits of the current compressed code data is calculated. The number of bits is set to RCT (step I3). As a result, the newly input compressed code data Code and the remaining data BYT are logically ORed to be byte-aligned. Further, the total number of bits RCT of the code that is ORed with the remaining data BYT
Is calculated. Here, RCT instead of Bits is an expression in which Bits corresponds to the length of each piece of compressed data, whereas RCT is the entire length of compressed code data that is not byte-aligned. It is because it expresses.

Subsequently, it is determined whether or not the RCT is greater than 15 bits (step I4). If the number of bits is larger than 15 bits, the compressed code data is sequentially output from the upper byte every two bytes (step I6).
The data lower than the upper 16 bits of the mode is shifted to the left as new remaining data BYT (step I7). On the other hand, when the RCT is 16 bits or less (step I4),
Further, it is determined whether the number is more than 7 bits (step I).
5). If there are more than 7 bits in step I5, the compressed code data is sequentially output one byte at a time from the upper byte (step I9), and as shown in FIG. BYT
And shift to the left (step I10). On the other hand, RCT
Is less than or equal to 8 bits (step I5),
Code itself is used as the remaining data BYT (step I
8).

Finally, the remainder obtained by dividing the total number of bits RCT by 8 is calculated as the number of effective bits RCT of the new remaining data BYT.
(Step I11). When the compressed code data is output using the compression conversion table of FIG. 16 as described above, the code length 1E (m) is introduced, so that the code length of the compressed code data can be made easier than in the case of using the conventional compression conversion table. And the speed of the byte alignment process can be increased.

When the above-mentioned compression code output processing is completed, the flow returns to step H1 in FIG. At the time of the processing in FIG. 18, the case where data is input in the order of B → C → E → A → B shown in FIGS. 19 and 20 will be continued.

First, when the compressed code data 001 corresponding to the input data B is output (step H1), the cumulative appearance number 1AB of the input data _B is increased by one, and is rewritten from zero to one. . That is, 1A _B =
1A _{B + 1} is set (step H2). In practice, when the cumulative number of appearances reaches a certain value, the cumulative number of appearances is normalized (Res
call) may be performed, but here, for simplicity, normalization is not described.
Next, m is stored in a variable START. Here, since the input data Xm is 1, B is put into START (step H3). Further, it is checked whether or not there is a component (input data m-1) higher than the input data Xm (step H).
4). If the input data B, it means that there is input data 1 is shown in the upper of the code pointer 1D _B is a two input data B, 1D _B = 1 is not established, the index 1Bm showing the upper component It is stored in the pointer variable NEXT. Upper component 1B of input data B _B FIG.
Since it is 0 as shown in the ascending index of (a),
0 is set in NEXT (step H5). On the other hand, if there is no higher-order component in the input data B in step H4, that is, if 1D _B = 1 holds, step H4 is executed.
Go to 9. Here, 0 is the input number of the input data A.

[0094] Next, the cumulative number of occurrences 1A _START input data START is checked whether more than the accumulated number of occurrences 1A _NEXT input data stored in the NEXT (Step H
6). In FIG. 19B, the cumulative number of appearances of the input data B is 1
A _B is 1 time, and the cumulative appearance frequency 1A of the input data _A is 0
Since more than times, the flow advances to step H7, the exchange of values between the codebook 1D _A of codebook 1D _B and the input data A of the input data B. That is, first a 2 is the value of the code pointer 1D _B placed in K, the 1D _B became empty add 1 is the value of the codebook 1D _A. Next, put in 1D _A became a 2, which had been placed in a K in the sky. As a result, FIG.
(B) As shown in FIG.
1D _B and 1D _A are rearranged in the descending order. As a method of exchanging 1D _B and 1D _A , in addition to the method shown in step H7, 1D _START = 1D _START -1, 1
The same objective can be achieved by setting D _NEXT = 1D _NEXT +1. Subsequently, NEXT is set to m (step H).
8). Here, NEXT is 0, so 0 is set to m.

Returning again to step H4, the input data B
Check if there is a component at a higher level. Input data B
Since the codebook 1D _B has a 1, as shown in FIG. _{19 (b), 1D B =} 1 is established, since the upper no component, the process proceeds to step H9.

In the following step H9, the input data m in the table is theoretically ordered by rearranging the ascending index and the descending index.

First, in step H9, the input data m
And START are checked. Here, m is 0 in step 8, and ST
Since ART is B, m and START are not the same, and the process proceeds to step H10 to continue the ascending index and descending index rearrangement processing based on the change in the rank of the input data. On the other hand, if m and START are the same, the process ends. This means that the cumulative appearance count 1 Am is 1
This means that no change occurs in the rank of the input data even if the number increases. Here, 1 is the input number of the input data B.

At step H10, it is determined whether START is the lowest order. In FIG. 19A, START
Since = 1 is not the lowest order, the process proceeds to step H11.
On the other hand, if it is the lowest, the process proceeds to step H12.

In step H11, the input data _START is removed from the logical chain, and the ascending index and the descending index of the input data before and after this are connected. First, the ascending index 1B _START is put in K. For example, FIG.
If (a) is used, 1B _START is A, so A is put in K. Thereby, the ascending index 1B of the input data B
_B is extracted from the ascending index column group. Then put the 1C _{S TART} in descending order index 1C _K. For example, 1C _K in FIG. 19 (a) is a C _A, also, 1C _START is 2, so add 2 to 1C _A. 2 is input data C
Is the input number. Then, descending index 1C _START
Into K. In FIG. 19A, 1C _START is 2, so 2 is put in K. Thereby, the extracted descending index 1B _B input data B from the descending index column group. Next, put 1B _START in ascending index 1B _K. 1BK In FIG. 19 (a) is a 1B _C, also
Since 1B _START is 0, 0 is inserted into 1B _C. As a result, the descending index 1C _A and the ascending index 1
B _C is rearranged, and the input data A and C are linked. That is, according to FIG. 19B, it can be seen that the data below the data _A has changed from 1 to 2 by the descending index 1CA. Further, the upper data of the data C by ascending index 1B _C, from 1 0
You can see that it has changed. Therefore, the storage positions of the data A and the data C in the input data string group remain fixed, and there is no data B between the data A and the data C in a logical manner, and the two have a continuous relationship.

Next, whether the input data START is the highest order
It is determined whether or not there is (Step H12). That is, 1D
_STARTIt is determined whether or not = 1. If at the top, enter
data_{S TART}Process to put the
(Step H13). That is, 1C_STARTNEX
Insert T, 1B_NEXTInto START. FIG.
In (b), 1D_START= 1 holds, so the next process
I do. First, 1C_STARTIs 1C _BNEXT is
Since it is 0 from step H5, 1C_BTo 0. Ma
1B_NEXTIs 1B_AAnd START is 1.
And 1B_APut 1 in. Thus, the table is shown in FIG.
9 (a) to 9 (b), and the input data m,
The storage positions of the code length 1E (m) and the compression code 1F (m) are
The input data m and the compression code 1F (m) remain fixed.
What is related by the magnitude of the cumulative appearance count of 1 Am
become. That is, comparing FIGS. 19A and 19B,
For example, ascending index 1A of input data A_BIs from 0
1 and the input data A is no longer at the top
And On the other hand, in step H12, S
If TART is not the highest, START is set to NE
A process of inserting between XT and m is performed (step H1).
4).

As described above, the table rewriting process when the data B is input ends. Subsequently, data C is input. Since the sign pointer 1D _C of the input data C is 3,
The third compressed code 010 from the top of the compressed code string group indicated by the number 3 is output (step H1). Since the cumulative number of occurrences 1A _C of data C will be increased one, 19
The cumulative appearance count 1A _C is rewritten from 0 to 1 as shown in (b) to (c) (step H2). Next, 2 is stored in the variable START (step H3). Further, it is checked from the code pointer 1Dm whether or not there is a component higher than the input data C (step H4). FIG. 19 (b)
In this case, the code book 1D _C is 3, and there are higher-order components, so that 1D _C = 1 does not hold (step H
4) Then, the process proceeds to step H5, where it is stored in the ascending index 1Bm pointer variable NEXT indicating the higher-order component.
That, NEXT 0 in ascending order index 1B _C
Put in.

[0102] Then the cumulative number of occurrences 1A _START input data START is examined cumulative occurrence count 1A _NEXT than or greater of the input data stored in the NEXT (Step H
6). In FIG. 19B, the cumulative number of appearances of the input data C is 1
Since A _C is one time, which is more than 0 times of the cumulative appearance number 1A _A of the input data A, the process proceeds to step H7, where the code pointer 1D _C of the input data _C and the code pointer 1D _A of the input data A are replaced. Do. Subsequently, NEXT is set to m. Here, since NEXT is 0, 0 is set to m (step H8).

Returning again to step H4, it is checked whether or not there is a component higher than the input data B. Although codebook 1D _C of the input data C is rewritten and 2 as shown in FIG. 19 (c), since the 1D _B = 1 does not hold, the process proceeds to step H5. Next, the ascending order index 1Bm is stored in the pointer variable NEXT. Here, since m is 0 in step H8, 1 which is the ascending index 1B _A is put in NEXT (step H5). continue,
The cumulative number of appearances 1A _{STAR T of the} input data START is N
Cumulative number of appearances of input data stored in EXT 1A _NEXT
It is checked whether the number is larger (step H6). Input data STA
The cumulative appearance count 1A _C of C which is RT is 1 and NE
Since the cumulative number of occurrences 1A _B of the stored data B to the XT also once, the cumulative number of occurrences 1A _C is not more than the cumulative number of occurrences 1A _B. Therefore, the process proceeds to step H9.

At step H9, input data m and S
Check whether TART is the same. Here, m is 0 in step H8, and STA
Since RT is 2, m and START are not the same, and the process proceeds to step H10. In step H10, the STA
Check if RT is lowest. In FIG. 19B, STAR
Since T = 2 is not the lowest, the process proceeds to step H11.

[0105] In step H11, remove the ascending index 1B _C and descending index 1C _C of the input data C from the logical chain of tables, linking ascending index 1B _D and descending index 1C _A input data of the front and rear. That is, first, ascending index 1B _C
Insert 0 indicated by. Next, 1C _A has descending index 1
Insert 3 which is _CC . 3 is the input number of the input data D. Subsequently, the descending index 1C is a _C 3 placed in K, add 0 is 1B _C to 1B _D. As a result, as shown in FIG.
It will be tied to the ascending index 1B _D of C _A and data D. That is, even if the storage position of the input data is fixed, the lower data of the data _A is D by the number 3 just by looking at the descending index 1C _A ,
The upper data of the data D is seen to be A by numbers 0 only see an ascending index 1B _D. Step H1
When the process in step 1 is completed, the process proceeds to step H12.

At step H12, it is checked whether or not the input data C is at the top. However, since the input data C is not at the top, the process proceeds to step H14. Step H1
4, the data START is replaced with the data NEXT and the data m
Is inserted between the and. That is, here, since the data START is 2, the data NEXT is 1, and the data m is 0, the process of assigning the data C between the data B and the data A, which have a continuous relationship in FIG. Will be done. First, ascending index 1B
Put NEXT in _START . 1B _START is 1B _C ,
Moreover, NEXT is 1, so that the add B to 1B _C as shown in FIG. 19 (c). Next, ascending index 1Bm
Into START. 1Bm is 1B _A , and
Since START is 2, put 2 to 1B _A as shown in FIG. 19 (c). Then, descending index 1C
_Insert 1C _NEXT into _START . 1C _START is 1C _C ,
Also, since 1C _NEXT is 0 as shown in FIG. 19B, if 0 is inserted into 1C _C , it becomes as shown in FIG. 19C. Further, START is entered in the descending order index 1C _NEXT . 1C _NEXT is 1C _B, also, START is 2, so add 2 to 1C _B as shown in FIG. 19 (c). Thereby, the data A in the input data string group
To C, the storage position is fixed, and the cumulative number of appearances is 1 Am
Are associated with the data B, C, and A in ascending order according to the magnitude of. For example, referring to FIG. 19 (c), the lower data of the data _B is found to be C from the number 2 when looking at the descending index 1CB, and the lower data of the data _C is represented by the number 0 by looking at the descending index 1CC. I know there is.

As described above, the table rewriting process when the data C is input ends. Subsequently, data E is input. Code pointer 1D _E of input data E Figure 19
Since (c) is 5, the fifth compressed code 1000 from the top of the compressed code string group indicated by number 5 is output (step H1). Since the cumulative number of occurrences 1A _E data E is that it has increased by 1, FIG. 19 (c) the cumulative occurrence count 1A _E as (d) is rewritten once from zero (step H2). Next, 4 is stored in the variable START (step H3). 4 is the input number of the input data E. Further, it is checked from the code pointer 1D _E whether or not there is a component higher than the input data E (step H4). FIG.
(C) the code pointer 1D _E is 5, it means that there is a component in the upper, 1D _E = 1 is not satisfied (step H4), the flow advances to step H5, ascending index 1B showing the upper component _E is stored in NEXT. That is, put 3 in ascending order index 1B _E to NEXT.

Next, the cumulative number of appearances 1A _E of the input data _E
Is greater than the cumulative number of appearances 1A _D of data D stored in NEXT (step H6). FIG.
In (d), since the cumulative appearance frequency 1A _E of the input data _E is one, which is larger than 0 of the cumulative appearance frequency 1A _D of the input data D, the process proceeds to step H7, where the sign pointer 1DE of the input data _E and the input data A process of replacing the code pointer _D with the code pointer 1DD is performed. Subsequently, NEXT is set to m. Here, since NEXT is 3, 3 is set to m (step H8).

Returning again to step H4, it is checked whether or not there is a component higher than the input data E. Since the sign pointer 1D _E of input data E is Tsu written as 4, 1D _E =
1 is not established, and the routine proceeds to step H5. Next, the ascending order index 1Bm is stored in NEXT. Here, m is since a 3 in the step H8, the 0 in ascending order index 1B _D as shown in FIG. 19 (c) Add to NEXT (step H5). Subsequently, the cumulative number of occurrences 1A _E of input data E, checks whether more than the accumulated number of occurrences 1A _A of the stored data A NEXT (step H6).
Since the cumulative appearance frequency 1A _E is one and the cumulative appearance frequency 1A _A is zero, the cumulative appearance frequency 1A _E is the cumulative appearance frequency 1
The number is larger than A _{A, and the} process proceeds to step H7 again.

In step H7, first, the code pointer 1D
The value 4 of _E is put into K, and the value 1 of the sign pointer 1D _A is put into 1D _E which has become empty. Next, put in 1D _A became the 4, which had been placed in a K in the sky. This results in cumulative occurrence count 1Am-rich order 1D _E and the 1A _A is Tsu lined conversion as shown in FIG. 28 (d). Next, NE
Let XT be m. Here, since NEXT is 0, 0 is set to m (step H8).

Returning to step H4, it is checked whether or not there is a component higher than the input data E. Since the sign pointer 1D _E of input data E is 3 and write Tsu, 1D _E = 1 is not established, the flow advances to step H5. Next, the ascending order index 1Bm is stored in NEXT. Here, since m is 0 in step H8, 2 as the ascending order index 1B _A is put in NEXT as shown in FIG. 19C (step H5). Subsequently, the cumulative number of occurrences 1A _E of input data E, checks whether more than the accumulated number of occurrences 1A _C of the stored data C in the NEXT (step H6). Cumulative number of occurrences 1A _E is once, since the cumulative number of occurrences 1A _A also once a equal to the cumulative occurrence count 1A _E and the accumulated occurrence counts 1A _A, the process proceeds to step H9.

At step H9, input data m and STA
Check whether the RT is the same. Here, m
Is 0 in step 8, and START is 4
Therefore, m and START are not the same, and the process proceeds to step H10. In step H10, it is checked whether START is the lowest. In FIG. 19C, START =
Since 4 is the lowest order, the process proceeds to step H12.

At step H12, it is checked whether or not the input data C is at the top. However, since the input data E is not at the top, the process proceeds to step H14. Step H1
In step 4, a process of inserting data E between data C and data A is performed. First, N in ascending index 1B _START
Insert EXT. 1B _START is 1B _E, also, NE
Since XT is 2, 1B _E as shown in FIG.
Put 2 in. Next, it is STAR to ascending index 1Bm.
Insert T. Put START in 1Bm. 1Bm is 1
Since B _A and START are 4, FIG.
Insert 4 into 1B _A as shown in (d). Subsequently, 1C _NEXT is entered in the descending order index 1C _START . 1C _START
A is 1C _E, also because 1C _NEXT, as shown in FIG. 27 (c) is 0, put a 0 to 1C _E Figure 28 (d)
It becomes as shown in. Furthermore, descending index 1C _NEXT
Into START. Since 1C _NEXT is 1C _C and START is 4, 4 is inserted into 1C _C as shown in FIG. As a result, the storage positions of the data A, C, and E in the input data string group remain fixed, and the data C,
E and A will be related in this order. For example, FIG.
0 (d) indicates that the lower data of the data C is E by the number 4 when viewed from the descending index 1C _C , and that the lower data of the data _E is A by the number 0 when viewed by the descending index 1C _E. You.

As described above, the table rewriting process when the data C is input ends. Subsequently, data A is input. The sign pointer 1D _A of the input data A is shown in FIG.
Since (d) is 4, the fourth compressed code 011 from the top of the compressed code string group indicated by the number 4 is output (step H1). That is, in the initial state of FIG. 19A, the compression code corresponding to data A is the first 00 in the compression code string group.
It can be seen that the value of 0 has changed according to the cumulative number of appearances of 1 Am.

[0115] Since the cumulative occurrence count 1A _A data A is to have increased one rewrites FIG 20 (d) the cumulative occurrence count 1A _A as (f) in once from zero (Step H2) . Next, 0 is stored in START (step H3). Furthermore, it examined from the code pointer 1D _A whether or not there is a component to the upper than the input data A (step H4). In FIG. 20 (d), the code pointer 1D _A is 4, and the ascending index 1B _A indicating the higher-order component is stored in NEXT. That is, ascending index 1
Since B _A is 4 in FIG. 20D, 4 is stored in NEXT (step H5). Cumulative occurrence count 1A _A of the next input data A is checked whether more than the accumulated number of occurrences 1A _E data E stored in NEXT (Step H
6). In FIG. 20F, the cumulative number of appearances of the input data A is 1
Since A _A is once and the cumulative appearance number 1A _E of the input data E is also one, both are the same, and the process proceeds to step H9.

In step H9, the input data m and S
Check whether TART is the same. Here, since m remains 0 and START is also 0, m and START are the same, and the data A is input without any change in the ascending index 1Bm and the descending index 1Cm. The table rewriting process is ended as it is.

Finally, the second data B is input. Code pointer 1D _B input data B is because in Figure 20 (f) from 1, and outputs a first compressed code 000 from the top of the compression code string group indicated by the number 1 (step H1). That is, unlike the first data B, the compression code 000 in accordance with the magnitude of the cumulative occurrence count 1A _B 001
Has changed to This is a feature of the dynamic adaptive data compression method. Since the cumulative number of occurrences 1A _B data B is increased by one, FIG. 20 (f) the cumulative occurrence count 1A _B as (e) rewriting the once to twice (Step H
2). Next, 1 is stored in START (step H).
3). Further, it is checked from the sign pointer 1D _B whether or not there is a component higher than the input data B (step H).
4). A diagram 20 (f) the code pointer 1D _B 1, since there is no component to the upper, 1D _E = 1 is satisfied (step H4), the flow advances to step H9. That is, here, the code book 1Dm does not need to be rewritten.

At step H9, input data m and STA
It is checked whether RT is the same or not. Here,
Since m and START are both 1, m and START
And the process of rewriting the table when the data B is input ends. The compression code data is 00
That is, the data is output to the general communication line 4 in the order of 1 → 010 → 1000 → 011 → 000.

As described above, when the compression encoding process of the input data m is performed using the dynamic adaptive data compression method, the input data sequence group, the cumulative appearance frequency sequence group,
Since the storage positions of the code length sequence group and the compressed code sequence group are fixed, it is easy to search out data matching the input data m, and the search speed is reduced. Further, the correspondence between the input data m that changes according to the cumulative number of appearances 1Am and the compression code 1F (1Dm) only needs to be exchanged with the value of the codebook sequence group. It is only necessary to exchange the values of the ascending index column group and the descending index column group, thereby making it easy to convert the input data m into the compression code 1F (1Dm), and to shorten the processing time.

In addition, when the code length 1E (m) is introduced as shown in the compression conversion table of FIG. 16, the compression code 1F (m)
Is output in a byte-aligned manner, the number of bits can be managed reliably and easily, and the byte alignment of the bit string of the compression code 1F (m) can be speeded up. Note that the item of the code length 1E in the table (b) is not always necessary.

Hereinafter, a method of expanding the code data compressed by the above method on the receiving side will be described with reference to FIG. 15 and FIGS.

In FIG. 22, this conversion table is composed of a decoding table (a) and an expanded data table (b). Decoding table (a) compressed code string group (m =
0-N) and decompressed data (index) column group (2B
m, m = 0 to N), and the decompression data table (b) has a cumulative appearance frequency sequence group (2A (m), m = 0 to N).
N) and a decompressed data string group (2 Cm, m = 0 to N). The compression code m corresponds to 1F (m) in FIG. 2, and the decompressed data (index) 2Bm is an index pointing to the decompressed data 2C (2Bm) corresponding to the compression code m. Cumulative appearance frequency 2A (2Bm)
Indicates the cumulative number of appearances of the decompressed data 2C (2Bm), and the decompressed data 2Cm is data obtained by decompressing the compressed code data and restoring it to the original data to be compressed. Note that the storage positions of the elements of the compression code string group, the cumulative appearance number string group, and the decompressed data string group are fixed. Decompressed data (index) 2
Based on Bm, the expansion data 2C (m) is ranked according to the magnitude of the cumulative appearance count 2A (m).

In FIG. 23, the conversion table is 0-2.
The data represented by 25 ASCII codes are to be decompressed. Therefore, the compressed code string group (m = 0 to 2)
25), decompressed data (index) column group (2Bm, m
= 0 to 225), the cumulative appearance frequency sequence group (2A (m), m =
0 to 225). It should be noted that the cumulative appearance frequency sequence group (2A (m), m = 0 to 225) is provided. It is assumed that the cumulative appearance frequency 2A (m) satisfies the following relationship.

2A (2B ₀ ) ≧ 2A (2B ₁ ) ≧ 〜 ≧ 2A
(2B _n ) ≧ 〜 ≧ 2A (2B ₂₂₅ ) In FIG. 24, the compression code m is five data (000,
001, 010, 011 and 1000), and the restored decompressed data are A, B, C, D and E. Further, 1 to 5 are numbers assigned to the storage positions of the respective elements of the decompression data table (b) in FIG.

The procedure for expanding received data will be described with reference to the flowchart shown in FIG. First, the compressed code data transmitted by the general communication line 4 is received by the data communication modem 23 and is temporarily stored in the input buffer 24. The data processing unit 19 takes out the compressed code data from the input buffer 24, converts it into character data such as ASCII code, and outputs it to the buffer 26. When the decompression unit 25 in the data processing unit 19 decompresses the compressed code data, first, the compression code data is specified by the bit string as to which compression code data m in the compression code string group, and then the compression code data corresponding to the compression code data m is specified. The expanded data 2Bm to be transferred to the buffer 26
And the cumulative number of appearances 2A (2B
m) has been updated. That is, the compression code data is set to 00
24, when the decompressed data B corresponding to the compressed code data 001 is output as shown in FIG. 24A (step J1), the cumulative appearance number corresponding to the storage position of the cumulative appearance number sequence group indicated by the number 2 is obtained. 2A _(B) is increased by one (step J2). As shown in FIG. 24 (b), the cumulative number of appearances 2A
_(B) is changed from 0 times to 1 time, and the process proceeds to step J3.

In step J3 and thereafter, the expanded data 2Bm
The storage position of each element in the expanded data (index) column group is rearranged so that the compression code m corresponds to the expanded data 2Bm in accordance with the magnitude of the cumulative appearance frequency 2A (2Bm).

First, in step J3, the compressed code data m
Is checking to see if it has reached the top of the table. If the compressed code data m is the highest-order element, there is no change in the table, and the process ends.
On the other hand, if the compression code m is not the highest-order element, the process proceeds to step J4. Since the compression code data 001 is not the highest-order element, the process proceeds to step J4.

At step J4, the compressed code data m
Corresponds to the cumulative number of appearances 2A of the expanded data 2Bm
(2Bm) and the cumulative appearance count 2A (2B
m-1) to determine whether the cumulative number of appearances 2A (2Bm) is greater than the cumulative number of appearances 2A (2Bm-1). The cumulative appearance frequency 2A (2Bm) is the cumulative appearance frequency 2A (2Bm
If not more than -1), the updating of the table contents is terminated as it is. On the other hand, if the cumulative number of appearances 2A (2Bm) is greater than the cumulative number of appearances 2A (2Bm-1), step J
Proceeding to step 5, the respective elements in the expanded data (index) column group are rearranged. In FIG. 24B, the compressed code data 0
Since the cumulative appearance count of the decompressed data B corresponding to 01 is larger than the accumulated appearance count of the decompressed data A corresponding to the compression code data 000 which is higher by one, the step J
Go to 5.

In step J5, first, the expanded data 2Bm is put into K, and the storage position of the expanded data 2Bm in the expanded data string group is made empty. Next, the empty decompressed data 2B
m and the expanded data 2Bm-1
Decompressed data 2 stored in K at the storage location of Bm-1
Add Bm. The compressed code data m-1 is set as the compressed code data m (step J6), and the process returns to step J3. As a result, each element of the decompressed data string group has a cumulative appearance frequency of 2A
This means that they are rearranged according to the magnitude of (m). That is, the use of FIG. 24 (a), K put the decompressed data B First, the putting decompressed data A to the extended data 2B _B. Then add decompressed data B that has been stored in the K in the decompressed data 2B _A. This means that the storage positions of the decompressed data A and the decompressed data B in the decompressed data string group have been switched as shown in FIG. Subsequently, the compression code data 000 is set as the compression code data m, and the process returns to step J3.

Returning to step J3, the compressed code data 0
Since 00 is the highest-order element, the processing ends.

Thus, the processing when the compressed code data 001 is restored to the decompressed data B is completed.

Next, the compression code data is specified as 010,
As shown in FIG. 24B, decompressed data C corresponding to the compressed code data 010 is output (step J1). Since the cumulative number of appearances 2A ( _c ) of the decompressed data C is increased by one, the value of the storage position of the group of cumulative number of appearances indicated by the number 3 is replaced. That is, as shown in FIG. 24C, the cumulative appearance number 2A ( _c ) is updated from 0 to 1 (step J).
2) Go to step J3.

At step J3, the compressed code data 0
It is checked whether 10 is at the top of the table. Since the compression code data 010 is not the highest-order element, the process proceeds to step J4.

At step J4, as shown in FIGS. 24 (b) and (c), the cumulative appearance count of the decompressed data C corresponding to the compressed code data 010 is changed to the next higher compressed code data 001 by one. Since the number of corresponding appearances of the decompressed data A is greater than 0, the process proceeds to step J5.

[0135] Proceeding to step J5, K put the decompressed data C first, empty the storage location of the decompressed data 2B _c decompression data sequence group. Next, place the decompressed data A to the extended data 2B _c emptied, then add decompressed data C that has been stored in K in the storage position of the expanded data 2B _A. This means that the storage positions of the decompressed data A and the decompressed data C in the decompressed data string group have been switched as shown in FIG. Subsequently, the compressed code data 001 is set as the compressed code data m (step J6), and the process returns to step J3.

Returning to step J3, the compressed code data 0
Since 01 is not the highest-order element, the process proceeds to step J4. At step J4, as shown in FIG. 24 (c), one occurrence of the cumulative appearance of the decompressed data C corresponding to the compressed code data 001 corresponds to the decompressed data B corresponding to the next higher compressed code data 000. Is the same as the cumulative number of appearances of 1. Therefore, the updating of the table contents is terminated as it is.

Thus, the processing when the compressed code data 010 is restored to the decompressed data C is completed.

Subsequently, the compression code data is specified as 1000, and as shown in FIG.
Is output (step J1).
Since the cumulative number of occurrences 2A of expanded data E _(E) is increased by one, the cumulative number of occurrences 2A _(E) pointed to number scan as shown in FIG. 35 (d) is updated to once from zero (step J
2) Go to step J3.

Proceeding to step J3, the compressed code data 1
000 is not the top of the table, so step J
Proceed to 4.

At step J4, as shown in FIGS. 24 (c) and (d), the cumulative appearance frequency of the decompressed data E corresponding to the compressed code data 1000 is changed to the next higher compressed code data 011 by one. Since the cumulative number of appearances of the corresponding decompressed data D is greater than 0, the process proceeds to step J5.

At step J5, the expanded data E is stored in K.
Placed, empty the storage location of the decompressed data 2B _E decompression data sequence group. Next, place the decompressed data D to the extended data 2B _E emptied, then add decompressed data E that has been stored in K in the storage position of the expanded data 2B _D. This means that the storage positions of the decompressed data D and the decompressed data E in the decompressed data string group have been switched. Subsequently, the compression code data 011 is set as the compression code data m, and the process returns to step J3.

Returning to step J3, the compressed code data 0
Since 11 is not the top element, the process proceeds to step J4.

At step J4, the compressed code data 0
11 is larger than the cumulative appearance number 0 of the decompressed data A corresponding to the compression code data 010, which is one higher order, so that step J5 is performed.
Proceed to.

At step J5, the expanded data E is stored in K.
Placed, empty the storage location of the decompressed data 2B _E decompression data sequence group. Next, place the decompressed data A to the extended data 2B _E emptied, then add decompressed data E that has been stored in K in the storage position of the expanded data 2B _A. This means that the storage positions of the decompressed data A and the decompressed data E in the decompressed data string group have been switched as shown in FIG. Subsequently, the compression code data 010 is set to the compression code data m (step J6), and the process returns to step J3.

Returning to step J3, the compressed code data 0
Since 10 is not the highest element, the process proceeds to step J4.

At step J4, as shown in FIG. 24D, the cumulative appearance count of the decompressed data E corresponding to the compressed code data 010 is one higher than the compressed code data 0
Since the cumulative number of times of decompression data C corresponding to 01 is the same as one, the updating of the table contents is terminated as it is.

Thus, the processing when the compressed code data 1000 is restored to the decompressed data E is completed.

Subsequently, the compressed code data is specified as 011 and the expanded data A corresponding to the compressed code data 011 is output as shown in FIG. 24D (step J1). Since the cumulative number of occurrences 2A extension data A _(A) is increased by one, 24 the accumulated occurrence counts 2A pointed to number service as (e) shows _(A) is updated to once from zero (step J
2) Go to step J3.

Proceeding to step J3, the compressed code data 0
Since 11 is not the top of the table, step J4
Proceed to.

At step J4, as shown in FIGS. 24 (d) and (e), the cumulative appearance frequency of the decompressed data A corresponding to the compressed code data 011 is changed to the next higher compressed code data 010 by one. Since the number of times of the corresponding appearance of the decompressed data E is equal to one, the updating of the table contents is terminated as it is.

Thus, the processing when the compressed code data 011 is restored to the decompressed data A is completed.

Finally, the compressed code data is specified as 000, and the expanded data B corresponding to the compressed code data 000 is output as shown in FIG. 24 (e) (step J1). Since the cumulative number of appearances 2A ( _B ) of the decompressed data B has increased by one, the cumulative number of appearances 2A ( _B ) as shown in FIG.
Is updated from once to twice (step J2), and the process proceeds to step J3.

Proceeding to step J3, the compressed code data 0
Since 00 is the top of the table, the process ends without any change in the table as shown in FIGS. Thereby, the compressed code data 00
The process when 0 is restored to the decompressed data B ends. The decompressed data is restored in the order of B, C, E, A, B, and DTE1
Is forwarded to.

As apparent from the above description, the received compressed code data 001, 010, 100, 001, 100
0 indicates that the input data B, C, E, A, and B, which are decompressed as decompressed data B, C, E, A, and B and are compression-encoded by the dynamic adaptive data compression method, are successfully reproduced on the receiving side. I understand. By performing the decompression process in this manner, data can be reliably reproduced when the dynamic adaptive data compression method and the dynamic adaptive data decompression method are used, and the compression conversion table is used to store the input data sequence group. Since the storage position of each element is fixed and the storage position of each element of the compression code string group is fixed in the decompression conversion table, the processing speed can be increased. Further, when performing the expansion processing of the compression code m by the dynamic adaptive data expansion method, the association between the compression code m to be changed by updating the cumulative appearance number 2A (m) and the expansion data 2C (m) and the expansion data (index) is performed. 2Bm, and the decompressed data (index) 2Bm is 2C with respect to the decompressed data 2C (2Bm).
When (2Bm) = 2Bm, the decompressed data (index) 2Bm can be used as decompressed data as it is, so that the processing time can be shortened accordingly.

What is important when restoring data compressed and encoded by the dynamic adaptive data compression method to the original data is the same in the dynamic adaptive data compression method and the dynamic adaptive data decompression method. That is, no algorithm is used. That is, when the data is compression-encoded by the compression conversion table of FIG. 18, the cumulative number of appearances of the data m currently processed in step A5 and thereafter is 1 Am
And input data m + whose cumulative appearance count is one more than data m
1 or the cumulative number of appearances 1A _NEXT of the input data m having the same cumulative number of appearances, the codebooks 1Dm and 1D
A process is performed to replace or leave _NEXT . On the other hand, when the data is expanded by the expansion conversion table of FIG. 25, the cumulative number of appearances 2A (2Bm) of the data m currently being processed in step C4 and the subsequent steps and the data of the storage position one higher order in the compression code string group By comparing the cumulative number of appearances 2A (2Bm-1) of m-1 with the expanded data (index) 2Bm and 2Bm corresponding to the codebook
A process of exchanging −1 with or keeping the same is performed. Here, as in the case of the compression code processing, a comparison is made between the cumulative appearance number 2Am and the input data m + 1 whose cumulative appearance number is one more than the data m or the cumulative appearance number 2A _NEXT of the input data m having the same cumulative appearance number. Not.
If the same algorithm as the compression code processing is used in the decompression processing, the result will be erroneously decoded as B, C, E, E, and the compressed code data will not be restored to the original data. As described above, by using separate algorithms for data encoding and decoding, it is possible to eliminate decoding errors and process data at high speed.

[0156]

As described above in detail, according to the first aspect, in the process of cutting out the encoded image data sequentially stored in the buffer in units of line data by a specific code and delivering it to the next stage data processing means, The inconvenience of increasing the load on the data processing unit (microprocessor) by repeating the search processing of the specific code at an interval shorter than necessary is eliminated. Are cut out and delivered collectively to the next stage of data processing, thereby eliminating the inconvenience of burdening the next stage of processing and increasing the waiting time of the next stage of processing.

[0157]

According to the second aspect of the present invention, even if the block size exceeds a predetermined block size, the block is completed when one line of code data is written, and if the next block is a sending block, the writing block is Since the block is completed even if the size does not reach the predetermined block size, the operation of transmitting data in the buffer is performed without delay, and it is possible to prevent a decrease in communication efficiency due to the transmission of 0 fill. Similarly, since the transmission of the 0 fill is stopped when data of one line is set in the buffer during the transmission of the 0 fill, the transmission time of the 0 fill can be minimized, and the communication efficiency can be improved. Can be. Also, it is possible to eliminate the transmission of 0 fill in the middle of the line data,
It is possible to prevent the destruction of the image information due to the transmission of the 0 fill in the middle of the line, and to guarantee the continuity of the communication without causing a code error.
If data is managed in units of lines when writing data to the output buffer, the data processing unit only needs to perform processing of not updating the write pointer unless data for one line is written, and manages data in units of blocks. The processing load is lighter than in the case.

According to the third aspect of the present invention, in a method of performing compression processing by assigning shorter compression code data to input data having a higher cumulative appearance number using the compression conversion table, the input data, the cumulative appearance number thereof, and the compression code Since the storage position of the data is fixed, and a compression conversion table having a code index for linking the input data and the compressed code data and an index indicating the ranking of the input data, and the storage position of the input data is always fixed, Processing for retrieving target input data from the table is facilitated, and the compression processing time is reduced. Also, the correspondence between the input data and the compressed code data, which fluctuates according to the cumulative number of appearances, is sequentially updated by rewriting the code index, and further, the ranking of the input data is sequentially updated by rewriting the index. Update processing is also easy.

In addition, by introducing code length information indicating the number of bits of the compressed code data into the compression conversion table, the number of bits when the compressed code is byte-aligned and output can be easily and reliably managed. Therefore, it is possible to speed up the process of arranging the compression codes.

Similarly, in a system in which shorter compressed code data is assigned to the expanded data having a larger number of cumulative appearances and restored by using the expansion conversion table, the storage positions of the expanded data and the compressed code data are fixed. and compressed code de - data and an extended de - extension conversion Te has a stretch index linking the data - using a table, extension de - since the always fix the storage position of the data, decompression conversion Te - purposes in Bull Search processing of expanded data of
Speed is reduced. Also, the expansion index and the expansion data
If data and are equivalent is directly decompressed de stretch index - it is possible to use as a data decompression de -
The table can be simplified by removing the data from the expansion conversion table, and the processing time can be further reduced.

According to the above inventions, efficient and reliable communication can be realized when communicating with a facsimile apparatus from a personal computer or the like or when communicating with another personal computer.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram showing an outline of a data processing apparatus according to an embodiment of the present invention in a facsimile communication mode.

FIG. 2 is a flowchart showing an embodiment of a control procedure of an EOL search and data cutout processing of the present invention.

FIG. 3 is a flowchart showing a buffer selection process at the time of size conversion in the present invention.

FIG. 4 is a state transition diagram of a buffer in the operation of FIG. 3;

FIG. 5 is a flowchart showing buffering control of the data processing unit of the present invention.

6 is a memory state diagram showing a buffer accumulation state in steps D10 to D13 in FIG.

FIG. 7 is a memory state diagram showing a state of a buffer at steps D10 to end of FIG. 5;

FIG. 8 is a memory state diagram showing the state of the buffer in steps D12 to end of FIG. 5;

FIG. 9 is a memory state diagram showing a state of a buffer in steps D12 to D13 of FIG. 5;

FIG. 10 is a state transition diagram showing a state transition of an output buffer in FIG. 8;

FIG. 11 is a flowchart showing a control operation of the modem of the present invention.

FIG. 12 is a flowchart showing another buffering control of the present invention.

13 is a state transition diagram showing the state transition of the output buffer under the control of FIG.

FIG. 14 is a flowchart showing another control operation of the modem of the present invention.

FIG. 15 is a schematic block diagram illustrating an outline of a data processing device according to an embodiment of the present invention in a data communication mode.

FIG. 16 is a table diagram showing a compression conversion table for converting input data used in the present invention into encoded data;

FIG. 17 is a flowchart showing a compression code output process;

FIG. 18 is a flowchart showing a table updating process accompanying output of a compression code;

FIG. 19 is a state transition diagram of a table exemplifying transition of the contents of the compression conversion table shown in FIG. 16;

FIG. 20 is a state transition diagram showing transition contents subsequent to the transition of the contents of the table shown in FIG. 16;

FIG. 21 is an explanatory diagram showing a data flow at the time of byte alignment processing in FIG. 17;

FIG. 22 is a table diagram of a decompression conversion table used when decompressing code data in the embodiment of the present invention.

23 is a table diagram showing another table configuration in the extension conversion table shown in FIG. 22;

FIG. 24 is a state transition diagram of a table exemplifying the transition of the contents of the expansion conversion table of FIG. 23;

25 is a flowchart showing a control operation of a decompression process of a code data processing unit performed by the data processing unit 19 in FIG.

FIG. 26 is a schematic block diagram schematically showing a conventional data processing device.

FIG. 27 is a state transition diagram showing decoding and encoding processing at the time of conventional size conversion.

FIG. 28 is a table diagram showing a conventional table for converting input data into a compression code;

FIG. 29 is a table showing a table for restoring conventional compression codes to decompressed data;

[Explanation of symbols]

 Reference Signs List 1 DTE 2 DCE 3 Private line 4 General communication line 5 Buffer 6 Data cutout unit 7 Decoding unit 8 Data buffer 9 Size conversion unit 10 Encoding unit 11 Data buffer after processing 12 Output buffer control unit 13 Output buffer 14 Facsimile transmission modem 15 NCU 16 input buffer 17 code check unit 19 data processing unit 21 compression unit 22 output buffer 23 data communication buffer 24 input buffer 25 decompression unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Nobuhiko Noma 2-3-8 Shimomeguro, Meguro-ku, Tokyo Inside Matsushita Electric Transmission Co., Ltd. (72) Osamu Noguchi 2-3-8 Shimomeguro, Meguro-ku, Tokyo No. Matsushita Electric Transmission Co., Ltd. (72) Inventor Sakae Miyamoto 2-3-8 Shimomeguro, Meguro-ku, Tokyo Matsushita Electric Transmission Co., Ltd. (56) References JP 4-324759 (JP, A) JP 58-175342 (JP, A) JP-A-3-102967 (JP, A) JP-A-2-75250 (JP, A)

Claims (11)

(57) [Claims] 1. A buffer for temporarily storing a data string sequentially transmitted, and a detection for searching whether or not a specific code indicating a line break is included in the data stored in the buffer. Means for reading out the data in the buffer in units of lines delimited by the specific code when the specific code is retrieved by the detection means, and transferring the data to the next-stage data processing means; When the number of data in the buffer that has not been processed by the cutout means is equal to or more than a predetermined number, the search means is started, and when the specific code is not detected by the search, an appropriate increase to the predetermined number is performed. And execution control means for a search process for repeatedly activating the search means until the specific code is detected, using the number of data sequentially added as search start conditions. A data processing device characterized by the following. 2. The data processing apparatus according to claim 1, wherein a predetermined number is set based on an average number of data of one line of data input to the buffer and a standard deviation thereof. 3. A buffer memory for storing code data for each block, transmitting data from a completed data block in the buffer memory to a communication line, and transmitting a completed data block in the buffer memory. When all data has been transmitted, transmitting means for transmitting 0 fills until the data block is completed in the buffer memory, and code data are written in the buffer memory, and the immediately preceding data block is being transmitted by the transmitting means. Data processing means for determining that the data block being written is completed even if the amount of write data is less than a predetermined block size and making the data block a transmittable block. apparatus. 4. The data processing apparatus according to claim 3, wherein said data processing means completes the data block being written at the stage of writing one line of code data. 5. A buffer memory for storing code data for each block, transmitting data from a completed data block in the buffer memory to a communication line, and transmitting a completed data block in the buffer memory. A transmission means for transmitting 0 fills until the data block is completed in the buffer memory when all data has been written, and writing coded data in the buffer memory, and writing even if the write data amount exceeds a predetermined block size. The next data block following the data block in the middle is transmitted by the transmitting means.
A data processing unit that continues to write to the data block as long as the total capacity of the buffer memory permits, even if the data exceeds a predetermined block size during data transmission. 6. The data processing means, if the data block next to the data block being written is a data block being sent, repeats the data block being written until the data is completely sent from the data block being sent. 6. The data processing device according to claim 5, wherein the data processing device is not completed. 7. A buffer memory for storing code data for each block, transmitting data from a completed data block in the buffer memory to a communication line, and transmitting a completed data block in the buffer memory. A transmission means for transmitting 0 fills until the data block is completed in the buffer memory when all the data blocks have been completed, and a code data being written in the buffer memory and exceeding a predetermined block size when completing the data block currently being written. The data processing device further comprises: data processing means for determining that the data block has been completed when the last one line of data has been written, and making the data block a transmittable block. 8. A first storage means for storing code data into which predetermined data is converted, and said predetermined data is stored at a predetermined storage location and indicates an address of said first storage means. Second storage means for storing a code index in association with the predetermined data; and rewriting the code index in accordance with the cumulative number of occurrences of the predetermined data to shorten the predetermined data in the order of the cumulative number of occurrences. Data processing means for allocating code data; and transmission means for transmitting the code data obtained from the first storage means to a communication line under the control of the data processing means. A data processing apparatus characterized by the above-mentioned. 9. An ascending index indicating an address of predetermined data whose cumulative appearance count is higher or equal to one in the second storage means, and the data processing means accumulates the first predetermined data. When the number of appearances increases by one, the ascending order index of the first predetermined data is searched to determine the second predetermined data whose cumulative appearance number is one higher, and the first predetermined data is determined. Is compared with the cumulative number of appearances of the second predetermined data, and when the cumulative number of occurrences of the first predetermined data is larger, the code index of the first predetermined data and the second
9. The data processing apparatus according to claim 8, wherein the code index of the predetermined data is rewritten to change the arrangement relationship of the code indexes. 10. A first storage means for storing decompressed data of each input code data in a predetermined order in combination with the cumulative number of appearances of the decompressed data, and fixing each of the input code data in a predetermined order. A second storage means for storing, in correspondence with each of the input code data, an index indicating an address of the first storage means and associating which cumulative appearance number corresponds to which input code data; Data processing means for rearranging the index in accordance with the cumulative number of appearances of the data and allocating short input code data to decompressed data having the largest cumulative appearance number. Processing equipment. 11. A first storage means for storing a fixed number of cumulative occurrences of each input code data in a predetermined order, and an address of said first storage means for storing said input code data fixed in a predetermined order. And a second storage means for storing an index for associating which cumulative appearance number corresponds to which input code data in association with each of the input code data, and according to the cumulative appearance number of the input code data. A data processing means for rearranging the index, wherein the index is constituted by code information indicating decompressed data, and the input code data is restored based on the index.