JP2004080634A - Code converter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a VLC coder which can be used commonly by MPEG2 and MPEG4. <P>SOLUTION: The VLC coder includes a table storage 34 for storing conversion tables in which a correspondence relation between a syntax and a VLC code string is described. A selection instruction for selecting any one of the conversion tables in the table storage 34 is provided from an external CPU to the VLC coder. A VLC coder/decoder selects any one conversion table according to the selection instruction received from the CPU, and performs VLC encoding or decoding operation on the basis of the selected conversion table. Consequently, when all the conversion tables associated with MPEG2 and MPEG4 are previously stored in the table storage 34, VLC coding corresponding to a stream of MPEG2 and MPEG4 can be realized by programming. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to variable-length coding, variable-length code decoding, fixed-length coding, and fixed-length code decoding applied to, for example, MPEG (Motion Picture Experts Group) 1, MPEG2, MPEG4, and H263. The present invention relates to a transcoder that performs the following.
[0002]
[Prior art]
At present, various types of image coding standards such as MPEG (Motion Picture Experts Group), MPEG2, MPEG4 and H263 have been proposed. In such an image coding standard, it is defined that various information (syntax) for restoring image data is coded. For example, a DCT coefficient generated by performing discrete cosine transform on an image signal in a spatial domain, a motion vector generated by performing motion prediction between frames, a macroblock type, a motion vector prediction direction, and the like. It is prescribed to encode. What kind of syntax is included in the encoded data is determined by the format of the encoded data, and the encoding order is also determined by the format.
[0003]
Further, in image coding standards such as MPEG1, MPEG2, MPEG4 and H263, various syntaxes are coded in a hierarchical structure. For example, in each of the above-described image coding standards, a sequence layer in which syntax relating to a series of screen groups having the same attribute is described, a GOP layer in which syntax relating to a screen group serving as a random access unit is described, and one screen Layer, which describes the syntax relating to a small screen obtained by slicing a single screen in the horizontal direction, and the syntax relating to a two-dimensional pixel block obtained by further subdividing the slice layer. It is stipulated that a hierarchical structure such as a macro block layer is formed.
[0004]
Further, the macroblock layer in each of the above-described hierarchical structures includes a syntax having the largest code amount, such as a DCT coefficient and a motion vector. Therefore, in each of the above coding standards, a variable length coding process is performed on the syntax included in the macroblock layer, and an efficient coding process is performed. In particular, with respect to DCT coefficients, run-length encoding in which a variable-length code string is assigned to a combination of 0 runs and levels is performed.
[0005]
When performing encoding and decoding of a variable-length code string, usually, a conversion from syntax to a code string or a reverse conversion thereof is performed by referring to a conversion table describing the correspondence between syntax and code string. Is performed. A conversion table is prepared for each type of syntax. That is, a conversion table for DCT coefficients, a conversion table for motion vectors, and a conversion table for each of various parameters are prepared.
[0006]
[Problems to be solved by the invention]
By the way, each image coding standard such as MPEG1, MPEG2, MPEG4, and H263 defines its own syntax, and the content of the syntax differs for each standard. Therefore, the conversion table is naturally different. Even if the same syntax is used between the standards, the relationship between the value of the syntax and the code string of the corresponding variable length code is different, and the encoding order is also different.
[0007]
Therefore, for example, when manufacturing a common encoder or decoder corresponding to a plurality of image coding standards, a variable length coding and decoding circuit is separately developed for each standard, and all the circuits corresponding to each standard are completely developed. It had to be mounted on the encoder and the decoder. That is, one encoder and one decoder have to be equipped with variable length coding and decoding circuits as many as the number of image coding standards.
[0008]
In addition, after developing an encoder or decoder that supports a certain image coding standard, if a new syntax is extended to that image coding standard, or if the coding order is changed, the variable length code The decoding and decoding circuit cannot support the expanded or changed standard, so that a new circuit design has to be performed.
[0009]
The present invention has been proposed in view of such a conventional situation, and can reduce the circuit scale even when supporting a plurality of standards. Further, a new coding standard has been proposed. It is an object of the present invention to provide a highly scalable code conversion device that can easily cope with the new standard even when the code conversion is performed.
[0010]
[Means for Solving the Problems]
The code conversion device according to the present invention is a conversion table storage unit that stores a plurality of conversion tables each describing a correspondence relationship between syntax indicating a specific information content and a code string for each type of the information content. Encoding processing for converting the generated syntax into a corresponding code string according to the description of the conversion table, and combining a plurality of code strings generated by the code conversion to generate one data stream; and / or Encoding / decoding means for performing a decoding process of extracting a code string from an input data stream and performing code conversion from the extracted code string into a corresponding syntax according to the description of the conversion table; Command storage means for storing command data indicating table selection information, wherein the encoding / decoding means is stored in the command storage means. The conversion table shown in and have instruction data selected one of the conversion table storing means, and performs the code conversion based on the conversion table selected.
[0011]
In this transcoder, a plurality of conversion tables classified according to the type of information content are stored in the conversion table storage means. The conversion table describes information describing the correspondence between the syntax and the code string.
[0012]
The transcoder stores command data from the outside. The instruction data describes at least selection information for selecting one conversion table from a plurality of conversion tables stored in the conversion table storage unit. The encoding / decoding means selects one conversion table based on the instruction data stored in the instruction storage means and refers to the selected conversion table to perform code conversion processing from syntax to a code string or a code string. Performs the conversion process from to the syntax.
[0013]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, as an embodiment of the present invention, a VLC code conversion device used in an image coding and decoding device compatible with both MPEG2 (ISO / IEC 13818-2) and MPEG4 (ISO / IEC 14496-2) systems. This will be described in detail with reference to the drawings.
[0014]
When performing the encoding process, the VLC code conversion device according to the embodiment of the present invention receives various information (hereinafter, referred to as syntax) specified in an MPEG2 or MPEG4 macroblock layer, and inputs them. Is subjected to a variable length encoding process (or a fixed length encoding process) to generate a predetermined code string. Then, the VLC code converter synthesizes the generated code strings to generate a data stream corresponding to MPEG2 or MPEG4. When performing decoding processing, the VLC code converter receives a data stream corresponding to MPEG2 or MPEG4 and divides the input data stream into individual code strings of variable-length codes (or fixed-length codes). I do. Then, the VLC code conversion device performs variable length code decoding processing (or fixed length code decoding processing) on each of the divided code strings, and performs syntax specified in the MPEG2 or MPEG4 macroblock layer. Is extracted.
[0015]
A code string generated by performing variable-length coding processing (or fixed-length coding processing) on the syntax is referred to as a VLC code string.
[0016]
FIG. 1 shows a block diagram of a VLC code converter 1 according to an embodiment of the present invention.
[0017]
As shown in FIG. 1, the VLC code converter 1 includes a code converter 11, a DCT coefficient buffer 12, an MV buffer 13, a parameter buffer 14, a bit stream buffer 15, a CPU (Central Processing Unit) 16, It is configured with.
[0018]
At the time of encoding, the code conversion unit 11 converts the syntax specified in the MPEG2 or MPEG4 macroblock layer into a VLC code string, and synthesizes the generated VLC code string to generate a bit stream. At the time of decoding, the code conversion unit 11 separates a bit stream corresponding to the MPEG2 or MPEG4 format into individual VLC code strings and performs a process of converting each VLC code string into syntax.
[0019]
The DCT coefficient buffer 12 stores a DCT coefficient defined as one of the syntaxes of MPEG2 and MPEG4. The MV buffer 13 stores a motion vector defined as one of the syntaxes of MPEG2 and MPEG4. The parameter buffer 14 stores the syntax other than the DCT coefficient and the motion vector among the syntaxes defined by MPEG2 or MPEG4, for example, the type of macroblock, the prediction direction of the motion vector, and the like. Note that, of the syntaxes defined by MPEG2 and MPEG4, syntaxes other than DCT coefficients and motion vectors are hereinafter referred to as parameters.
[0020]
The bit stream buffer 15 stores a bit stream of MPEG2 or MPEG4. The CPU 16 controls the operation of the code conversion unit 11.
[0021]
At the time of encoding, the VLC code conversion device 1 configured as described above reads the syntax (DCT coefficient, motion vector, and parameter) stored in the DCT coefficient buffer 12, the MV buffer 13, and the parameter buffer 14 and performs variable reading. A VLC code string is generated by performing long coding (or fixed length coding), and the generated VLC code string is combined with a bit stream corresponding to MPEG2 or MPEG4 on the bit stream buffer 15. Also, at the time of decoding, the VLC code conversion device 1 reads a bit stream corresponding to MPEG2 or MPEG4 stored in the bit stream buffer 15 to separate individual VLC code strings, and separates each VLC code string. A variable-length code is decoded (or a fixed-length code is decoded) to generate a syntax, and the generated syntax is stored in a buffer (DCT coefficient buffer 12, MV buffer 13, or parameter buffer 14). Buffer).
[0022]
Next, the internal configuration of the code conversion unit 11 will be described.
[0023]
The code conversion unit 11 includes a run label analysis unit 21, an MV analysis unit 22, an encoding / decoding unit 23, an instruction register 24, a control register 25, and a control unit 26.
[0024]
The run-level analyzer 21 reads out the DCT coefficients from the DCT coefficient buffer 12 in a predetermined scan order at the time of encoding, and performs a conversion process of converting the DCT coefficients arranged in the read order into run-level information. , And its inverse conversion processing. The so-called run-level two-dimensional variable length coding is performed on the DCT coefficient. The run is the number of leading zero values in the sequence of DCT coefficients, and the level is the value of the non-zero DCT coefficient following the run. Run-level information is coded information representing a combination of one run and one level. In MPEG2 and MPEG4, variable-length coding is performed on DCT coefficients based on the run-level information. At the time of encoding, the run-level analysis unit 21 reads out DCT coefficients from the DCT coefficient buffer 12 in a predetermined scan order, generates run-level information, and supplies the run-level information to the encoding / decoding unit 23. Also, at the time of decoding, the run level analysis unit 21 restores the values of a plurality of DCT coefficients arranged in a predetermined scan order from the run-level information supplied from the encoding / decoding unit 23, and restores the restored DCT coefficients. Is stored in the DCT coefficient buffer 12. In MPEG2 and MPEG4, the DCT coefficient of the DC component of an intra macroblock is not subjected to run-level two-dimensional variable-length coding, but is subjected to fixed-length coding for the level. Therefore, only the level value of the DCT coefficient of the DC component of the intra macroblock is exchanged with the encoding / decoding unit 23.
[0025]
At the time of encoding, the MV analysis unit 22 calculates a difference between the values of the motion vectors of the continuous macroblocks, and converts the motion vector into a difference code based on the difference value. Further, at the time of decoding, the MV analysis unit 22 performs a process of restoring a motion vector from a difference code. At the time of encoding, the MV analysis unit 22 reads a motion vector from the MV buffer 13 and performs differential coding, and supplies the differentially coded motion vector to the encoding / decoding unit 23. At the time of decoding, the MV analysis unit 22 restores a motion vector from the difference code supplied from the encoding / decoding unit 23, and stores the restored motion vector in the MV buffer 13. Note that differential coding is not performed on the head macroblock, intra macroblock, non-MC, or the like of a slice in a P picture, and the value of the motion vector is VLC-coded as it is.
[0026]
The encoding / decoding unit 23 performs a conversion process between the syntax and the VLC code string with reference to a conversion table stored therein.
[0027]
At the time of encoding, the encoding / decoding unit 23 receives DCT coefficients, motion vectors, and parameters as syntax. The DCT coefficient is input as run-level information by the run-level analysis unit 21, and the motion vector is input as a differential code by the MV analysis unit 22. The parameters are input directly from the parameter buffer 14. At the time of encoding, the encoding / decoding unit 23 converts the input syntax into a corresponding VLC code string by referring to the conversion table selected by the control unit 26, combines them, and complies with MPEG2 or MPEG4. Generate a bitstream.
[0028]
At the time of decoding, the encoding / decoding unit 23 receives a bit stream corresponding to MPEG2 or MPEG4 from the bit stream buffer 15. The encoding / decoding unit 23 separates the input bit stream into respective VLC code strings. The encoding / decoding unit 23 converts each VLC code string into a corresponding syntax with reference to the conversion table selected by the control unit 26. The encoding / decoding unit 23 stores the DCT coefficients in the converted syntax in the DCT coefficient buffer 12 via the run-level analysis unit 21 and stores the motion vectors in the MV buffer 13 via the MV analysis unit 22. , And the parameters are stored in the parameter buffer 14.
[0029]
The instruction register 24 is supplied with instruction data from the CPU 16 and stores the instruction data. The details of the instruction data will be described later.
[0030]
The control register 25 is supplied with control data from the CPU 16 and stores the control data. Details of the control data will be described later.
[0031]
The control unit 26 controls the operation of the encoding / decoding unit 23 with reference to the instruction data stored in the instruction register 24 and the control data stored in the control register 25. Details of the operation control process of the control unit 26 will be described later.
[0032]
Next, the configuration of the encoding / decoding unit 23 will be described in detail. FIG. 2 shows a block diagram of the encoding / decoding unit 23.
[0033]
The encoding / decoding unit 23 includes a table converter 31, a stream analyzer 32, and a stream combiner 33.
[0034]
An MPEG2 or MPEG4 bit stream is input to the stream analyzer 32 at the time of encoding. The MPEG2 and MPEG4 bit streams are streams formed by arranging various types of data in a predetermined hierarchical data structure. In particular, the macroblock layer is generated by performing variable-length coding or fixed-length coding on syntax. Data strings (VLC code strings) are arranged according to a predetermined data structure. The macroblock layer portion in the bit stream includes, for example, a VLC code string indicating information contents such as a DCT coefficient, a motion vector, a macroblock address, and a macroblock address. However, the maximum code length of a VLC code string (or the code length of fixed-length code data) differs for each information content. The stream analyzer 32 receives information indicating the maximum code length of the VLC code string contained in the bit stream from the control unit 26, and converts the input bit stream into individual VLC codes based on the maximum code length information. Separate into columns. The information indicating the maximum code length is given by the instruction data stored in the instruction register 24.
[0035]
Upon decoding, the VLC code string output from the table converter 31 is input to the stream combiner 33. The stream combiner 33 combines the input VLC code strings according to the format of MPEG2 or MPEG4 to generate a bit stream corresponding to MPEG2 or MPEG4. The combined bit stream is output to the outside as hierarchical data of the MPEG2 or MPEG4 macroblock layer portion, and combined with a predetermined portion of the upper layer data stream.
[0036]
At the time of encoding, the table converter 31 receives a syntax and outputs a VLC code string. Also, at the time of decoding, the table converter 31 receives the individual VLC code strings or fixed-length code data divided by the stream analyzer 32 and outputs the corresponding syntax.
[0037]
The table converter 31 refers to a conversion table that internally stores in advance code conversion processing from syntax to a VLC code string at the time of encoding and code conversion processing from a VLC code string to syntax at the time of decoding. Do it.
[0038]
FIG. 3 shows the internal configuration of the table converter 31.
[0039]
The table converter 31 includes an encoding / decoding table storage unit 34, an offset table storage unit 35, an address generation unit 36, an offset table temporary storage unit 37, an address combining unit 38, and an encoding / decoding table temporary unit. The storage unit 39 is provided.
[0040]
The encoding / decoding table storage unit 34 stores a plurality of encoding conversion tables (encoding tables) and a plurality of decoding conversion tables (decoding tables).
[0041]
In the encoding table, the value of the addressed syntax is described as an index word. Addressed syntax is data obtained by converting syntax values so that it can be easily handled as table data. The value of the syntax is, for example, a minus value or a value that is not simply incremented, and it is difficult to handle as it is as table data. Therefore, in the present encoding table, the values of these syntaxes are simplified according to a predetermined conversion method, and are addressed so that they can be used as index words of the table. In the encoding table, a corresponding VLC code string is assigned to each index word. The assigned VLC code string is a unique code string in the table. Therefore, when the syntax of an arbitrary value is input to the encoding table, one VLC code string corresponding to the syntax can be output.
[0042]
In the decoding table, the value of the addressed VLC code string is described as an index word. The VLC code string converted into an address is data obtained by converting the value of the VLC code string so as to be easily handled as table data. The values of the VLC code string have different code lengths, and if they are used as they are as the index words of the table data, the table becomes very large and the blank area becomes enormous, making it very difficult to handle. Therefore, in the present decoding table, the values of these VLC code strings are simplified according to a predetermined conversion method, and are addressed so that they can be used as index words of the table. In the decoding table, a corresponding syntax value is assigned to each index word. For this reason, when a certain arbitrary VLC code string is input, the decoding table can output one syntax value corresponding to the code string.
[0043]
In addition, the encoding / decoding table storage unit 34 stores independent conversion tables for at least each type of syntax included in the macroblock layer, each type of bit stream, and each encoding / decoding. . For example, a DCT coefficient conversion table corresponding to MPEG2, a DCT coefficient conversion table corresponding to MPEG4, a motion vector conversion table corresponding to MPEG2, a motion vector conversion table corresponding to MPEG4, a macroblock type corresponding to MPEG2 And a conversion table generated independently for each type of information, each type of stream, and each encoding / decoding, such as a conversion table of a macroblock type corresponding to MPEG4.
[0044]
The offset table storage unit 35 stores various offset tables. In the offset table, information of the upper part (upper address) of the addressed syntax or the addressed VLC code string is described as an index word, and a predetermined value in the conversion table for each index word is described. 5 is a table to which an offset address indicating an index position is assigned.
[0045]
As shown in FIG. 4, the information in the conversion table can be divided into regions in which valid index values are continuously described. In particular, the VLC code string converted into an address can be divided into regions in which effective values are gathered discretely since the data length may vary. In the conversion table, an upper address is assigned to each area, and a lower address indicating the order in the area is assigned to each piece of information in the area. The offset table is a table that, when an upper address is input, outputs a head address of an area specified by the upper address.
[0046]
The address generation unit 36 receives data before conversion (syntax for encoding, VLC code string for decoding), and outputs an upper address specifying an area in the conversion table and a lower address indicating the order in the area. Generate. The upper address is supplied to an offset table temporary storage unit 37, and the lower address is supplied to an address synthesizing unit 38.
[0047]
To the offset table temporary storage unit 37, one of the plurality of offset tables stored in the offset table 35, which is selected by a table selection signal given from the control unit 26, is transferred. Store the table. The offset table temporary storage unit 37 outputs an offset address corresponding to the input upper address with reference to the stored offset table. The offset address is supplied to the address synthesis unit 38. The offset table transferred to the offset table temporary storage unit 37 is selected by the control unit 26 based on the instruction data stored in the instruction register 24.
[0048]
The address combining unit 38 receives the offset address and the lower address. The address synthesizer 38 generates an address obtained by adding the offset address and the lower address. The added address becomes an address for specifying the index word of the conversion table. The address generated by the address synthesizing unit 38 is supplied to an encoding / decoding table temporary storage unit 39.
[0049]
The encoding / decoding table temporary storage unit 39 stores one of a plurality of conversion tables stored in the encoding / decoding table storage unit 34 selected by a table selection signal given from the control unit 26. The encoding / decoding table is transferred and stores the transferred conversion table. The encoding / decoding table temporary storage section 39 refers to the stored conversion table, and converts the converted data (VLC code string at the time of encoding, thin Tax). The conversion table transferred to the encoding / decoding table temporary storage unit 39 is selected by the control unit 26 based on the instruction data stored in the instruction register 34.
[0050]
As described above, the encoding / decoding unit 23 stores a plurality of various conversion tables in advance, reads one conversion table necessary for performing variable-length encoding (or decoding), and reads the conversion table. Based on the read conversion table, conversion from syntax to VLC code string or conversion processing from VLC code string to syntax is performed. As described above, the encoding / decoding unit 23 is configured to read out the conversion table as needed. Therefore, even if only the correction and addition of the conversion table are performed without changing the hardware configuration itself, a new conversion table can be obtained. It can support various syntaxes and formats.
[0051]
Further, when a VLC code string is used as an index of a table, a useless variable area increases due to the nature of the variable length code, and the table data becomes large. However, the encoding / decoding unit 23 divides the data in the conversion table into regions, and generates addresses specifying each region in the offset table. Since the encoding / decoding unit 23 uses the two-stage code conversion table as described above, even if variable-length codes are handled, there is no need to create a useless blank area in the table. Efficiency can be improved.
[0052]
Next, the instruction data and control data stored in the instruction register 24 and the control register 25 will be described.
[0053]
Instruction data is written into the instruction register 24, and control data is written into the control register 25. It is the CPU 16 that writes the command data and the control data. The CPU 16 has various processing programs according to the processing status, and stores an instruction set (a set in which a plurality of instruction data are listed) and control data necessary for the processing programs in the instruction registers 24 and 25. Store.
[0054]
It is the control unit 26 of the code conversion unit 11 that performs operation control with reference to the written command data and control data.
[0055]
FIG. 5A shows the data structure of the instruction data at the time of encoding, and FIG. 5B shows the data structure of the instruction data at the time of decoding.
[0056]
The instruction data at the time of encoding includes mode information 41, maximum bit length information 42, parameter identification information 43, buffer selection / fixed length encoded value information 44, and the number of loops 45. The instruction data at the time of decoding includes mode information 41, maximum bit length information 42, and the number of decodings 46.
[0057]
The mode information 41 indicates the type of syntax to be encoded or decoded. The control unit 26 of the code conversion unit 11 refers to the mode information 41 and selects one necessary conversion table (and offset table) from a plurality of conversion tables (and a plurality of offset tables). The control unit 26 of the code conversion unit 11 issues a control instruction specifying the selected conversion table (and offset table) to the encoding / decoding unit 23, and refers to the specified conversion table (offset table). To perform code conversion.
[0058]
6 to 9 show specific code examples defined as the mode information 41. The code example shown here is a 5-bit code example. FIG. 6 is a code example of mode information in a case where encoding processing is performed in the MPEG2 system. FIG. 7 is a code example of mode information in a case where encoding processing is performed by the MPEG4 system. FIG. 8 is a code example of mode information in a case where decoding processing is performed by the MPEG2 method. FIG. 9 is a code example of mode information in a case where decoding processing is performed by the MPEG4 system.
[0059]
The maximum bit length information 42 is information indicating the maximum bit length of a VLC code string to be encoded or decoded. The control unit 26 of the code conversion unit 11 issues control information designating the maximum bit length of the VLC code string to the encoding / decoding unit 23 based on the maximum bit length information 42, and individually outputs the MPEG2 or MPEG4 bit stream. Is divided into VLC code strings.
[0060]
The parameter identification information 43 is information for identifying whether the syntax to be encoded is a parameter, a DCT coefficient or a motion vector. When the parameter identification information 43 indicates that the syntax to be encoded is a parameter, the control unit 26 performs control to transfer data from the parameter buffer 14 to the encoding / decoding unit 23. Note that the parameter identification information is set only at the time of encoding.
[0061]
The buffer selection / fixed-length coded value information 44 has different information contents according to the description of the parameter identification information 43. When the syntax to be encoded is indicated in the parameter identification information 43 which is a parameter, the value of the syntax to be fixed-length encoded is described in the buffer selection / fixed-length encoded value information 44. In this case, when the value of the syntax matches the value indicated in the buffer selection / fixed-length coded value information 44, the control unit 26 performs the fixed-length coding process on the syntax. On the other hand, if the parameter identification information 43 indicates that the syntax to be coded is not a parameter, the buffer selection / fixed-length coded value information 44 indicates whether the syntax to be coded is a DCT coefficient or Identification information as to whether the vector is a vector is described. The control unit 26 controls the transfer of data from the DCT coefficient buffer 12 to the encoding / decoding unit 23 if the syntax to be encoded in the buffer selection / fixed-length encoding information 44 is indicated as a DCT coefficient. If the syntax to be encoded in the buffer selection / fixed-length encoding information 44 indicates that the motion vector is a motion vector, control is performed to transfer data from the MV buffer 13 to the encoding / decoding unit 23. Note that the buffer selection / fixed-length coded value information 44 is set only at the time of coding.
[0062]
The number-of-loops information 45 indicates the number of indexes for continuously performing the encoding process using the conversion table specified by the instruction data. For example, DCT coefficients and motion vectors are continuously encoded in a stream. Therefore, when consecutive indices of the same type are encoded with reference to one conversion table, the encoding process is repeated for the number of times described in the loop number information 45. As a result, encoding processing for a plurality of syntaxes can be performed based on one instruction data, and the number of instructions can be reduced. The loop number information 45 is set only at the time of encoding.
[0063]
The number-of-decoding times information 46 indicates the number of times of performing the decoding process using the conversion table specified by the instruction data. For example, DCT coefficients and motion vectors are continuously encoded in a stream. Therefore, when decoding consecutive indices of the same type with reference to one conversion table, the decoding process is repeated for the number of times described in the decoding number information 46. As a result, a plurality of decoding processes can be performed based on one instruction data, and the number of instructions can be reduced. Note that the decoding count information 46 is set only at the time of decoding.
[0064]
FIG. 10 shows the data structure of the control data. The control data is set for each code value described as mode information. That is, control data is set for each of the 5-bit values shown in FIGS. The control data is written into the control register 25 by the CPU 16 and is referred to by the control unit 26. After reading the instruction data from the instruction register 24, the control unit 26 first reads the control data corresponding to the mode information of the instruction data from the control register 25, and executes the contents described in the control data.
[0065]
The control data includes an END flag 47, an FLC flag 48, an INTRA flag 49, a DCT flag 50, an MV flag 51, and a PARAM flag 52.
[0066]
The END flag 47 is a flag indicating the last macroblock. That is, if the bit of the END flag 47 is set, the control data indicates that the code conversion processing performed based on the mode information is processing for the last macroblock. This indicates that the code conversion process performed based on is not the process for the last macroblock. When the END flag 47 of the control data is set, the control unit 26 performs processing on the last macroblock when performing processing based on the instruction data. When the END flag 47 is not set, the control unit 26 performs processing based on the instruction data. When processing is performed, processing is performed on a normal macroblock.
[0067]
The FLC flag 48 is a flag indicating whether to perform fixed-length processing or variable-length processing. That is, if the bit of the FLC flag 48 is set, the control data indicates that the code conversion process performed based on the mode information is a fixed-length coding process or a fixed-length decoding process. If not, it indicates that the code conversion process performed based on the mode information is a variable length coding or a variable length decoding process. If the FLC flag 48 of the control data is set, the control unit 26 performs a process corresponding to the fixed-length encoding process, for example, when performing an analysis of a bit stream, an address generation process of a conversion table, and the like. If the FLC flag is not set to 48, a process corresponding to the variable-length encoding process is performed, for example, when performing bit stream analysis or conversion table address generation.
[0068]
The INTRA flag 49 is a flag indicating whether to perform processing corresponding to an intra macroblock or to perform processing corresponding to an inter macroblock. That is, if the bit of the INTRA flag 49 is set, it indicates that the code conversion process performed based on the mode information is a process for an intra macroblock. If the bit is not set, the code conversion process is performed based on the mode information. This indicates that the code conversion process is a process for an inter macroblock. If the INTRA flag 49 of the control data is set, the control unit 26 controls the run level analysis unit 21 and the MV analysis unit 22 to perform processing corresponding to the intra macro block, for example. If the INTRA flag 49 of the data is not set, for example, control is performed so that the run level analysis unit 21 and the MV analysis unit 22 perform processing corresponding to the inter macro block.
[0069]
The DCT flag 50, the MV flag 51, and the PRAM flag 52 are flags for selecting a buffer for storing the decoded syntax. That is, the control unit 26 stores the decoded syntax in the DCT coefficient buffer 12 if the bit of the DCT flag 50 is set, and stores the decoded syntax in the MV buffer 13 if the bit of the MV flag 51 is set. If the bit of the PRAM flag 52 is set, the decoded syntax is stored in the parameter buffer 14.
[0070]
As described above, in the VLC code conversion device 1, the instruction data and control data are written by the CPU 16, and the control unit 26 selects a conversion table and controls various code conversion processes based on the instruction data and control data. I have. For this reason, for example, it is possible to freely program the types and the order of the indexes to be encoded or decoded. Therefore, for example, encoding processing and decoding processing corresponding to a plurality of streams, correspondence to a new encoded stream, and the like can be performed only by changing the program.
[0071]
Next, the contents of processing performed by the code conversion unit 11 during encoding will be described with reference to the flowchart in FIG.
[0072]
First, writing of control data to the control register 25 (step S11) and writing of an instruction set to the instruction register 24 are performed by the CPU 16 in advance (step S12). The instruction set is a data group composed of a plurality of instruction data, and indicates the instruction data and the execution order.
[0073]
Subsequently, the control unit 26 of the code conversion unit 11 receives an encoding start command from the CPU 16. The control unit 26 starts the encoding process when the encoding start command is given (Step S13).
[0074]
Subsequently, the control unit 26 of the code conversion unit 11 reads out the first one instruction data from the instruction set stored in the instruction register 24 (Step S14).
[0075]
Subsequently, the control unit 26 of the code conversion unit 11 refers to the mode information described in the read instruction data, and reads control data corresponding to the mode information from the control register 25. The control unit 26 performs setting control of the content described in the read control data (step S15).
[0076]
Subsequently, the control unit 26 of the code conversion unit 11 selects an offset table and a conversion table specified by the mode information of the command data, and stores the offset table and the conversion table in the encoding / decoding table storage unit 34 and the offset table storage unit 35. The data is transferred to the temporary storage 37 and the conversion table temporary storage 39 (step S16).
[0077]
Subsequently, when the syntax for performing the encoding process is a parameter, the control unit 26 of the transcoding unit 11 reads one parameter from the parameter buffer 14 and transfers the parameter to the encoding / decoding unit 23. Alternatively, when the syntax for performing the encoding process is a DCT coefficient or a motion vector, the control unit 26 of the transcoding unit 11 causes the run-level analysis unit 21 or the MV analysis unit 22 to perform an arithmetic process. The difference code of the information or the motion vector is transferred to the encoding / decoding unit 23 (Step S17).
[0078]
Subsequently, the control unit 26 of the code conversion unit 11 issues a control command to the encoding / decoding unit 23 to perform a conversion process from the input syntax (DCT coefficient, motion vector or parameter) to address information. (Step S20), a process of converting into a VLC code string corresponding to the syntax converted into address information (Step S21), and a bit stream synthesizing process of the VLC code string (Step S22) are performed.
[0079]
Subsequently, the control unit 26 of the code conversion unit 11 determines whether code conversion processing has been performed for the number of loops indicated in the instruction data (step S23). If the code conversion process for the number of loops has not been performed, the process returns to step S17, and the coding process for the next syntax is performed again using the same conversion table.
[0080]
When the code conversion processing for the number of loops has been completed, subsequently, it is determined whether or not the code conversion processing has been completed for all the instruction data in the instruction set. If the code conversion has not been completed for all the instruction data in the instruction set, the process returns to step S14, reads the next instruction data from the instruction register 24, and continues the processing. When the encoding has been completed, the present encoding processing ends.
[0081]
Next, the processing contents of the code conversion unit 11 at the time of decoding will be described with reference to the flowchart of FIG.
[0082]
First, writing of control data to the control register 25 (step S31) and writing of an instruction set to the instruction register 24 are performed by the CPU 16 in advance (step S32).
[0083]
Subsequently, the control unit 26 of the code conversion unit 11 receives an encoding start command from the CPU 16. When the coding start command is given, the control unit 26 of the code conversion unit 11 starts coding processing (step S33).
[0084]
Subsequently, the control unit 26 of the code conversion unit 11 reads out the first one instruction data from the instruction set stored in the instruction register 24 (Step S34).
[0085]
Subsequently, the control unit 26 of the code conversion unit 11 refers to the mode information described in the read instruction data, and reads control data corresponding to the mode information from the control register 25. The control unit 26 performs setting control of the content described in the read control data (step S35).
[0086]
Subsequently, the control unit 26 of the code conversion unit 11 selects an offset table and a conversion table specified by the mode information of the instruction data, and outputs the offset table and the offset table from the decoding / decoding table storage unit 34 and the offset table storage unit 35. The data is transferred to the temporary storage section 37 and the conversion table temporary storage section 39 (step S36).
[0087]
Subsequently, the control unit 26 of the code conversion unit 11 analyzes the bit stream in the bit stream buffer 15 and extracts one VLC code string (Step S37).
[0088]
Subsequently, the control unit 26 of the code conversion unit 11 issues a control command to the decoding / decoding unit 23 to convert the extracted VLC coded sequence into address information (step S38), A conversion process (step S39) to a syntax corresponding to the VLC code string thus performed is performed.
[0089]
Subsequently, the control unit 26 of the code conversion unit 11 determines whether the decoded syntax is a DCT coefficient or a motion vector (Step S40). If it is not a DCT coefficient or a motion vector, that is, if it is a parameter, the decoded parameter is stored in the parameter buffer 14 (step S41). When the decoded syntax is a DCT coefficient or a motion vector, the control unit 26 of the code conversion unit 11 causes the run-level analysis unit 21 or the MV analysis unit 22 to perform an arithmetic process (Step S42), and The motion vector is stored in the DCT coefficient buffer 12 or the MV buffer 13 (Step S43).
[0090]
Subsequently, the control unit 26 of the code conversion unit 11 determines whether or not code conversion processing has been performed for the number of decodings indicated in the instruction data (step S44). If the code conversion process for the number of decoding has not been performed, the process returns to step S37, and the decoding process for the next VLC code string is performed again using the same conversion table.
[0091]
When the code conversion process for the number of times of decoding has been completed, subsequently, it is determined whether the code conversion process has been completed for all the instruction data in the instruction set. If the code conversion has not been completed for all the instruction data in the instruction set, the process returns to step S34, reads the next instruction data from the instruction register 24, and continues the processing. If the decoding has been completed, the present decryption process ends.
[0092]
Next, an example of a control processing method of the code conversion unit 11 by the CPU 16 will be described. Here, a decoding process corresponding to data partitioning defined by MPEG4 will be described.
[0093]
In MPEG4, a transmission method called data partitioning in which the transmission order of information is changed so that important information is transmitted first can be applied instead of a simple hierarchical structure of data included in a macroblock layer. In data partitioning, error resilience can be improved by transmitting important information first. In the data partitioning of MPEG4, the transmission order is changed so that particularly important information of each macroblock of a video packet is placed in the first half of the video packet, and other information is placed in the second half. Then, a specific bit string (dc marker in the case of an intra VOP, motion marker in the case of an inter VOP) is arranged between the first half and the second half of the data string.
[0094]
In the case of an intra VOP, the information inserted in the first half is an encoding mode, a quantization difference value, and an intra DC component, and the information inserted in the second half is further divided into two loops. I have. The first loop in the latter half is a coding pattern and an AC prediction flag necessary for decoding the AC component of the macroblock, and the second loop is the AC component of the macroblock. That is, the data of the macroblock layer in the video packet is transferred in a total of three loops.
[0095]
In the case of an inter VOP, the information inserted in the first half is a flag indicating whether it is a coded MB or an uncoded MB, a coding mode, and a motion vector. , And the inside is divided into two loops. The first loop of the latter half is a coding pattern, an AC prediction flag, a quantization difference value, and an intra DC component, and the second loop is an AC component of a macroblock. That is, the data of the macroblock layer in the video packet is transferred in a total of three loops.
[0096]
Therefore, when performing a decoding process corresponding to such data partitioning, a plurality of instruction sets are prepared for each loop for one video packet, and these instruction sets are set according to the position of the loop of the stream. It is necessary to write to the instruction register 24 while switching appropriately.
[0097]
The control flow described below is an example of a program for performing a decoding process corresponding to such data partitioning.
[0098]
FIG. 13 shows a control flow of data partitioning for an intra VOP, and FIGS. 14 to 16 show instruction sets used in a control flow for an intra VOP.
[0099]
First, the CPU 16 initializes a variable m indicating the number of macroblocks included in one video packet to 0 (step S51).
[0100]
Subsequently, the CPU 16 writes the control data into the control register 25 (Step S52).
[0101]
Subsequently, the CPU 16 writes the first instruction set into the instruction register 24 (Step S53). FIG. 14 shows the first instruction set. This first instruction set is an instruction set for decoding an encoding mode, a quantization difference value, and an intra DC component in one macroblock.
[0102]
Subsequently, the CPU 16 issues a decoding start command to the code conversion unit 11 (step S54). When the decoding start instruction is issued, the code conversion unit 11 performs a decoding process corresponding to the first instruction set (Step S55).
[0103]
Subsequently, when the decoding process corresponding to the first instruction set by the code conversion unit 11 ends, that is, when decoding of the encoding mode, quantization difference value, and intra DC component for one macroblock ends, the CPU 16 The variable m is incremented by 1 (step S56).
[0104]
Subsequently, the CPU 16 determines whether the next data in the data stream is “dc marker” (Step S57). That is, it is determined whether or not all data in the first half of data partitioning has been decoded. If it is not "dc marker", the process returns to step S54, issues a decoding start instruction again, and repeats the processing. Note that since the first instruction set is still stored in the instruction register 24, when the decoding start instruction is given again, the code conversion unit 11 performs the decoding process corresponding to the first instruction set again. repeat. That is, the encoding mode, the quantization difference value, and the intra DC component for the next macroblock are decoded.
[0105]
If the next data in the data stream is “dc marker”, that is, if all the data in the first half of the data partitioning is decoded, then the process proceeds to step S58, where the first loop of the second half of the data partitioning is performed. Is decrypted. At this time, the value of the number of macroblocks in the video packet is stored in the variable m.
[0106]
In step S58, the CPU 16 writes the second instruction set in the instruction register 24 (step S58). FIG. 15 shows the second instruction set. This second instruction set is an instruction set for decoding an encoding pattern and an AC prediction flag necessary for decoding an AC component in one macroblock.
[0107]
Subsequently, the CPU 16 initializes a variable i to 0 (Step S59). This variable i is a value indicating the number of decoded macro blocks.
[0108]
Subsequently, the CPU 16 issues a decoding start command to the code conversion unit 11 (Step S60). When the decoding start instruction is issued, the code conversion unit 11 performs a decoding process corresponding to the second instruction set (Step S61).
[0109]
Subsequently, when the decoding process corresponding to the second instruction set by the code conversion unit 11 ends, that is, when decoding of the coding pattern and the AC prediction flag for one macroblock ends, the CPU 16 increments the variable i by one. (Step S62).
[0110]
Subsequently, the CPU 16 determines whether or not the variable i matches m (step S63). That is, it is determined whether or not all the data in the first loop of the latter half of data partitioning has been decoded. If the variables i and m do not match, the process returns to step S60, issues a decoding start instruction again, and repeats the processing. Note that since the second instruction set is still stored in the instruction register 24, when the decoding start instruction is given again, the code conversion unit 11 performs another decoding process corresponding to the second instruction set. repeat. That is, the decoding of the coding pattern and the AC prediction flag necessary for decoding the AC component for the next macroblock is performed.
[0111]
If the variables i and m match, that is, if all the data in the first loop of the second half of data partitioning is decoded, then the process proceeds to step S64, where decoding of the second loop of the second half of data partitioning is performed thereafter. Processing is performed.
[0112]
In step S64, the CPU 16 writes the third instruction set in the instruction register 24 (step S64). FIG. 16 shows the third instruction set. This third instruction set is an instruction set for decoding an AC component in one macroblock.
[0113]
Subsequently, the CPU 16 initializes a variable i to 0 (Step S65). This variable i is a value indicating the number of decoded macro blocks.
[0114]
Subsequently, the CPU 16 issues a decoding start instruction to the code conversion unit 11 (Step S66). When the decoding start instruction is issued, the code conversion unit 11 performs a decoding process corresponding to the third instruction set (Step S67).
[0115]
Subsequently, when the decoding process corresponding to the third instruction set by the code conversion unit 11 ends, that is, when decoding of the AC component in one macroblock ends, the CPU 16 increments the variable i by one (step S68). ).
[0116]
Subsequently, the CPU 16 determines whether or not the variable i matches m (step S69). That is, it is determined whether or not all data in the second loop of the latter half of data partitioning has been decoded. If the variables i and m do not match, the process returns to step S66, issues a decoding start instruction again, and repeats the processing. Since the third instruction set is still stored in the instruction register 24, when the decoding start instruction is given again, the code conversion unit 11 performs the decoding process corresponding to the third instruction set again. repeat. That is, decoding of the AC component for the next macroblock is performed.
[0117]
When the variables i and m match, that is, when all the data of the second loop in the second half of the data partitioning is decoded, the present processing is terminated.
[0118]
FIG. 17 shows a control flow of data partitioning for an inter VOP. The control flow for this inter VOP is the same as that for the intra VOP except for the difference in whether the processing in step S57 is “dc marker” or “motion marker”. Is omitted.
[0119]
However, the first instruction set, the second instruction set, and the third instruction set are different from each other. FIG. 18 shows a first instruction set used in the inter-VOP control flow, FIG. 19 shows a second instruction set used in the inter-VOP control flow, and FIG. 20 shows a second instruction set used in the inter-VOP control flow. 1 shows an instruction set.
[0120]
The first instruction set (FIG. 18) used in the control flow of the inter-VOP includes a flag indicating whether it is a coded MB or an uncoded MB, which is inserted in the first half of data partitioning, This is an instruction set for decoding a mode and a motion vector. The second instruction set (FIG. 19) used in the control flow of the inter VOP includes the coding pattern, the AC prediction flag, the quantization difference value, and the intra DC inserted in the first loop of the latter half of the data partitioning. Instruction set for decoding components. A third instruction set (FIG. 20) used in the control flow of the inter VOP is an instruction set for decoding an AC component of a macroblock.
[0121]
【The invention's effect】
As described above, in the code conversion device according to the present invention, one conversion table is selected based on the instruction data stored in the instruction storage means, and the code from the syntax to the code string is referred to with reference to the selected conversion table. A conversion process or a conversion process from a code string to syntax is performed.
[0122]
For this reason, in the transcoder according to the present invention, even when a plurality of standards are supported, transcoding can be performed by one circuit without separately creating hardware for each standard. Further, even when a new coding standard is proposed, it is possible to cope with it by adding a conversion table and changing a software program that issues instruction data.
[Brief description of the drawings]
FIG. 1 is a block diagram of a VLC code conversion device according to an embodiment of the present invention.
FIG. 2 is a block diagram of an encoding / decoding unit of the VLC code conversion device.
FIG. 3 is a block diagram of a table conversion unit of the encoding / decoding unit.
FIG. 4 is a diagram illustrating an offset table.
FIG. 5 is a diagram for explaining a data configuration of instruction data.
FIG. 6 is a diagram illustrating a code example of mode information of instruction data in a case where encoding processing is performed according to the MPEG2 method.
FIG. 7 is a diagram illustrating a code example of mode information of instruction data in a case where encoding processing is performed in the MPEG4 system.
FIG. 8 is a diagram illustrating a code example of mode information of instruction data in a case where a decoding process is performed by the MPEG2 method.
FIG. 9 is a diagram illustrating a code example of mode information of instruction data when performing a decoding process according to the MPEG4 system.
FIG. 10 is a diagram for explaining a data configuration of control data.
FIG. 11 is a flowchart illustrating processing performed by a code conversion unit during encoding.
FIG. 12 is a flowchart showing processing contents of a code conversion unit at the time of decoding.
FIG. 13 is a flowchart showing a processing program for decoding data of a macroblock layer from an intra VOP transmitted according to data partitioning.
FIG. 14 is a first instruction set for decoding data in one macroblock from an intra VOP transmitted according to data partitioning.
FIG. 15 is a second instruction set for decoding data in one macroblock from an intra VOP transmitted according to data partitioning.
FIG. 16 is a third instruction set for decoding data in one macroblock from an intra VOP transmitted according to data partitioning.
FIG. 17 is a flowchart showing a processing program for decoding data of a macroblock layer from an inter VOP transmitted according to data partitioning.
FIG. 18 is a first instruction set for decoding data in one macroblock from an inter VOP transmitted according to data partitioning.
FIG. 19 is a second instruction set for decoding data in one macroblock from an inter VOP transmitted according to data partitioning.
FIG. 20 is a third instruction set for decoding data in one macroblock from an inter VOP transmitted according to data partitioning.
[Explanation of symbols]
1 VLC code converter, 11 code converter, 12 DCT coefficient buffer, 13 MV buffer, 14 parameter buffer, 15 bit stream buffer, 21 run level analyzer, 22 MV analyzer, 23 encoder / decoder, 24 instruction register , 25 control register, 26 control unit

Claims (14)

Conversion table storage means for storing a plurality of conversion tables in which a correspondence relationship between syntax indicating a specific information content and a code string is described for each type of the information content,
An encoding process of performing code conversion from the input syntax to a corresponding code string according to the description of the conversion table, synthesizing a plurality of code strings generated by code conversion to generate one data stream, and / or Encoding / decoding means for extracting a code string from the input data stream, and performing decoding processing for performing code conversion from the extracted code string to a corresponding syntax according to the description of the conversion table;
Instruction storage means for storing at least instruction data indicated by the selection information of the conversion table,
The encoding / decoding means selects one of the conversion tables indicated by the instruction data stored in the instruction storage means from the conversion table storage means, and performs the code conversion based on the selected conversion table. A transcoding device characterized by the above-mentioned. The command storage means stores one or more command data from the outside,
The encoding / decoding means sequentially reads out each instruction data stored in the instruction storage means, and executes a code conversion by selecting a conversion table indicated in the instruction data every time the instruction data is read out. The transcoder according to claim 1, wherein: In the instruction data, the selection information of the conversion table, the data length of the code string, and the number of processes for performing continuous code conversion are indicated.
3. The apparatus according to claim 2, wherein said encoding / decoding means performs an encoding process in accordance with the contents indicated in said instruction data. 4. The apparatus according to claim 3, wherein said encoding / decoding means extracts a code string having a predetermined data length from an input data stream based on a data length of the code string. 4. The code conversion apparatus according to claim 3, wherein the coding / decoding means performs the code conversion by using the selected conversion tables for the number of processes for continuously performing the code conversion. Further comprising data control means for generating the instruction data,
The data control means generates an instruction set including a plurality of instruction data according to a format of a data stream to be input / output, and stores the generated instruction set in the instruction storage means. Item 3. The transcoder according to Item 2. The conversion table storage means stores a conversion table for encoding in which a code string is indicated for the syntax, and a conversion table for decoding in which syntax is indicated for the code string. The transcoder according to claim 1, wherein: The information described in the conversion table is divided for each predetermined information group,
The syntax or code string before the code conversion is input, and the syntax or the code string before the code conversion is converted into an upper address specifying each information group in the conversion table, and the order of each information in the information group. Address generating means for converting the data into a lower address to be indicated,
An offset table describing the address of the leading information in the information group specified by the upper address in association with the upper address,
The encoding / decoding means includes:
A conversion table and an offset table indicated in the instruction data stored in the instruction storage means are selected one by one from the conversion table storage means and the offset table;
The conversion from the upper address output from the address generator to the address of the head information in the information group is performed according to the description of the selected offset table,
Extracting information from the selected conversion table and performing code conversion based on the address of the leading information in the information group output from the offset table and the lower address output from the address generation means. The code conversion device according to claim 1, wherein Variable-length coding and fixed-length coding processing and / or variable-length code decoding and fixing for a data stream that includes at least moving image data coded using the discrete cosine transform processing and the motion prediction processing In a code conversion device that performs decoding processing of a long code,
Conversion table storage means for storing a plurality of conversion tables in which a correspondence relationship between syntax indicating a specific information content and a code string is described for each type of the information content,
A DCT coefficient buffer that stores a DCT coefficient that is a syntax generated by performing the discrete cosine transform process on the moving image data;
A motion vector buffer that stores a motion vector that is a syntax generated by performing a motion prediction process on the moving image data,
A parameter buffer for storing a parameter that is a syntax other than the DCT coefficient and the motion vector,
The syntax stored in the DCT buffer, the motion vector buffer, and the parameter buffer is read, and code conversion from the read syntax to a corresponding code string is performed according to the description in the conversion table, and code conversion is performed. Encoding processing for generating a single data stream by combining the plurality of code strings thus obtained, and / or extracting a code string from the input data stream and performing code conversion from the extracted code string to a corresponding syntax. Encoding / decoding means for performing decoding according to the description of the conversion table and storing syntax in the DCT buffer, the motion vector buffer, and the parameter buffer, respectively;
Command storage means for storing at least command data indicating the type of the information content,
The encoding / decoding means selects one conversion table from a plurality of conversion tables stored in the conversion table storage means based on a type of information content indicated in the instruction data stored in the instruction storage means. And a code conversion device for performing the code conversion based on the selected conversion table. The conversion table storage means stores a conversion table for encoding in which a code string is indicated for the syntax, and a conversion table for decoding in which syntax is indicated for the code string. 10. The transcoder according to claim 9, wherein: The command storage means stores one or more command data from the outside,
The encoding / decoding means sequentially reads out each instruction data stored in the instruction storage means, and executes a code conversion by selecting a conversion table indicated in the instruction data every time the instruction data is read out. The transcoder according to claim 10, wherein: The instruction data given at the time of encoding includes type information of information content, data length of a code string, selection information of a parameter buffer, a DCT coefficient buffer or a motion vector buffer, and information indicating a value to be fixed-length encoded. , And the number of times of continuous code conversion are shown,
12. The transcoder according to claim 11, wherein said encoding / decoding means performs an encoding process according to the contents indicated in said instruction data. In the instruction data given at the time of decoding, the type information of the information content, the data length of the code string, and the number of times of continuous code conversion are shown,
13. The apparatus according to claim 12, wherein said encoding / decoding means performs an encoding process in accordance with the contents indicated in said instruction data. Control data corresponding to each type information of information content indicated in the command data stored in the command data storage means includes control data storage means stored from the outside,
The control data includes termination processing information indicating that the processing based on the instruction data is processing for the end of the stream, and whether the processing based on the instruction data performs processing on fixed-length codes or processing on variable-length codes. Fixed / variable-length code processing identification information indicating whether or not the processing based on the instruction data performs processing on an intra macroblock or an inter macroblock, and the instruction data And buffer selection information that specifies a buffer in which to store the syntax decoded based on
12. The transcoder according to claim 11, wherein said encoding / decoding means performs an encoding process according to the contents indicated in said instruction data.

Images