JPH0311883A - Decoding system for variable length code, and facsimile equipment and still picture transmission system - Google Patents

Abstract

PURPOSE:To reduce the retrieval time and to decrease the capacity of a decoding table by retrieving a decoded data from a variable length code at a batch. CONSTITUTION:A decode table group is provided, in which a variable length code group and its relevant decoding information are listed while being classified with a consecutive number of the same bit from the head or a variable length code, and as to the variable length code to be decoded, a relevant code is selected from the decode table group classified with a consecutive number of the same bit from the head and the decode table retrieved by a remaining code of the relevant variable length code except the consecutive part to obtain decoded information. For example, consecutive '0s' bits number is counted from the head of, e.g. an MH code, a relevant code is selected from the decode table group classified in advance and the decode table is retrieved by remaining code except consecutive '0' bit and a bit '1' stopping the consecutive '0' bits.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for decoding variable length codes such as run-length codes used in facsimile machines, etc.
(Modified Hoffman) Concerning a decoding method suitable for speeding up code decoding processing and reducing the size of the code. A run-length code encoded based on the probability of occurrence is used.

This run-length code has a variable word length, and an MH code is used as the -dimensional encoding method.

FIG. 2 shows the codeword length of the MH code used in facsimile machines.

Conventionally, there has been a method for decoding such an MH code.

A common method is to prepare a decoding table in advance and search the table corresponding to the MH code. An example of such a table search method is a bit sequential search method in which the MH code to be decoded is searched bit by bit and a table is searched to determine whether the run length is determined. Figures 3 and 4 show the MH code decoding method using this bit sequential search method.
This will be explained using figures.

FIG. 3 is a tree-like diagram showing a decoding table for white run (a) and black run (b) of the MH code used in facsimile. Further, FIG. 4 is a diagram showing a part of an example of a decoding table for an MH code using the bit sequential search method. The table tree in FIG. 3 shows that Ml (code rQJ or rl) is branched until the run length is determined. FIG. 4 shows an example in which a decoding table corresponding to this table tree is arranged, for example, on a ROM.

In the decoding of the MH code using the decoding table shown in 2, rQJ is decoded bit by bit from the first bit of the MH code.
or “1”, and if “Oj”, it is an even address, “1”
In the case of , the decoding table corresponding to the detection result is sequentially searched, such as an odd address, and the bit search is repeated until the run length is determined.

For example, M H code ro corresponding to black run length "6"
In the decoding of 010j, the first bit is "rQ", so the data at position 0 of the black run decoding table shown in FIG. 4 is searched. Note that if the bit is "1", it is called address 1, and so on, and is divided into even addresses and odd addresses depending on the value of the bit.

The decoding table contains a flag indicating whether the run length has been determined, and if the run length is not determined, address data indicating the destination address of the next table, and if the run length has been determined, the decoding table contains a flag indicating whether the run length has been determined. Data indicating the run length is recorded in each table.

In the table at address 0 and 11 of the black run, the run length has not yet been determined, and address data "2" indicating the next destination address is read. Since the second bit is also rQJ, the data at the second address corresponding to the address data "2" read out in the search for the first bit is searched. Even with the data at the second address, the run length has not yet been determined, and the next jump destination is Read address data r6) indicating the address.
Since the bit is "1", the data at address 7, which is the addition of address 1 to the 6th address corresponding to the address data "6" read in the search for the second bit, is read, and in this case, the run length has not yet been determined. Therefore, the address data “12” indicating the next destination address is
” is read out. The fourth bit is "O", and the data at address 12 corresponding to the address data "12" read out in the search for the third bit is read.

This data has a flag bit "1" set indicating that the run length has been determined, and the run length data "6" is read out and -
From the MH code rQ Q 10J, run length decoding with a run length of "6" has been achieved.

However, in the conventional bit-sequential-search MH code decoding system, the table search is performed bit by bit, so the decoding process time required by the table search is long.

A decoding method that recognizes the shortcomings of the prior art and realizes high-speed decoding has been proposed, for example, in JP-A-61-2], 2921.
It is shown in the publication No.

This decoding method attempts to obtain run length data by arranging a decoding table for MH codes consisting of a maximum of 13 bits in order of numerical values of the codes, and performing a table search for multiple bits of Ml codes at once. It is something.

[Problem to be solved by the invention]

However, although the above conventional technology has the effect of shortening the search time, it requires 2" = 8192 decoding tables to correspond to the maximum 13 bits of the MH code, which is difficult to implement in hardware. This method requires a large capacity ROM, which increases the circuit scale, and when it is realized by software, it requires a large capacity memory.

An object of the present invention is to provide a decoding method that can reduce search time and reduce the table capacity of a decoding table in decoding a variable length code such as an MH code.

[Means to solve the problem]

The above purpose is to have a decoding table group in which variable-length codes and corresponding decoding information are classified according to the number of consecutive bits of the same value from the beginning of the variable-length code, and to Select the corresponding decoding table from the group of decoding tables classified above according to the number of consecutive identical values P1 to 1 from the beginning, and search the selected decoding table using the remaining codes excluding the continuous part of the variable length code. This is achieved by obtaining decoding information.

Specifically, the above purpose is, for example, to count the number of ``0'' bits from the beginning of a variable length code such as an MH code, and to perform corresponding decoding from a group of decoding tables classified in advance according to the number of consecutive bits. Select the table and consecutive "0'"
bit and pa0'''Bit to stop the series of bits''
This is achieved by searching the decoding table for the remaining codes excluding 1'''.

In the variable length code decoding method of the present invention, as the decoding table group, the variable length code group and the decoding information corresponding to each variable length code are determined by the number of consecutive bits of the same value from the beginning of the variable length code. It has a classified decoding table for each consecutive number, and each decoding table is information representing the maximum code length of the remaining codes excluding the continuous part among all variable length codes having the same consecutive number. It is preferable to use one having the following information: , the remaining code for each variable length code, and information representing the actual remaining code length of the remaining code.

In the variable length code decoding method of the present invention, preferably, the number of consecutive identical value bits from the beginning is counted using the above decoding table, and based on the counted value, the identical value bits from the beginning of the variable length code are calculated. Select a decoding table with the same number of consecutive codes, and then, for the remaining codes, select the decoding table for each variable-length code in the decoding table within the range of the number of bits corresponding to the information representing the maximum code length of the selected decoding table. The remaining codes are searched, and decoding is performed by obtaining decoding information for variable-length codes that partially or completely match from the beginning.

Further, in the variable length code decoding method of the present invention, preferably, information representing the actual remaining code length of the remaining codes is used to detect the beginning of the next code of the continuously input variable length codes.

In the above-mentioned invention, among the variable-length codes, the bits with the same value consecutively from the beginning and the bits that stop this continuation are considered as the continuous part of the variable-length code, and the remaining code excluding this part, Preferably, the decoding table is searched.

The above purpose is to ensure that the variable length code group and the corresponding decoding information have m consecutive bits of the same value from the beginning of the variable length code.
It has a decoding table group classified according to bits or more or less than m bits, and the variable length code to be decoded is classified according to whether the number of consecutive bits of the same value from the beginning is In bits or more or less than m bits. Select a corresponding decoding table from a group of decoding tables, and if it is m bits or more, select the corresponding decoding table from the decoding table group, and if it is m bits or more, for the remaining variable length codes excluding the m bits and 8 bits.

Search the corresponding decoding table above within a preset number of bits, while if the number is less than m bits.

This can also be achieved by searching a corresponding decoding table for the variable length code using a preset range of bit numbers from the beginning and decoding the variable length code.

In this variable length code decoding method, the number of consecutive bit values from the beginning of the variable length code is m as the decoding table group.
It is distinguished by whether it is more than bits, and if it is more than m bits,
Search for a variable-length code within the maximum code length range of the remaining codes after excluding the m-bit code, and if it is less than 1m bits, search for variable-length codes within a certain number of code lengths from the beginning. It is preferable to use one having a decoding table for each variable-length code, and each decoding table having information representing the total bit length of each variable-length code.

The variable length code decoding method of the present invention can be applied to, for example, facsimile machines, still image transmission systems, and the like.

[Effect]

The operation of the present invention will be explained using a variable length code such as an MH code as an example.

Calculate the consecutive number of ii OI+ bits from the beginning of the MH code and calculate the consecutive number. Next, a decoding table corresponding to the counted consecutive number and the maximum code length of the remaining codes uniquely determined by the counted consecutive number are selected. The run length corresponding to the remaining maximum code length code is determined from the selected decoding table.

Since the actual code length of the remaining variable length codes is also prepared in the decoding table, by subtracting this code length from the maximum code length, the remaining codes required for decoding can be obtained. Therefore, even if decoding is possible without reaching the maximum code length, the beginning of the primary MH code can be determined.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings. 1st
The figure shows an example of a decoding table group for decoding an MH code using the variable length code decoding method according to an embodiment of the present invention. (b)
shows the case of black run. Further, FIG. 5 is a flowchart of the decoding process of the MH code according to this embodiment, and FIG. 6 is a diagram showing an example of the MH code and its decoding method II.

As shown in FIG. 2, the MH code is divided into white runs and black runs, and run length data is encoded. In this embodiment, decoding processing is performed by using decoding tables for white run and black run MH codes as shown in FIGS. 1(a) and 1(b).

In the tables shown in FIGS. 1(a) and (b), "
c" is the number of consecutive bits + 1011 that continues from the beginning of the MH code. Among MH codes that consist of a maximum of 13 bits, excluding the EOL (End of Line) code, bits 1101 + continue up to a maximum of 6 bits 1-. .

Further, rQJ is the maximum code length of the remaining code excluding the bit 110 II from the beginning and the bit +11 II that stops the series, and the number differs depending on the number of consecutive bits ``0''''. For example, fJj indicates that if the number of consecutive white run bits "0""c" is 1, the remaining code is the maximum number of bits, and if "c" is 5, the maximum number of bits is 2 bits. ing. rPJ is an MH code with a maximum remaining code length rQJ pi1~, and "n" and "q" are the actual remaining code length and run length corresponding to the MH code of that rPJ.

In the white run decoding table shown in FIG. 1(a),
The consecutive number C of pid i 0 +'' from the beginning is 7 from 0 to 6.
The table is determined such that when C=O, n=5 and 32 ways, and when n=1, n=7 and 1.28 ways. In the case of white run, there are 222 decoding tables in total, and the capacity of the comparison table can be reduced to 2g=512, which corresponds to the maximum code length of 9 bits for the MH code of white run.

Likewise, in the black run decoding table shown in FIG.
When c = 1, there are 2 ways, and when c = 6, there are 64 ways when n = 6, and so on.

The decoding table for the black run is 2 in total.
There are 66 ways, and the capacity can be reduced by 1,000 pulls compared to 2''=8192 ways corresponding to the maximum code length of 13 pino i~ of the black run MH code.

Next, the procedure of the decoding process will be explained using FIGS. 5 and 6. FIG. 5 is a flowchart of the MH code decoding process when the MH code decoding method according to the present embodiment is used, and the process flow will be explained using an example of a continuous MH code string as shown in FIG. . , as shown in FIG. 6, the MH code string rotoooooi.

. . ” is input, 5. First, to distinguish whether it is a white run or a black run, normally it starts from the white run and repeats black and white inversion alternately, but this process is not directly related to the present invention, so the 5th
It is omitted in the flowchart in the figure. ), when it is determined that it is a white run, the number C of consecutive bits "0" is counted (step 1). In this case, the consecutive number c=1.

Next, the remaining maximum code length Q corresponding to n=1 is read from the table shown in FIG. 1(a) to obtain Q;7 (step 2).

Next, the consecutive bits “0” from the beginning and bit 11
Excluding the bit ``1'' that stopped the series of 0''
Remaining Q=7 bits of MH code string P=r00000
Obtain 10J (step 3).

Then, from the table shown in FIG. 1(a), bit II
Remaining code P=rooo when the number of consecutive O11s is c=1
Corresponds to "ooto". Obtain the remaining code length n = 3 and run length q = 11 as the actual MH code (step 4).

Here, n is data indicating the remaining code length of 7, so
This is also data that informs the starting position of the next MH code. Therefore, in the case of Q#n, the pointer to the beginning of the next MH code is specified. In this case, Q='1. Since n = 3, of the 7-bit code used for table reference, Q-n = 4 bits will be the next MH code, so move the start pointer forward by these 4 bits. Return (steps 5 and 6).

q is run length data for generating data before being compressed by the MH code, which enables data expansion.

This is a series of processes for converting the MH code into a run length.

Next, the decoding of the MH code of the next black run following the above-mentioned white run MH code determined by specifying the head pointer of the MH code will be explained.

As shown in FIG. 6, the MH code string rootoooooo
When "il..." is input, it is the same as in the case of white run.

Count the number C of consecutive bits "'0" and obtain c=2 (step 1).

Next, the remaining maximum code rCQ corresponding to c=2 is read from the table shown in FIG. 1(b) to obtain A=1 (step 2).

Next, consecutive bits II O11 and bit r from the beginning
An MH code P= rQJ for the remaining Q=] bits excluding the bit u 1 ++ for which the continuation of r Ou has been stopped is obtained (step 3).

Then, from the table shown in FIG. 1(b), bit L(
Actual remaining code length n = 1 and 5-run length q corresponding to remaining code P = “0” when the number of consecutive O11s is c = 2
=6 is obtained (step). In this case, Q=n,
There is no need to shift the pointer to the starting position of the next MH code (steps 5 and 6).

In this way, q=l of white run 1. The black run q=6 can be decoded, and the subsequent <MH code can be decoded in the same manner by repeating the white run and the black run alternately to the run length. Note that the E○L code has bit II
The consecutive number of OII is 11, but in this case bit 1/
This can be handled as a1)-gal processing when the number of consecutive 011s is 7 or more.

In this way, according to the MH decoding using the MH decoding table according to the present embodiment, the power of 5 bits II O++ is
- It is possible to decode to run length data at high speed only by processing and batch table search processing of remaining codes, and furthermore,
When the table capacity is white run, it is 222 ways, and when it is black run, it is 266 ways, which can be achieved with a small amount of indigo.

In the decoding tables shown in FIGS. 11(a) and 11(b), the correspondence between the number of consecutive bits ``0'' C and the remaining maximum code length Q is set as the first table (14 bytes in total for black and white runs), Remaining code P, remaining code length n, and run length q
Correspondence to the second daple (black and white run total 488 words)
You can also use it as

Furthermore, although the present embodiment describes the case of M and H codes, the present invention can also be applied to decoding other variable word length codes.

FIG. 7 is a decoding table in the second embodiment of the present invention. The second embodiment is an application of the present invention to a still image transmission device for natural images (multivalued images).

The encoding method of the second embodiment will be explained.

First, orthogonal transformation is performed on the luminance level of 8 pixels vertically by 8 pixels horizontally. Here, orthogonal transformation is a method of encoding image signals, dividing the image into blocks of appropriate pixels, decomposing the image into frequency components for each block, and assigning the number of bits depending on the size of the components. This is a method to encode the data. For example, trigonometric functions (sin,
A transformation is performed using an orthogonal function called (cos).

Next, the data after orthogonal transformation is quantized to ±127 levels and entropy encoded according to the encoding table shown in FIG. The data after orthogonal transformation encodes the consecutive number of 0s, that is, the run length, since the frequency of 0s is 8N.
By repeating block-by-block encoding using 8 x 8 pixels as one program.

Encodes one screen. In the 8th IA, E○B is E
Indicates that all data after OB is O.

B in the code is an extension bit and represents the offset value,
S represents positive and negative.

For example, when encoding level -6, the offset value is 2 in the code 0OOOOOOIBBS representing 4 to 7.
(6-4) & Expressed in binary, BB = (10) 7. Since it is a negative value, S = t, and the encoded code is 0000 in the end.
It becomes 001101.

Next, a method of decoding using the decoding table of FIG. 7 will be explained using the flowchart of FIG. 9.

After loading the encoded data (SIO), first, a decoding table is determined based on whether or not the first five bits of the encoded data are all O's (Sll).

Next, for encoded data where the first 5 bits are all 0, refer to the decoding table using the 6th and subsequent 7 bits (513), and for other encoded data, use the first 7 bits. Please refer to the decoding table using
2) Obtain the zero run/level identification R1 value D, total code length, and extended bit length M.

Next, when the extended bit length M is not O (S14),
Add the value indicated by the M bits after the (total code length plus one extended bit length M) bit of the encoded code to the value (815)
. Next, if the total code length L-th bit S of the encoded code is 1 (S16), the value is set to a negative value (S17), and the value and zero run/level identification R are obtained (S18).

As an example, a procedure for decoding encoded data 0000001101 will be described.

(2) Since the first 5 bits are all O, the decoding table is referred to using 7 bits 01101Xx from the 6th bit (xX is the subsequent encoded data). Number of consecutive O's at the beginning 5
From the table contents of bits or more 0IXXXXX, zero run/level identification R = level, value D = 4, total code length L =
10. Obtain the extended bit length M=2.

(2) Since the extension bit is not 0, extract M=2 bits from L−M=8th bit to obtain extension bit (10)2.

Add the value of the extension bit part to the value and get D=4+ (10
) 2=4+2=6.

(2) Total code length L=Since the 10th bit (indicating positive/negative) is 1, the value is set to a negative value, and the value D=-6 and zero run/level identification R=level are obtained.

As described above, in the second embodiment, the decoding tables are classified into two types, so the selection of the decoding table involves checking at which bit from the beginning there is a code that is not O. The decoding process is characterized in that the decoding process can be performed easily and at high speed, since it is only necessary to check whether O continues for a certain value or more even if the decoding process does not occur. Furthermore, since there is no need to search the table for extension bits representing the offset amount of the decoded value, the size of the table can be small.

〔Effect of the invention〕

As explained above, according to the variable length code decoding method of the present invention, the search time for searching decoded data from the variable length code is not repeated one bit at a time as in the conventional method, but is Since you can search, you can save time. In addition, the table capacity for searching is also the capacity corresponding to the maximum code length of a variable length code (for example, 2
13=8192 ways).

[Brief explanation of drawings]

Figures 1 (1) and (b) are explanatory diagrams showing an example of a decoding table group for decoding an MH code using the 11 variable length code decoding method according to an embodiment of the present invention, and Figure 2 is a facsimile machine. 3 and 4 are explanatory diagrams showing a tree-shaped MH decoding table, and FIG. 5 is an explanatory diagram showing an encoding table of the M l-1 code used in A flowchart showing a decoding process, FIG. 6 is an explanatory diagram showing an example of an MH code string and the flow of its decoding procedure, FIG. 7 is an explanatory diagram showing a decoding table in the second embodiment of the present invention,
FIG. 8 is an explanatory diagram showing an encoding table according to the second embodiment of the present invention, and FIG. 9 is a flowchart of decoding processing according to the second embodiment of the present invention. collection! Figure (0-) Figure compilation 1 (b) Figure Figure (0-) Self (b) Black compilation + Figure (α) Self <b) $, Franc compilation Figure 8 illustrated book diagram

Claims (1)

[Claims] 1. The variable length code group and the corresponding decoding information have a decoding table group classified by the number of consecutive bits of the same value from the beginning of the variable length code, and the variable length code to be decoded. Regarding the code, depending on the number of consecutive bits with the same value from the beginning,
The method is characterized in that a corresponding decoding table is selected from the group of classified decoding tables, and the selected decoding table is searched using the remaining code excluding the continuous part of the variable length code to obtain decoding information. Variable length code decoding method. 2. A variable length code decoding method that decodes a variable length code by referring to a decoding table group having a variable length code group and corresponding decoding information, wherein the decoding table group is a variable length code The group and the decoding information corresponding to each variable-length code are classified by the number of consecutive bits of the same value from the beginning of the variable-length code, and each decoding table has a decoding table for each consecutive number, and each decoding table includes: Information indicating the maximum code length of the remaining codes excluding the continuous part among all variable length codes having the same number of consecutive parts, the remaining code for each variable length code, and the actual remaining code length of the remaining code. 1. A variable length code decoding method characterized by having information representing. 3. In the variable length code decoding method according to claim 2,
The number of consecutive identical value bits from the beginning is counted, and based on the counted value, a decoding table with the same consecutive number of identical value bits from the beginning of the variable-length code is selected, and then, for the remaining codes, the selected decoding table is Search the remaining codes for each variable-length code in the decoding table within the range of the number of bits corresponding to the information representing the maximum code length of the decoding table, and search for variable-length codes that partially or completely match from the beginning. A variable length code decoding method characterized by obtaining decoding information. 4. In the variable length code decoding method according to claim 2 or 3, the beginning of the next code of the continuously input variable length codes is detected using information representing the actual remaining code length of the remaining codes. It is characterized by Variable length code decoding method. 5. A facsimile apparatus that decodes a received signal using the variable length code decoding method according to claim 1, 2, 3 or 4. 6. The variable length code group and the corresponding decoding information have a decoding table group classified according to whether the number of consecutive bits of the same value from the beginning of the variable length code is m bits or more or less than m bits, and the decoding For the variable-length code to be converted, a corresponding decoding table is selected from the group of decoding tables classified above depending on whether the number of successive bits of the same value from the beginning is m bits or more or less than m bits, and if it is m bits or more, The m
For the remaining variable length code excluding bits, search the corresponding decoding table above within the preset number of bits, and if the number is less than m bits, the bits in the preset range from the beginning of the variable length code are searched. A variable length code decoding method characterized in that a variable length code is decoded by searching a corresponding decoding table numerically. 7. A variable length code decoding method that decodes a variable length code by referring to a decoding table group having a variable length code group and corresponding decoding information, wherein the decoding table group is a variable length code It is distinguished by whether the number of consecutive bit values from the beginning of If the search code is less than m bits, there is a decoding table that uses a variable length code as the search code within a certain number of code lengths from the beginning, and each decoding table is A variable length code decoding method characterized by having information representing bit length. 8. A still image transmission system that decodes transmitted still image data using the variable length code decoding method according to claim 6 or 7. 9. Of the variable-length code, the bits with the same value consecutively from the beginning and the bit that stops this continuation are considered as the continuous part of the variable-length code, and the remaining code excluding this part,
Claim 1, characterized in that the decoding table is searched.
6. The variable length code decoding method according to claim 2, 3 or 4, or the facsimile apparatus according to claim 5.