JP2002246914A - Code conversion device, code conversion method, code conversion program, and storage medium recoding the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a code conversion device and a method which receive data encoded, through a method of using EOB is used and is capable of outputting the same code with a 3DVLC method, and avoiding redundant procedures. SOLUTION: A code detector 111 detects codes or EOB codes from an input bit string, referring to the bit arrangement of a detection code table 110. An EOB verifying part 112 stores the codes in a code storage 113, when it detects codes other than EOB codes. A code output changing means 114 outputs the codes, stored in the code storage 113 into a halfway code converter 118, when the code detector 111 detects the EOB codes and into a code converter 115, when the code detector 111 detects the EOB codes. The halfway code converter 118 retrieves and outputs the codes of a midway code primary table 17 to input codes, and the code converter 115 retrieves and outputs the codes of the a primary table 116. In this way, a bit arrangement is replaced, and the same codes with a 3DVLC method are outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a variable-length code conversion method and apparatus in moving image coding transcoding.

[0002]

2. Description of the Related Art Generally, in moving picture coding such as MPEG-1, a DCT (Discrete Cosine Transform) is applied to image information and its coefficient is quantized. The DC component of the quantized DCT coefficient is subjected to variable-length coding as it is, but the AC component uses run-length coding because not all of the DCT coefficient arrays have valid values. That is, when the DCT coefficient that is an invalid quantization level is represented by 0, the value of the valid DCT coefficient is represented by a level
The run number is defined as the run number of the coefficient 0 immediately before that, and the variable length code table corresponding to the two-dimensional amount of each run and level prepared in advance is used as a look-up table to perform variable length coding. In this way, the image information is represented by a continuous variable length code.

[0003] There is not always a valid coefficient at the end of the DCT coefficient. In this case, the value of the level cannot be obtained. Therefore, it is necessary to indicate that all coefficients after a certain valid DCT coefficient are not valid. This method is roughly classified into two types. Encoding schemes MPEG-1 and M
PEG-2 uses a method of arranging a code of a unique bit pattern called EOB (end of block) after the last variable length code. On the other hand, the encoding method 26
In 3 or MPEG-4, a method of preparing a variable length code table for the last variable length code separately in advance and using the codeword of the table to indicate the last variable length code (3DVLC method) use. In the latter case,
A variable-length code table expressed in three dimensions of a run, a level, and an information last indicating that variable-length coding is the last is used.

For example, image information is converted into one type of luminance component (Y
Component) and two types of color components (U component and V component),
When an image in the 420 format in which the U and V components each have an information amount of の of the Y component is encoded by the MPEG-1 system, a macroblock which is a unit for encoding the image is encoded. In this case, a two-dimensional DC is performed by six blocks (eight Y components, one U component, and one V component) of eight pixels in the vertical and horizontal directions.
T, quantization and variable length coding are performed. The order of encoding each block is predetermined. First, the upper left Y component block is encoded. In the case of inter-frame prediction coding,
For the block, after two-dimensional DCT and quantization,
Zigzag scan DCT coefficients as shown in FIG.
Find the run and level of the AC component. A variable-length code corresponding to the run and level is searched for in a variable-length code table for a two-dimensional amount of run and level as shown in FIG. 11, and is output as encoded data. If 0 continues until the last DCT coefficient, the block has a valid DCT
Since there is no coefficient, the EOB code is output. Subsequently, similar processing is performed on the remaining blocks such as the Y component block at the upper right of the screen. In this way, the image information in the macro block is encoded.

[0005] Fig. 12 shows an example of the configuration of the variable length coding portion. Zero count section 101 counts the number of DCT coefficients having a value of 0 until a coefficient having a value other than 0 is found.
When a coefficient having a value other than 0 is found in the DCT coefficient array, the run level setting unit 103 determines the number of coefficients having a value of 0,
The subsequent coefficient values other than 0 are output to the 2D code output unit 104 as run and level values. The 2D code output unit 104 outputs a code with reference to the two-dimensional lookup table corresponding to the run and the level recorded in the two-dimensional table unit 105. The EOB code output unit 10
6 outputs the EOB code by the change by the EOB coding changing means 102 when the value of the DCT coefficient is 0 to the end in the zero count unit 101.

Under such a configuration, the operation when the value of the DCT coefficient becomes as shown in FIG. 13 is as follows. There are twelve DCT coefficients, and the array is DCT [i] (i
= 0,... 11). First, the zero count unit 101 measures the number of coefficients up to a coefficient other than 0 (referred to as a 0 run length). DCT [0] = 0, DCT [1]
= 11, the 0 run length is 1. The run level setting unit 103 outputs run 1 and level 11. The 2D code output unit 104 outputs the codes of run 1 and level 11 with reference to the table of the two-dimensional table unit 105. Similarly, the 2D code output unit 104 outputs DCT [6] (= 53)
The code corresponding to the run and level is output. DCT [7]
Thereafter, since the value is all 0, the EOB code output unit 106
Outputs an EOB code.

On the other hand, when an image of the same 420 format is encoded by the MPEG-4 system, two-dimensional DCT, quantization and variable length encoding are similarly performed in six blocks, but variable length encoding is performed. The method is different from the MPEG-1 method and uses the 3DVLC method. First, the DCT coefficient is zigzag-scanned as shown in FIG. 10 to determine the run and level of the AC component. Next, it is confirmed whether there is a valid DCT coefficient after the AC component and the last information is obtained. Thereafter, a variable-length code corresponding to the run, level, and last is searched for in a variable-length code table for a three-dimensional amount of run, level, and last as shown in FIG. 14, and is output as encoded data. When the last valid DCT coefficient run, level, and last in the block are obtained, the encoding of the block is completed. Subsequently, similar processing is performed on the remaining blocks.

FIG. 15 shows a configuration example of the 3DVLC method. Zero count section 101 counts the number of DCT coefficients having a value of 0 until a coefficient having a value other than 0 is found. D
If a coefficient having a value other than 0 is found in the CT coefficient array, the run level setting unit 103 determines the number of coefficients having a value of 0 (0 run length) and the subsequent non-zero coefficient value as a run. The value is output to the 3D code output unit 109 as a level value. The next code coefficient detection unit 107 checks whether or not a coefficient having a value other than 0 exists after the coefficient having a value other than 0 found by the zero counting unit 101. If a coefficient with a value other than 0 exists, 0 is output to the 3D code output unit 109 as a last value if it does not exist. 3D code output unit 1
09 outputs a code by referring to a three-dimensional lookup table corresponding to the run, level, and last recorded in the three-dimensional table unit 108.

Under such a configuration, the operation when the DCT coefficient value becomes as shown in FIG. 13 is as follows. There are twelve DCT coefficients, and the array is DCT [i] (i
= 0,..., 11). First, the zero count unit 101 measures the number of coefficients up to a coefficient other than 0 (zero run length). DCT [0] = 0, DCT [1] = 1
Since it is 1, the 0 run length is 1. Runlevel setting unit 1
03 outputs run 1 and level 11. Next code coefficient detection section 107 searches for a coefficient other than 0 in the coefficients after DCT [1]. Since the value of DCT [3] is 22, which is other than 0, 0 is output as the last code. The 3D code output unit 109 converts the code of run 1, level 11, and last 0 to 3
The data is output with reference to the table of the dimension table unit 108.
Similarly, the 3D code output unit 109 outputs DCT [3] (= 2
In response to 2), codes corresponding to the run, level, and last are output. For DCT [6] (= 53), run level setting section 103 outputs run 2 and level 53. Next code coefficient detection section 107 searches for a coefficient other than 0 in coefficients after DCT [6]. Since all coefficients are 0, 1 is output as the last code. 3D code output unit 109
Outputs the codes of run 2, level 53, and last 1 with reference to the table of the three-dimensional table unit 108.

Generally, since the EOB code has a long code length, the method using EOB is effective when the number of effective DCT coefficients in a block increases and the encoding speed is high, and the 3DVLC method is effective. DCT coefficient decreases,
This is effective when the encoding speed is low.

[0011]

The transcoding method for converting from a method using EOB such as MPEG-1 encoded data to a method using 3DVLC method such as MPEG-4 encoded data corresponds to a method using 3DVLC method. Last information is needed to output the code to be executed. However, since the last information is not known until an EOB code in the method using EOB is detected, the last information cannot be known from the variable length code encoded by the method using EOB. Therefore, from the code of the method using EOB, 3
Transcoding cannot be realized by a method in which a look-up table is created in which a one-to-one correspondence with codes according to the method using DVLC is made, and codes are converted with reference to the table. Therefore, it is once decoded to DCT coefficients,
It was necessary to encode the T coefficient. That is, a procedure for temporarily converting the arrangement of variable length codes into run and level information as two-dimensional information, a procedure for converting such information into DCT coefficients, and a run, level, and last as three-dimensional information from DCT coefficients again. A procedure for converting the information into information and a procedure for converting the information into a variable length code are required.

Therefore, the present invention can output a code similar to that using the 3DVLC method, and converts the above-mentioned sequence of variable-length codes into run and level information which is two-dimensional information once. There is no need for a procedure for converting these information into DCT coefficients, a procedure for converting the DCT coefficients into run, level, and last information, which is three-dimensional information, and a procedure for converting these information into variable-length codes. It is an object of the present invention to provide a code conversion device and a code conversion method.

[0013]

In order to solve the above-mentioned problems, a code conversion apparatus according to a first aspect of the present invention comprises a plurality of patterns other than an EOB code which is a unique bit array and a bit array of an EOB code. A code conversion device for inputting a bit sequence composed of a code capable of taking a bit sequence of a code sequence and outputting a bit sequence different from the input bit sequence, wherein the bit sequence of the code and the bit sequence of an EOB code are recorded. A detection code table section, a code detection section for detecting the code or the EOB code by referring to a bit array recorded in the detection code table section from an input bit string, and a code detection section for detecting the code or the EOB code. Code accumulating section for accumulating a code input thereto, and EOB confirming means for inputting a code to the code accumulating section when the code detecting section detects a code other than an EOB code. , The bit sequence of each code,
A one-dimensional table unit that stores a table in which codes of different bit arrays are associated with each other; and a code conversion unit that searches and outputs a code stored in the one-dimensional table unit for an input code. When the code detection unit detects a code other than the EOB code, the code stored in the code storage unit is output. When the code detection unit detects the EOB code, the code stored in the code storage unit is output. Code output changing means for inputting to the code conversion unit.

Alternatively, the code conversion apparatus according to the second invention of the present invention comprises an EOB code which is a unique bit array,
A code conversion device that inputs a bit string composed of a code that can take a bit array of a plurality of patterns other than the bit array of the OB code, and outputs a bit string different from the input bit string. A detection code table section in which an array and a bit array of the EOB code are recorded;
From the input bit string, the code or EO is referred by referring to the bit arrangement recorded in the detection code table section.
A code detection unit that detects a B code, a code storage unit that stores a code detected and input by the code detection unit, and stores a code in the code storage unit when the code detection unit detects a code other than the EOB code. EOB confirmation means for inputting a code, a one-dimensional table section storing a table corresponding to a code of another bit array for each code bit array, and a one-dimensional table for storing the input code. A code conversion unit for searching and outputting the codes stored in the unit, and a table in which the bit array of each code is associated with a code of a bit array different from the bit array recorded in the one-dimensional table unit. An intermediate code one-dimensional table unit storing the intermediate code, an intermediate code conversion unit for searching and outputting a code stored in the intermediate code one-dimensional table unit for an input code, and a code detection unit. EOB When a code is detected, the code stored in the code storage unit is input to the code conversion unit, and when the code detection unit detects a code other than the EOB code, the code stored in the code storage unit is converted to the intermediate code. Code output changing means for inputting to the conversion unit.

Further, the code conversion method according to the first invention of the present invention provides an EOB code which is a unique bit array,
A code conversion method for inputting a bit string composed of a code that can take a bit array of a plurality of patterns, other than a code bit array, and outputting a bit string different from the input bit string, wherein the code bit array And a detection code table in which a bit array of EOB codes are recorded, and a one-dimensional table storing a table corresponding to a code of another bit array with respect to the bit array of each code. A code detecting step of detecting the code or the EOB code from the bit string by referring to a bit array recorded in the detection code table; and a code storing step when a code other than the EOB code is detected in the code detecting step. And an EOB confirmation step of inputting the EOB and storing the code inputted from the EOB confirmation step Outputting a code stored in the code storing step when an code other than an EOB code is detected in the code storing step, and outputting the code stored in the code storing step when the code detecting step detects an EOB code. A code output changing step of inputting the code stored in step 1 into a code converting step; and a code converting step of searching and outputting a code stored in the one-dimensional table for the code input from the code output changing step. And characterized in that:

Alternatively, the code conversion method according to the second invention of the present invention comprises the steps of:
A code conversion method for inputting a bit sequence composed of a code that can take a bit array of a plurality of patterns, other than the bit sequence of the OB code, and outputting a bit sequence different from the input bit sequence, comprising: A detection code table in which an array and a bit array of an EOB code are recorded; a one-dimensional table in which a table corresponding to a code of another bit array is stored for each bit array of each code; An intermediate code one-dimensional table that stores a table corresponding to a code of a bit array different from the bit array recorded in the one-dimensional table for the array; A code detecting step of detecting the code or the EOB code by referring to the bit array recorded in the table; An EOB checking step of inputting a code to a code storing step when a code other than a code is detected; a code storing step of detecting the code in the code detecting step and storing the code input from the EOB checking step; If a code other than the EOB code is detected in the step, the code stored in the code storage step is input to the intermediate code conversion step, and the EO code is detected in the code detection step.
When a B code is detected, a code output change step of inputting the code stored in the code storage step to a code conversion step, and a code input from the code output change step are stored in the intermediate code one-dimensional table. The intermediate code conversion step of searching for and outputting the stored code,
A code conversion step of searching for and outputting a code stored in the one-dimensional table with respect to the code input from the code output changing step.

Further, the recording medium on which the code conversion program according to the first invention of the present invention is recorded is an EOB code which is a unique bit array and a code which can take a bit array of a plurality of patterns other than the bit array of the EOB code. And a recording medium storing a code conversion program for outputting a bit sequence different from the input bit sequence, wherein the bit sequence of the code and the bit sequence of the EOB code are recorded. A detection code table, and a one-dimensional table storing a table corresponding to a code of another bit array for a bit array of each code. A code detecting procedure for detecting the code or the EOB code by referring to a certain bit array;
An EOB confirmation procedure for inputting a code to the code accumulation procedure when a code other than a code is detected; a code accumulation procedure for detecting the code in the code detection procedure and accumulating the code input from the EOB confirmation procedure; If a code other than the EOB code is detected in the procedure, the code stored in the code storage procedure is output, and if the EOB code is detected in the code detection procedure,
A code output changing step of inputting the code stored in the code storing step to a code converting step; and searching and outputting a code stored in the one-dimensional table for the code input from the code output changing step. The code conversion procedure is recorded on a computer-readable recording medium in order to execute the code conversion program provided by the computer.

Alternatively, the recording medium on which the code conversion program according to the second invention of the present invention is recorded is a code that can take a bit array of a plurality of patterns other than the EOB code that is a unique bit array and the bit array of the EOB code. And a recording medium storing a code conversion program for outputting a bit sequence different from the input bit sequence, wherein the bit sequence of the code and the bit sequence of the EOB code are recorded. A detection code table, a one-dimensional table storing a table corresponding to a code of another bit array with respect to a bit array of each code, and a one-dimensional table storing a bit array of each code in the one-dimensional table. An intermediate code one-dimensional table storing a table corresponding to a code of a bit arrangement different from the existing bit arrangement. A code detection procedure for detecting the code or the EOB code by referring to the bit array recorded in the detection code table from the sequence of codes, and a code storage procedure when a code other than the EOB code is detected in the code detection procedure. An EOB confirmation procedure for inputting a code, the code accumulation procedure for detecting the code detected in the code detection procedure, and accumulating the code input from the EOB confirmation procedure, A code output changing step of inputting the code stored in the code storing procedure to an intermediate code converting procedure and inputting the code stored in the code storing procedure to the code converting procedure when an EOB code is detected in the code detecting step; For the code input from the code output change procedure, the intermediate code conversion procedure of searching and outputting the code stored in the intermediate code one-dimensional table, A code conversion step of searching and outputting a code stored in the one-dimensional table with respect to a code input from the code output change procedure, and executing a code conversion program provided by a computer. It is recorded on a computer-readable recording medium.

Further, the code conversion program according to the first invention of the present invention comprises: an EOB code which is a unique bit array;
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
A code conversion program for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; And a one-dimensional table storing a table corresponding to the code of the bit array of the above. The code or the EOB code is detected from the input bit string by referring to the bit array recorded in the detection code table. A code detection procedure, an EOB confirmation procedure for inputting a code to a code storage procedure when a code other than an EOB code is detected in the code detection procedure, and a code input from the EOB confirmation procedure detected in the code detection procedure. EO is stored in the code storing procedure and the code detecting procedure.
A code for outputting a code stored in the code storage procedure when detecting a code other than the B code, and for inputting the code stored in the code storage procedure to a code conversion procedure when detecting an EOB code in the code detection procedure. It is characterized by comprising an output change procedure, and the code conversion procedure of searching and outputting a code stored in the one-dimensional table with respect to a code input from the code output change procedure.

Alternatively, the code conversion program according to the second aspect of the present invention comprises an EOB code which is a unique bit array and a code which can take a bit array of a plurality of patterns other than the bit array of the EOB code. A code conversion program for inputting a bit sequence and outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit array of the code and a bit array of an EOB code are recorded; In contrast, a one-dimensional table storing a table corresponding to codes of different bit arrays, and a bit array of a bit array different from the bit array recorded in the one-dimensional table for the bit array of each code. An intermediate code one-dimensional table that stores a table corresponding to codes, and records the detected code table from the input bit string. A code detection procedure for detecting the code or the EOB code by referring to the bit array, and an EOB confirmation procedure for inputting a code to a code storage procedure when a code other than an EOB code is detected in the code detection procedure. A code storing step of detecting codes in the code detecting procedure and storing codes input from the EOB checking procedure; and, when a code other than an EOB code is detected in the code detecting procedure, the codes stored in the code storing procedure are intermediary codes. A code output changing step of inputting the code stored in the code storing step to a code converting step when an EOB code is detected in the code detecting step, and a code input from the code output changing step. The intermediate code conversion procedure for searching for and outputting the code stored in the intermediate code one-dimensional table, and the code input from the code output change procedure. With respect to, the one
And a code conversion step of searching and outputting a code stored in a dimension table.

According to the first aspect of the present invention, a bit string composed of an EOB code which is a unique bit array and a code which can take a bit array of a plurality of patterns other than the bit array of the EOB code is input. In the code conversion for outputting a bit sequence different from the bit sequence that has been performed, the bit sequence is exchanged, so that the bit sequence of the EOB code and the code immediately before it can be converted to another bit sequence. Thereby, from the encoded data encoded by the method using EOB,
In image coding for outputting coded data that can be decoded by a decoder for the 3DVLC method, a procedure for temporarily converting a sequence of variable length codes into run and level information as two-dimensional information, and converting the information into DCT coefficients It is possible to omit a procedure, a procedure for converting the DCT coefficients into run, level, and last information, which are three-dimensional information, and a procedure for converting these information into a variable-length code.

According to a second aspect of the present invention, in addition to the first aspect, the bit arrangement of the EOB code and the code immediately preceding the EOB code are converted to another bit arrangement by exchanging the bit arrangement.
The bit array of a code that is not immediately before the EOB code is also converted to another bit array. Accordingly, in the first invention, only the bit array of the code of the last valid DCT coefficient in the block is converted, whereas in the second invention, other valid DCT coefficients in the block are converted. The bit arrangement of the sign of the coefficient can be converted.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to the drawings.

[0024] Bits different from the first encoding scheme, which can be decoded by the second encoding scheme, using the encoded data encoded by the first encoding scheme using the motion compensation prediction encoding scheme as an input. An operation example when the code conversion method of the present invention is applied when converting the bit array of a macroblock in an image coded data conversion device that outputs coded data of an array will be described. DCT in input encoded data
The coefficients are coded using EOB, which is
The output is converted to a code that can be decoded by the DVLC method.

FIG. 1 shows an outline of an image coded data converter. The image coded data conversion device includes a data input unit 201 that inputs coded data of a macroblock,
A coding mode decoding unit 203 for decoding the coding mode of the macroblock, a coding mode coding unit 204 for coding the coding mode again, and a bit array of the motion vector information of the first coding method in the second coding mode. And a motion vector table unit 206 that stores a lookup table corresponding to the bit array of the encoding method in a one-to-one correspondence with a bit array of motion vector information in an input bit string.
And a motion vector conversion unit 205 that converts the bit array with reference to the table stored in the table 06, and the bit array of the quantization scale of the first encoding system corresponds to the bit array of the second encoding system on a one-to-one basis. The quantization scale table unit 208 storing the lookup table thus obtained, and the quantization scale conversion for converting the bit array of the bit array of the quantization scale information in the input bit string with reference to the table stored in the quantization scale table unit 208 Unit 207, DCT bit pattern conversion unit 209 for converting DCT coefficient encoded data, and DCT bit pattern conversion counting unit 210 for counting the number of times DCT coefficient encoded data is converted
And the bit string input by the data input unit 201 into the encoding mode decoding unit 203 or the motion vector conversion unit 20.
5, or the quantization scale conversion unit 207, or DCT
A data input changing unit 202 that switches input to any of the bit pattern conversion units 209 and a data output unit 211 that outputs encoded data are provided.

The DCT bit pattern conversion unit 20
9 is shown in FIG. DCT bit pattern converter 2
09 denotes a detection code table unit 110 and a code detection unit 11
1, EOB confirmation means 112, code accumulation section 113,
A code output changing unit 114, a code conversion unit 115, a one-dimensional table unit 116, an intermediate code conversion unit 118, and an intermediate code one-dimensional table unit 117 are provided.

The encoding mode includes the first encoding method and the second encoding method.
Information on whether motion vector information is included in the encoded data of the macroblock,
It is assumed that the information indicates the position of the block containing the DCT coefficient encoded data. In the first encoding method, it is assumed that the encoded data in the encoding mode has a fixed length of 7 bits.
When the first 1 bit is 0, it indicates that there is no motion vector information, and when it is 1, it indicates that there is motion vector information. The following 6 bits indicate the presence or absence of DCT coefficient information of each block in the order of Y component upper left, Y component upper right, Y component lower left, Y component lower right, U component, and V component.
When each bit is 0, it indicates that there is no DCT coefficient information of the corresponding block, and when each bit is 1, it indicates that there is DCT coefficient information. In the second encoding method, it is assumed that encoded data in the encoding mode has a fixed length of 7 bits. The first 6 bits indicate the presence or absence of DCT coefficient information of each block in the order of Y component upper left, Y component upper right, Y component lower left, Y component lower right, U component, and V component. When each bit is 0, it indicates that there is no DCT coefficient multiple information of the corresponding block, and when each bit is 1, it indicates that there is DCT coefficient information. When the subsequent 1 bit is 0, it indicates that there is no motion vector information, and when it is 1, it indicates that there is motion vector information.

In the motion vector information, in the first encoding method, it is assumed that the value of the vertical component and the value of the component are each encoded at a fixed length of 7 bits. When the first bit is 0, the value is positive, and when it is 1, it is negative. In the second encoding method, it is assumed that the value of the vertical component and the value of the component are each encoded with a fixed length of 7 bits. When the first bit is 1, the value is positive, and when it is 0, the value is negative. The quantization scale information is encoded with a fixed length of 7 bits represented by a binary number in the first encoding scheme, and encoded with a fixed length of 8 bits represented by a binary number in the second encoding scheme. It is assumed that

FIG. 3 shows an example of the bit arrangement correspondence table stored in the motion vector table section 206. FIG. 4 shows an example of a bit array correspondence table stored in the quantization scale table unit 208. FIG. 5 shows an example of a bit array correspondence table stored in the detection code table section 110.
FIG. 6 shows an example of the bit array correspondence table stored in the one-dimensional table unit 116.
FIG. 7 shows an example of the bit array correspondence table stored in the table 17. The bit arrangement of the EOB code is 0011111000
And It is assumed that bit 0 is stored in the code storage unit 113 in advance. Each time a new bit array is stored, the stored code is output. The code output changing means 114 does not output anything when the output from the code storage unit 113 is bit 0. If the EOB code output unit 106 does not output the EOB code, the code output changing unit 114 outputs the bit array stored in the code storage unit 113 to the intermediate code conversion unit 118. When the EOB code output unit 106 outputs the EOB code, the bit array stored in the code storage unit 113 is output to the code conversion unit 115, and bit 0 is stored in the code storage unit 113. As an initial state, the data input change unit 202 converts the data input by the data input unit 201 into the encoding mode decoding unit 2.
03 is set to be input.

Under the above assumption, when macroblock encoded data having the following bit arrangement is input,
The image encoding device operates as follows.

1010100 0001111 1000001 0001111 0011100 0011110 0011111000 0011110 001011000 0011111000 First, the coding mode decoding section 203 comprises a data input section 201.
7 bits are extracted from the coded data input from. It can be seen that the encoding mode information is 1010100.
By decoding the encoding mode, the encoded data of the macroblock includes motion vector information, and includes DCT coefficient information of the upper right component and the lower right component of the Y component, that is, DCT information for two blocks. You can see that.
The encoding mode encoding unit 204 converts the decoded encoding mode into a bit array 0101001. Since the motion vector information is in the encoded data, the motion vector conversion unit 2
Reference numeral 05 refers to a table in which the input data is stored in the motion vector table unit 206, and converts the bit array. Bit array 1001 for bit array 0001111
111, and a bit array 0000001 for the bit array 1000001. The quantization scale conversion unit 207 refers to the table stored in the quantization scale table unit 208, and refers to the bit array 0001111
Is converted into a bit array 000011111 and output.

Thereafter, the DCT bit pattern converter 20
9 is 3DVL from the coded data obtained by coding two blocks of coded data by the method using EOB.
It is converted to the one coded by the C method. The method is as follows.

The code detecting section 111 detects a code by referring to the table stored in the detected code table section 110. First, the bit array 0011100 is detected. Since this is not an EOB, the code storage unit 113 outputs bit 0 stored in advance, and stores this bit array.
The sign output changing means 114 does not output a bit because the bit 0 is input.

Next, the code detecting unit 111 outputs the bit sequence 001
1110 is detected. Since this is not an EOB, the code storage unit 113 stores the stored bit array 001110.
0 is output and the bit array 0011110 is stored. Since the code output changing unit 114 does not receive the EOB code, the code output changing unit 114 converts the input from the code storage unit 113 into the intermediate code conversion unit 11.
8 is output. The intermediate code conversion unit 118 refers to the table stored in the intermediate code one-dimensional table unit 117,
The bit array 0011100 is replaced with the bit array 1111100
And output.

Next, the code detecting section 111 outputs the bit sequence 001
1111000 is detected. Since this is an EOB code,
The code storage unit 113 stores the stored bit array 0011
Output 110 and store bit 0. Since the code output changing means 114 receives the EOB code, the code output
13 is output to code conversion section 115. The code conversion unit 115 converts the bit array 0011110 into a bit array 1111111, with reference to the table stored in the one-dimensional table unit 116, and outputs it.

DCT bit pattern conversion counting section 21
0 means that the conversion of one block has been completed, so the count is 1
To

Subsequently, the code detection unit 111 sets the bit arrangement 0
011110 is detected. Since this is not EOB,
The code storage unit 113 outputs the bit 0 stored in advance, and stores this bit array. Sign output changing means 11
No. 4 does not output a bit because the bit 0 is input.

Next, the code detecting section 111 outputs the bit sequence 001
011000 is detected. Since this is not EOB,
The code storage unit 113 stores the stored bit array 0011
110 and accumulate the bit array 001011000. Since the code output changing unit 114 does not receive the EOB code, the code output changing unit 114 outputs the input from the code storage unit 113 to the intermediate code conversion unit 118. The intermediate code conversion unit 118 refers to the table stored in the intermediate code one-dimensional table unit 117 and converts the bit array 0011110 into the bit array 1111.
Converted to 110 and output.

Next, the code detecting section 111 outputs the bit sequence 001
1111000 is detected. Since this is an EOB code,
The code storage unit 113 stores the stored bit array 0010
Output 11000 and store bit 0. Since the code output changing unit 114 receives the EOB code, it outputs the input from the code storage unit 113 to the code conversion unit 115. The code conversion unit 115 refers to the table stored in the one-dimensional table unit 116 and
Is converted to a bit array 111011001 and output.

DCT bit pattern conversion counting section 21
0 means that the conversion of one block has been completed, so the count is 1
Increase to 2. Transform DCT coefficients for two blocks,
Since all the encoded data of the macro block has been obtained, the data output unit 211 outputs the encoded data.
The encoded data at this time is as follows.

0101001 1001111 00000001 000011111 1111100 1111101 1111110 111011001 As described above, there is no EOB code, and a bit array that can be decoded by a decoder corresponding to the 3DVLC method can be obtained.

In the image coding apparatus according to the present invention as in this embodiment, when coded data that can be decoded by a decoder for the 3DVLC method is output from coded data coded by the method using EOB, A procedure for temporarily converting a sequence of long codes into run and level information as two-dimensional information, a procedure for converting the information into DCT coefficients, and a procedure for again converting DCT coefficients into run, level and last information as three-dimensional information And a procedure for converting such information into a variable-length code can be omitted.

The DCT bit pattern converter 209
In this case, a configuration without the intermediate code conversion unit 118 and the intermediate code one-dimensional table unit 117 is also possible. FIG. 8 shows the configuration. According to this, it is also possible to convert only the bit arrangement of the sign of the last valid DCT coefficient in the block.

Also, by removing the quantization scale decoding unit and the quantization scale encoding unit from the configuration of the present embodiment, even for encoded data in which macroblock encoded data does not include quantization scale information. , DCT coefficients can be converted. The configuration is shown in FIG.
According to this, even when a non-quantized DCT coefficient array is encoded, when encoded data that can be decoded by a decoder for the 3DVLC method is output from encoded data encoded by a method using EOB, A procedure for temporarily converting a sequence of variable-length codes into run and level information which is two-dimensional information, a procedure for transforming the information into DCT coefficients,
It is possible to omit the procedure of converting the DCT coefficients into run, level, and last information, which are three-dimensional information, and the procedure of converting these information into variable-length codes.

It should be noted that the MPEG-4 encoded data
Regarding the conversion from the method using the 3D VLC method to the method using the EOB such as EG-1 encoded data, 3DVL
1 from code of method using C to code of method using EOB
Since a lookup table having a one-to-one correspondence can be created, it can be easily realized.

Also, the processing of some or all of the units in the image coded data conversion device and DCT bit pattern conversion unit shown in FIGS. 1, 2, 8, and 9 and the flow of the processing are described by a computer program. The present invention can be implemented by executing the program using a computer, or the processing procedure of the image coded data conversion method and the DCT bit pattern conversion method described with reference to these drawings. It is needless to say that the computer program can be executed by a computer, and it is needless to say that the program can be executed by a computer. The program is stored in a computer-readable storage medium such as an FD. Floppy disk (registered trademark))
It can be recorded on an MO, a ROM, a memory card, a CD, a DVD, a removable disk, or the like, stored and provided, and can be distributed via a network such as the Internet.

[0047]

According to the above invention, in a code conversion apparatus for outputting a code similar to that using the 3DVLC method, a DC conversion in a block required in the conventional 3DVLC method is performed.
It is possible to avoid a redundant procedure such as a check operation of the next code coefficient detection unit and a three-dimensional table search which are necessary every time the code of the T coefficient is output.

Also, in a code conversion device which receives coded data obtained by coding DCT coefficients by a method using EOB and outputs coded data that can be decoded by a decoder for the 3D VLC method, the arrangement of variable length codes is A procedure for converting the information into run and level information, which is dimensional information, a procedure for converting the information into DCT coefficients,
It is possible to omit a procedure for converting the information into run, level, and last information, which are dimensional information, and a procedure for converting the information into a variable length code.

[Brief description of the drawings]

FIG. 1 is a configuration diagram illustrating an embodiment of an image coded data conversion device according to the present invention.

FIG. 2 is a configuration diagram showing an embodiment of a DCT bit pattern converter according to the present invention;

FIG. 3 is a configuration diagram of a table stored in a motion vector table unit according to the embodiment of the present invention;

FIG. 4 is a configuration diagram of a table stored in a quantization scale table unit according to the embodiment of the present invention;

FIG. 5 is a configuration diagram of a table stored in a detection code table unit according to the embodiment of the present invention;

FIG. 6 is a configuration diagram of a table stored in a one-dimensional table unit according to the embodiment of the present invention;

FIG. 7 is a configuration diagram of a table stored in an intermediate code one-dimensional table unit according to the embodiment of the present invention;

FIG. 8 is a configuration diagram showing another embodiment of the DCT bit pattern conversion unit of the present invention.

FIG. 9 is a configuration diagram showing another embodiment of the image coded data conversion device of the present invention.

FIG. 10 is an explanatory diagram of a zigzag scan on 8 × 8 DCT coefficients.

FIG. 11 is a configuration diagram of a two-dimensional table corresponding to runs and levels.

FIG. 12 is a configuration diagram of a method using EOB.

FIG. 13 shows an example of a DCT coefficient array.

FIG. 14 is a configuration diagram of a three-dimensional table corresponding to a run, a level, and a last

FIG. 15 is a configuration diagram of a 3DVLC method.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 101 ... Zero count part 102 ... EOB coding change means 103 ... Run level setting part 104 ... 2D code output part 105 ... Two-dimensional table part 106 ... EOB code output part 107 ... Next code coefficient detection part 108 ... Three-dimensional table part 109 ... 3D code output unit 110 ... detection code table unit 111 ... code detection unit 112 ... EOB confirmation unit 113 ... code storage unit 114 ... code output change unit 115 ... code conversion unit 116 ... one-dimensional table unit 117 ... intermediate code one-dimensional Table section 118: Intermediate code conversion section 201: Data input section 202: Data input change section 203: Encoding mode decoding section 204: Encoding mode encoding section 205: Motion vector conversion section 206: Motion vector table section 207: Quantization Scale conversion unit 208: quantization scale table unit 209: DCT Ttopatan conversion unit 210 ... DCT bit pattern conversion counting unit 211 ... data output unit

Continued on front page F-term (reference) 5C059 KK41 MA00 MA23 MC11 MC24 MC38 ME01 ME13 ME17 PP04 PP16 RC14 SS20 UA02 UA05 5J064 BB13 BC01 BC02 BC05 BC16 BC29 CA02 CB12 CC01

Claims (8)

[Claims] 1. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
What is claimed is: 1. A code conversion device for outputting a bit sequence different from said input bit sequence, comprising: a detection code table section storing a bit sequence of said code and a bit sequence of an EOB code; The code or EO is referred to by referring to the bit arrangement recorded in the table section.
A code detection unit that detects a B code, a code storage unit that stores a code detected and input by the code detection unit, and a code stored in the code storage unit when the code detection unit detects a code other than the EOB code. EOB confirmation means for inputting a code, a one-dimensional table unit storing a table in which a code of another bit array is made to correspond to a bit array of each code, A code conversion unit that searches and outputs a code stored in a table unit; and, when the code detection unit detects a code other than an EOB code, outputs the code stored in the code storage unit. A code conversion device comprising: code output changing means for inputting a code stored in the code storage unit to the code conversion unit when a code is detected. 2. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
What is claimed is: 1. A code conversion device for outputting a bit sequence different from said input bit sequence, comprising: a detection code table section storing a bit sequence of said code and a bit sequence of an EOB code; The code or EO is referred to by referring to the bit arrangement recorded in the table section.
A code detection unit that detects a B code, a code storage unit that stores a code detected and input by the code detection unit, and a code stored in the code storage unit when the code detection unit detects a code other than the EOB code. EOB checking means for inputting a code, a one-dimensional table section storing a table corresponding to a code of another bit array for each code bit array, and a one-dimensional table for the input code. A code conversion unit for searching and outputting the codes stored in the unit, and a table in which the bit array of each code is associated with a code of a bit array different from the bit array recorded in the one-dimensional table unit. An intermediate code one-dimensional table unit that stores the following: an input code, an intermediate code conversion unit that searches and outputs a code stored in the intermediate code one-dimensional table unit, and the code detection unit E When an OB code is detected, the code stored in the code storage unit is input to the code conversion unit, and when the code detection unit detects a code other than the EOB code, the code stored in the code storage unit is transmitted to the middle. A code conversion device comprising: code output changing means for inputting to a code conversion unit. 3. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
A code conversion method for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; And a one-dimensional table that stores a table corresponding to the code of the bit array. The code or EOB is referred to from the input bit string by referring to the bit array recorded in the detection code table.
A code detection step of detecting a code, an EOB confirmation step of inputting a code to a code storage step when a code other than an EOB code is detected in the code detection step, and a detection of the code in the code detection step, and an input from the EOB confirmation step. The code accumulation step of accumulating the obtained code, and, when a code other than the EOB code is detected in the code detection step, the code accumulated in the code accumulation step is output, and the EOB code is detected in the code detection step. A code output change step of inputting the code stored in the code storage step to a code conversion step; and searching and outputting a code stored in the one-dimensional table for the code input from the code output change step. And a code conversion step. 4. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
A code conversion method for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; A one-dimensional table storing a table corresponding to the codes of the bit array of the above, and a table corresponding to the bit array of each code with a code of a bit array different from the bit array recorded in the one-dimensional table. And a one-dimensional table of intermediate codes that stores the code or EOB from the input bit sequence with reference to the bit array recorded in the detection code table.
A code detection step of detecting a code, an EOB confirmation step of inputting a code to a code storage step when a code other than an EOB code is detected in the code detection step, and a detection of the code in the code detection step, and an input from the EOB confirmation step. The code storing step of storing the obtained code; and, when a code other than the EOB code is detected in the code detecting step, the code stored in the code storing step is input to the intermediate code converting step, and the EOB code is detected in the code detecting step. When a code is detected, a code output change step of inputting the code stored in the code storage step to a code conversion step, and storing the code input from the code output change step in the intermediate code one-dimensional table. The intermediate code conversion step of searching for and outputting a code that has To the code input from flop, said code conversion step and a code conversion method characterized by comprising the to the search code stored in the one-dimensional table output. 5. An EOB code which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
What is claimed is: 1. A recording medium storing a code conversion program for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; A one-dimensional table storing a table corresponding to the code of another bit array, and referring to the bit array recorded in the detection code table from an input bit string, EOB
A code detection procedure for detecting a code, and when a code other than an EOB code is detected in the code detection procedure,
An EOB confirmation procedure for inputting a code in the code accumulation procedure, the code accumulation procedure for detecting in the code detection procedure and accumulating the code input from the EOB confirmation procedure, and a code other than an EOB code being detected in the code detection procedure In case,
A code output changing step of outputting the code stored in the code storing procedure, and inputting the code stored in the code storing procedure to a code converting procedure when an EOB code is detected in the code detecting step; The code conversion procedure of searching and outputting the code stored in the one-dimensional table with respect to the code input from the procedure; and executing the code conversion program provided by the computer. A recording medium on which a code conversion program characterized by being recorded on a simple recording medium is recorded. 6. An EOB code which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
What is claimed is: 1. A recording medium storing a code conversion program for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; In contrast, a one-dimensional table storing a table corresponding to codes of different bit arrays, and a bit array of a bit array different from the bit array recorded in the one-dimensional table for the bit array of each code. An intermediate code one-dimensional table storing a table corresponding to a code, and referring to a bit array recorded in the detection code table from an input bit string, the code or EOB.
A code detection procedure for detecting a code, and when a code other than an EOB code is detected in the code detection procedure,
An EOB confirmation procedure for inputting a code in the code accumulation procedure, the code accumulation procedure for detecting in the code detection procedure and accumulating the code input from the EOB confirmation procedure, and a code other than an EOB code being detected in the code detection procedure In case,
A code output changing step of inputting the code stored in the code storing procedure to an intermediate code converting procedure and inputting the code stored in the code storing procedure to the code converting procedure when an EOB code is detected in the code detecting step; For the code input from the code output change procedure, the intermediate code conversion procedure of searching and outputting the code stored in the intermediate code one-dimensional table, and the code input from the code output change procedure. The code conversion procedure for searching for and outputting the code stored in the one-dimensional table is recorded on a computer-readable recording medium in order to execute the code conversion program provided on the computer. The recording medium which recorded the code conversion program characterized by the above. 7. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
A code conversion program for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; And a one-dimensional table that stores a table corresponding to the code of the bit array. The code or EOB is referred to from the input bit string by referring to the bit array recorded in the detection code table.
A code detection procedure for detecting a code, and when a code other than an EOB code is detected in the code detection procedure,
An EOB confirmation procedure for inputting a code in the code accumulation procedure, the code accumulation procedure for detecting in the code detection procedure and accumulating the code input from the EOB confirmation procedure, and a code other than an EOB code being detected in the code detection procedure In case,
A code output changing step of outputting the code stored in the code storing procedure, and inputting the code stored in the code storing procedure to a code converting procedure when an EOB code is detected in the code detecting step; A code conversion step of searching for a code stored in the one-dimensional table for a code input from the procedure and outputting the code. 8. An EOB code, which is a unique bit array,
In addition to the bit array of the EOB code, a bit string composed of a code that can take a bit array of a plurality of patterns is input,
A code conversion program for outputting a bit sequence different from the input bit sequence, comprising: a detection code table in which a bit sequence of the code and a bit sequence of an EOB code are recorded; A one-dimensional table storing a table corresponding to the codes of the bit array of the above, and a table corresponding to the bit array of each code with a code of a bit array different from the bit array recorded in the one-dimensional table. And a one-dimensional table of intermediate codes that stores the code or EOB from the input bit sequence with reference to the bit array recorded in the detection code table.
A code detection procedure for detecting a code, and when a code other than an EOB code is detected in the code detection procedure,
An EOB confirmation procedure for inputting a code in the code accumulation procedure, the code accumulation procedure for detecting in the code detection procedure and accumulating the code input from the EOB confirmation procedure, and a code other than an EOB code being detected in the code detection procedure In case,
A code output changing step of inputting the code stored in the code storing procedure to an intermediate code converting procedure and inputting the code stored in the code storing procedure to the code converting procedure when an EOB code is detected in the code detecting step; For the code input from the code output change procedure, the intermediate code conversion procedure of searching and outputting the code stored in the intermediate code one-dimensional table, and the code input from the code output change procedure. A code conversion procedure for searching for and outputting a code stored in the one-dimensional table.