JP3993213B2 - Image decoding device - Google Patents

Description

  The present invention relates to an image decoding apparatus that handles encoded bit streams of different encoding systems.

  Currently, the MPEG-4 (Moving Picture Experts Group Phase-4) system, which is being standardized by ISO / IEC JTC11 / SC29 / WG11, and the ITU-T recommendation H.264. The header information (information signal for decoding) added to the encoded bit stream as the transmission signal is different from the system based on the H.263 standard.

  FIG. H.263 based on the H.263 standard. 2 is a diagram illustrating the structure of a H.263 encoded bitstream 201, in which header information 211 and H.264 are encoded. Macroblock data 225 which is image encoded data encoded by the H.263 encoding method is multiplexed. FIG. 1B is a diagram showing the structure of an MPEG-4 encoded bitstream 202, in which header information 212 and macroblock data 239 that is image encoded data encoded by the MPEG-4 encoding method are multiplexed. Has been. As shown in the figure, the structure of the encoded bit stream is different. H.263 does not include header information related to VO (Video Object), VOL (Video Object Layer), VOP (Video Object Plane), and the like necessary for MPEG-4 decoding. For this reason, separate image decoding devices and image encoding devices are required to perform image communication using both systems.

  H. The GOB start code 223, the GOB header information 224 of the H.263 encoded bit stream 201, the resynchronization instruction code 237 of the MPEG-4 encoded bit stream 202, and the re-synchronization information 238 are not always inserted. Inserted.

Thus, since the conventional encoded bit stream is configured as described above, for example, in an MPEG-4 compatible image decoding apparatus, H.264 is used. H.263 generated based on the H.263 standard. There is a problem that the H.263 encoded bitstream 201 cannot be decoded.
MPEG-4 compatible and H.264. In order to decode both encoded bit streams based on the H.263 standard, it is necessary for the image decoding apparatus to include both types of decoding circuits, and there is a problem that the apparatus becomes complicated.

  The present invention has been made to solve the above-described problems. An object of the present invention is to obtain an image decoding apparatus capable of decoding the H.263 encoded bit stream 201.

The image decoding apparatus according to the present invention is at least H.264. H.263 encoding method header information and H.264 encoding method. A first encoded bit stream obtained by multiplexing image encoded data encoded by the H.263 encoding method, or an image encoded by the header information of the MPEG-4 encoding method and the MPEG-4 encoding method In the decoding of the second encoded bit stream multiplexed with the encoded data, reception is performed based on the header information of the first encoded bit stream or the header information of the second encoded bit stream. Encoding method determining means for determining whether the encoded bit stream is the first encoded bit stream or the second encoded bit stream, and the encoding method determining means includes the first encoding When it is determined that the bit stream has been received, H.263 encoding method header information included in the H.263 encoding method header information. H.264 format image encoding information defined based on H.263 format image encoding information is used. When the H.263 encoded bit stream is decoded in accordance with the syntax of the MPEG-4 encoded bit stream, and the encoding scheme determination means determines that the second encoded bit stream has been received, the received second The encoded image data included in the second encoded bit stream is converted to the encoded image data included in the second encoded bit stream using the MPEG-4 encoded image encoded information included in the header information of the encoded MPEG-4 encoding method. In the decoding process of decoding, the quantization scale used for the inverse quantization process of the DC component of the intra DCT coefficient is a fixed value when the determination unit determines that the first encoded bitstream has been received. set, if it is determined that it has received the second coded bit stream quantized determined for each macroblock Those having setting means for setting a value corresponding to the value of the step size, and an inverse quantization processing unit for performing inverse quantization process of the DC component of the intra DCT coefficient using a quantization scale determined by the setting means is there.

  According to the present invention, H.P. Even with H.263 and MPEG-4 encoded bitstreams, the encoding scheme can be easily identified and decoded.

Hereinafter, in order to describe the present invention in more detail, the best mode for carrying out the present invention will be described with reference to the accompanying drawings.
Embodiment 1 FIG.
2 is a diagram showing the structure of an encoded bit stream received by the image decoding apparatus according to Embodiment 1, and FIG. The H.263 encoded bit stream 203 and FIG. 2B are the MPEG-4 encoded bit stream 204. MPEG-4 compatible H.264 in FIG. The H.263 encoded bit stream 203 is the same as the H.263 encoded bit stream 203 shown in FIG. 263 encoded bit stream 201, VO start code 231, VO identification number 232, VOL start code 233, and H.264 code. H.263 compatible identification information 226 is added. Also, the MPEG-4 encoded bit stream 204 in FIG. 2B is added to the conventional MPEG-4 encoded bit stream 202 shown in FIG. H.263 compatible identification information 226 is added. MPEG-4 compatible H.264 H.263 encoded bit stream 203 and MPEG-4 encoded bit stream 204 have been added. The H.263 compatible identification information 226 is information that can be distinguished from each other. If the H.263 compatible identification information is a “0” bit, the other is a “1” bit.

  FIG. 3 is a block diagram showing a configuration of an image decoding apparatus for decoding a VO (Video Object) according to the first embodiment. In the figure, 1 is a received encoded bit stream, 2 is a syntax analysis / variable length decoding unit, which analyzes the syntax (multiplexed video signal) from the encoded bit stream 1 and also generates shape encoded data. 3. Output texture encoded data 6 and texture motion data 7. 4 is a shape decoding unit that decodes the shape encoded data 3 to obtain decoded shape data 5, 8 is a motion compensation unit that performs motion compensation based on the texture motion data 7 and obtains predicted texture data 9, and 10 is a texture encoded data 6 This is a texture decoding unit that performs decoding based on the predicted texture data 9 and obtains decoded texture data 11.

Next, the operation will be described.
Here, the MPEG-4 compatible H.264 of FIG. The decoding operation of the H.263 encoded bit stream 203 will be mainly described. Accordingly, a case will be described in which the shape of each VOP is rectangular, that is, in the case where shape encoded data is not included in the bitstream, and texture data and information on motion are encoded for each macroblock.
Note that the decoding operation of the MPEG-4 encoded bit stream 204 in FIG. 2B is basically the same as the conventional one.

  First, the syntax analysis / variable length decoding unit 2 separates the input encoded bitstream 1 from the binary bitstream into meaningful data. This syntax analysis / variable length decoding unit 2 is compatible with MPEG-4 compatible H.264. The H.263 encoded bit stream 203 can be decoded. The motion compensation unit 8 performs motion compensation based on the texture motion data 7 output from the syntax analysis / variable length decoding unit 2 and outputs predicted texture data 9. The texture decoding unit 10 receives the texture encoded data 6 output from the syntax analysis / variable length decoding unit 2 and the predicted texture data 9 output from the motion compensation unit 8 to obtain decoded texture data 11.

Next, the operation of the syntax analysis / variable length decoding unit 2 will be described.
FIG. 4 is a block diagram showing the configuration of the syntax analysis / variable length decoding unit 2. In the figure, 21 is a header information analysis unit that extracts header information added to the encoded bitstream 1 and sets various header information necessary for subsequent decoding control, and 22 is a texture encoding from the encoded bitstream 1. It is a macroblock layer syntax analysis unit for obtaining data 6 and texture motion data 7.

  FIG. 5 is a block diagram showing the configuration of the header information analysis unit 21. In the figure, 30 is a VO start code detector for detecting the VO start code 231 in the encoded bit stream 1, 31 is a VOL start code detector for detecting the VOL start code 233 from the encoded bit stream 1, and 32 is the encoding method. A determination unit, and the encoded bitstream 1 is MPEG-4 compatible H.264. H.263 encoded bit stream 203 or MPEG-4 encoded bit stream 204, The H.263 compatible identification information 33 is output. 34 is a switching unit that is switched according to the determined encoding method, and 35 is an MPEG-4 compatible H.264. In the case of the H.263 coded bit stream 203, the picture header information 222, which is image coding information specific to the H263 system, is decoded, and the VOL header information 234 and VOP header information 236, which are image coding information specific to the MPEG-4 system. Set H. H.263 picture header information analysis unit 36 is MPEG-4 compatible H.264. In the case of the H.263 encoded bit stream 203, H.264 is used. 263 GOB (Group of Block) header information 224 is decoded, and H.264 GOB (Group of Block) header information 224 is decoded based on the decoded GOB header information 224. H.263 which changes the VOP header information 236 set by the H.263 picture header analysis unit 35. 263 GOB header information analysis unit, 37 is a VOL header information decoding unit for decoding VOL header information 234 in the case of the MPEG-4 encoded bit stream 204, and 38 is a VOP header information 236 in the case of the MPEG-4 encoded bit stream 204. It is a VOP header information analysis part to decode.

Next, the operation of the header information analysis unit 21 will be described.
The VO start code detection unit 30 is compatible with the MPEG-4 compatible H.264 shown in FIG. When the VO start code 231 in the H.263 encoded bit stream 203 or the MPEG-4 encoded bit stream 204 is detected, the following decoding operation is started. The VOL start code detector 31 detects the VOL start code 233 from the encoded bit stream 1. The encoding method determination unit 32 then converts the encoded bitstream 1 to H.264. H.263 compatible identification information 226 is decoded, According to the H.263 compatible identification information 226, the encoded bit stream 1 is MPEG-4 compatible H.264. H.263 encoded bit stream 203 or MPEG-4 encoded bit stream 204 is determined. The H.263 compatible identification information 33 is output.
The encoded bitstream 1 is MPEG-4 compatible H.264. In the case of the H.263 encoded bit stream 203, the switching unit 34 changes the encoded bit stream 1 to H.264. It is input to the H.263 picture header information analysis unit 35.

  FIG. 3 is a block diagram illustrating a configuration of a H.263 picture header information analysis unit 35. FIG. H. When the H.263 picture start code detection unit 41 detects the picture start code 221 from the encoded bitstream 1, the H.263 picture start code detection unit 41 next detects the picture start code 221. The H.263 picture header information decoding unit 42 decodes the picture header information 222 from the encoded bitstream 1. Then, the MPEG-4 header information setting unit 43 sets VOL header information 234 and VOP header information 236 based on the decoded picture header information 222.

  FIG. It is a block diagram which shows the structure of the H.263 picture header information decoding part. The temporal reference (TR) decoding unit 51 is an H.264 standard. The bitstream 1 from the H.263 picture start code detection unit 41 is input, and the number of pictures (TR) that are not skipped or referred to between transmitted pictures is decoded. This information is used as necessary when displaying.

  Next, the picture type (PTYPE) decoding unit 52 decodes the picture type (PTYPE). The picture type includes information such as a picture format 301, a picture encoding type 302, and an optional mode instruction flag 303. The decoded picture format 301 and picture encoding type 302 are output to the MPEG-4 header information setting unit 43 in FIG.

  Also, the picture type (PTYPE) decoding unit 52 determines whether or not the optional mode instruction flag 303 is ON. H. In H.263, some optional modes are prepared. However, in the image decoding apparatus described in this embodiment, it is assumed that compatibility is not allowed for a bitstream including these optional modes. The encoded bit stream for which the optional mode is ON (valid) is input to the decoding operation end unit 54 by the switching unit 53. Then, the decoding operation end unit 54 ends the decoding operation for the encoded bit stream. The picture type includes other information that defines the display, etc., which can be used as necessary.

  On the other hand, the bit stream for which the optional mode is OFF (invalid) is input to the picture quantization step size (PQUANT) decoding unit 55 by the switching unit 53. The picture quantization step size (PQUANT) decoding unit 55 decodes the picture quantization step size (PQUANT) 304. The decoded picture quantization step size 304 is output to the MPEG-4 header information setting unit 43 in FIG. The picture header information after the picture quantization step size 304 is skipped because it is not necessary for the subsequent decoding.

Next, the operation of the MPEG-4 header information setting unit 43 in FIG. 6 will be described.
The MPEG-4 header information setting unit 43 sets VOL shape information and object size as VOL header information 234 based on the decoded picture header information 222. MPEG-4 compatible H.264 In the case of an H.263 encoded bit stream, each bit stream corresponds to a frame, and the MPEG-4 header information setting unit 43 sets that the shape information is rectangular. Furthermore, since the object size corresponds to the frame size, the MPEG-4 header information setting unit 43 obtains the frame size from the picture format 301 which is one of the picture header information 222, and sets it to the object size. Also, it is set whether or not the gradation per pixel is 8 bits. H. In H.263, since the gradation per pixel is not assumed to be other than 8 bits, the gradation per pixel is set to 8 bits.

  Next, the MPEG-4 header information setting unit 43 invalidates sprite coding, error resilience coding, intra AC / DC prediction, and scalability coding, which are MPEG-4 based coding conditions. In MPEG-4, the quantization method is H.264. H.263, MPEG-1 / 2 can be selected from two methods, so that MPEG-4 compatible H.264 can be selected. In the case of the H.263 encoded bit stream 203, the quantization method is H.264. Set to H.263.

  Further, the MPEG-4 header information setting unit 43 sets VOP header information 236. As VOP header information 236, VOP prediction type information and quantization step size are set. There are two types of VOP prediction types: intra, which is encoded using only data in the VOP, and inter, which is encoded using data of preceding and succeeding VOPs. The VOP prediction type information is set based on the picture coding type 302 that is one of the picture header information 222. Similarly, the VOP quantization step size is set from the picture quantization step size 304 which is the picture header information 222.

  Further, in MPEG-4, since a motion vector search range can be selected from seven types, a code for specifying a motion vector search range is prepared. On the other hand, H.H. Since H.263 supports only one type of search range, the MPEG-4 header information setting unit 43 sets the motion vector search range designation code to H.264. It is necessary to set a code corresponding to the motion vector search range used in H.263. MPEG-4 is compatible with interlaced images. Since the H.263 does not support interlace, the interlace mode instruction information is always set to invalid.

  In FIG. When the analysis of the picture header information by the H.263 picture header information analysis unit 35 is completed, when the GOB start code 223 and the GOB header information 224 are inserted in the encoded bit stream, the H.264 The GOB header information analysis unit 36 analyzes the GOB header information 224. When the GOB start code 223 and the GOB header information 224 are not inserted in the encoded bit stream, The H.263 GOB header information analysis unit 36 does not operate.

  FIG. 8 shows the H.264 in FIG. 3 is a block diagram illustrating a configuration of a H.263 GOB header information analysis unit 36. FIG. When the GOB start code detection unit 61 detects the GOB start code 223 added to the encoded bitstream 1, the GOB header information decoding unit 62 decodes the GOB header information 224.

  FIG. 9 is a diagram for explaining GOB. As shown in the figure, the GOB is an image divided into several macroblock columns, and the GOB header information 224 includes information necessary for resynchronization on the decoding side. When an error occurs in a certain bit of the encoded bit stream, when variable-length encoding and predictive encoding are performed, the subsequent macroblock data cannot be correctly decoded and the error propagates. When GOB header information is detected, the encoded bitstream is resynchronized before decoding the first macroblock of the GOB, and information necessary for decoding the subsequent macroblocks is reset. Can prevent error propagation. The quantization step size and motion vector of each macroblock are re-synchronized because the prediction encoding is performed to encode the difference from the quantization step size and motion vector of the already encoded macroblock. If this happens, it is necessary to reset these information.

  FIG. 10 is a block diagram showing the configuration of the GOB header information decoding unit 62. The GOB number decoding unit 71 decodes the GOB number (GN) from the encoded bit stream 1. Then, the GOB frame identification number decoding unit 72 decodes the identification number (GFID) of the picture to which the GOB belongs. Further, the GOB quantization step size decoding unit 73 decodes the GOB quantization step size (GQUANT) 305 and outputs it to the MPEG-4 header information changing unit 63 in FIG.

  The MPEG-4 header information changing unit 63 changes the VOP header information 236 set by the MPEG-4 header information setting unit 43 based on the decoded GOB header information 224. The information changed by the GOB header information 224 is the quantization step size, and the GOB quantization step size is set to the VOP quantization step size. Each set information is output to the macroblock layer syntax analysis unit 22 of FIG.

  The encoding method determination unit 32 in FIG. If the H.263 compatible identification information 226 indicates MPEG-4, it is determined that the encoded bitstream 1 is the MPEG-4 encoded bitstream 204, and The H.263 compatible identification information 33 is output. The MPEG-4 encoded bit stream 204 is input to the VOL header information decoding unit 37 by the switching unit 34. The VOL header information decoding unit 37 decodes the VOL header information 234 from the encoded bit stream, and the VOP header information analysis unit 38 decodes the VOP header information 236 and outputs it to the macroblock layer syntax analysis unit 22 of FIG.

  When the above information is set, the macroblock layer syntax analysis unit 22 decodes the macroblock data 225 and 239 by analysis based on the MPEG-4 syntax. However, the encoding method of block data is MPEG-4 and H.264. Since it is slightly different in H.263, it is necessary to switch the operation on the decoding side.

  FIG. 11 shows MPEG-4 compatible H.264 in the first embodiment. It is a figure which shows the layer structure of the macroblock data 225 in the 263 bit stream 203. FIG. The macro block is composed of four luminance blocks and two color difference blocks. As shown in the figure, for each macroblock, the attribute information of macroblock skip determination information 251, macroblock type / effective color difference block identification information 252, effective block identification information 253, differential quantization step size 254, and motion data 255 is included. Multiplexed.

  Here, the macroblock skip determination information 251 indicates that the inter-VOP has zero motion vector and all the coefficient data in the macroblock (input image signal (original signal in the case of intra, and reference VOP in the case of inter). This is information indicating whether or not the data quantized after the DCT of the difference signal) is zero. If the motion vector is zero and the coefficient data is all zero, the subsequent information regarding the macroblock is included in the bitstream. I can't. Therefore, it is skipped to the next macroblock.

  The macroblock type of the macroblock type / effective color-difference block identification information 252 is encoded when the macroblock data is encoded using the original signal of the macroblock (intra) or motion compensation prediction is performed. Information indicating the coding type of a macroblock, such as when encoding a differential signal with a reference macroblock (inter), coding with a quantization step size different from the quantization step size of the previous macroblock, etc. is there.

Furthermore, the effective block identification information 253 is information indicating whether the coefficient data of each block is all zero. Although coefficient data is multiplexed for each block after the attribute information (corresponding to block data 256), if this valid block identification information 253 indicates an invalid block, coefficient data for that block exists. do not do.
The differential quantization step size 254 is information that is multiplexed when the macroblock type indicates that it is different from the quantization step size of the previous macroblock, and is the quantum of the previous macroblock. The difference value from the conversion step size is shown.

  FIG. 12 is a block diagram showing a configuration of the macroblock layer syntax analysis unit 22. In the figure, 81 is a switching unit switched by the shape information 311 set by the MPEG-4 header information setting unit 43, 82 is a shape-encoded data decoding unit for decoding shape-encoded data in the encoded bitstream, and 83 is A switching unit that is switched by the VOP prediction type 312 set in the MPEG-4 header information setting unit 43; and 84, a skip determination information decoding unit that decodes the macro block skip determination information 251 when the VOP prediction type is other than intra. 85 is a switching unit that is switched based on the skip determination information 251, 86 is a skip data setting unit that sets all motion vectors and texture data of the macroblock to zero when skipping, and 87 is a case where the VOP prediction type 312 is intra or skipping. If not, macroblock type 313 and A macroblock type valid color difference block identification information decoder for decoding the valid color difference block identification information.

  88 is a switching unit that is switched by the intra AC / DC prediction mode instruction information 315 set by the MPEG-4 header information setting unit 43, 89 is an AC prediction instruction information decoding unit that decodes AC prediction instruction information, and 90 is an effective block. An effective block identification information decoding unit 91 that decodes the identification information 253 is a switching unit that is switched by the macroblock type 313 output from the macroblock type / effective color difference block identification information decoding unit 87.

  Further, 92 is a difference quantization step size zero setting unit that sets the difference quantization step size to zero, 93 is a difference quantization step size decoding unit that decodes the difference quantization step size 317, and 94 is a decoded difference quantization step. The addition unit 95 adds the size 317 and the VOP quantization step size 318 of the previous block and outputs the quantization step size 319 to the texture decoding unit 10 in FIG. 3, and 95 indicates an interlace mode from the MPEG-4 header information setting unit 43 Switching unit switched by instruction information 316, 96 is an interlace information decoding unit that decodes interlace information, 97 is a macroblock type 313 and MPEG-4 header output from the macroblock type / effective color difference block identification information decoding unit 87 VOP prediction type 3 output from the information setting unit 43 2 and a motion vector decoding unit that decodes a motion vector (texture motion data 7) based on the motion vector search range designation information 320, and 98 decodes the encoded block data and outputs the texture encoded data 6 to the texture decoding unit 10. It is a block data decoding unit.

Next, the operation of the macroblock layer syntax analysis unit 22 will be described.
Here, the encoded bitstream 1 is MPEG-4 compatible H.264. A case of the H.263 encoded bit stream 203 will be described. The case of the MPEG-4 encoded bit stream 204 is omitted because it is described in ISO / IEC JTC1 / SC29 / WG11 MPEG-4 Video VM8.0.
The output of the encoded bitstream 1 is switched by the switching unit 81 according to the shape information 311 set by the MPEG-4 header information setting unit 43. The encoded bitstream 1 is MPEG-4 compatible H.264. In the case of the H.263 encoded bit stream 203, since the shape information 311 is set to a rectangle, the bit stream 1 is input to the switching unit 83 without passing through the shape encoded data decoding unit 82.

  Next, the switching unit 83 is switched by the VOP prediction type 312 set by the MPEG-4 header information setting unit 43. When the VOP prediction type 312 is intra, the macroblock type / effective color difference block identification information decoding unit 87 decodes the macroblock type 313 and the effective color difference block identification information. If the VOP prediction type is other than intra, the skip determination information decoding unit 84 decodes the macro block skip determination information 251. When the switching unit 85 is switched based on the decoded skip determination information 251 to indicate that the macro block is skipped, the skip time data setting unit 86 sets the motion vector of the macro block and the macro block in the macro block. All the texture data is set to zero and the decoding for the macroblock is terminated. When the skip determination information 251 indicates that the macro block is not skipped, the macro block type / effective color difference block identification information decoding unit 87 decodes the macro block type 313 and the effective color difference block identification information.

  Next, the switching unit 88 is switched by the intra AC / DC prediction mode instruction information 315 set by the MPEG-4 header information setting unit 43. MPEG-4 compatible H.264 In the case of the H.263 encoded bit stream 203, since there is no function of performing intra AC / DC prediction, when setting the VOL header information 234, the intra AC / DC prediction is set to be invalid, and the AC prediction instruction The information is input to the effective block identification information decoding unit 90 without going through the information decoding unit 89.

Next, the effective block identification information decoding unit 90 decodes the effective block identification information 253 for the luminance block in the macroblock. The switching unit 91 is switched by the macroblock type 313 decoded by the macroblock type / effective color-difference block identification information decoding unit 87, and the quantization of the macroblock whose quantization step size of the macroblock is decoded one before is performed. When the difference is different from the step, the differential quantization step size decoding unit 93 decodes the differential quantization step size 317 with the quantization step size of the macroblock encoded immediately before. The decoded difference quantization step size 317 is added to the VOP quantization step size 318 of the previous macroblock by the adder 94 and output to the texture decoding unit 10 in FIG. 3 as the quantization step size 319.
On the other hand, if the quantization step size of the macroblock indicates that it is the same as the quantization step size of the previously decoded macroblock, the difference quantization step size zero setting unit 92 sets the difference quantization step size to Set to zero.

  Next, the switching unit 95 is switched by the interlace mode instruction information 316 from the MPEG-4 header information setting unit 43. MPEG-4 compatible H.264 In the case of the H.263 coded bit stream 203, since the interlaced image is not supported, the interlace mode is set to be invalid, and the motion vector decoding unit 97 does not pass through the interlace information decoding unit 96. Entered. Then, when the VOP prediction type 312 set by the MPEG-4 header information setting unit 43 is inter, the motion vector decoding unit 97 uses the macroblock type / macroblock type decoded by the effective color difference block identification information decoding unit 87. Based on the motion vector search range designation information 320 set in 313 and the MPEG-4 header information setting unit 43, the motion vector (texture motion data 7) is decoded and output to the motion compensation unit 8 in FIG.

  Next, the block data decoding unit 98 decodes the encoded block data in the encoded bit stream. FIG. 13 is a block diagram showing the configuration of the block data decoding unit 98. In the figure, reference numeral 101 denotes a switching unit which inputs encoded block data and is switched by the macroblock type 313 output from the macroblock type / effective color difference block identification information decoding unit 87, and 102 is an MPEG-4 header information setting unit 43. A switching unit that is switched by the set intra AC / DC prediction mode instruction information 315, and 103 is a DC coefficient based on the gradation 321 per pixel from the MPEG-4 header information setting unit 43 when the intra AC / DC prediction is OFF. A DC coefficient fixed-length decoding unit that performs fixed-length decoding and outputs a decoded intra DC coefficient 111, and 104 is a DC coefficient decoding unit that decodes the DC coefficient and outputs the decoded intra DC coefficient 111 when intra AC / DC prediction is ON. is there.

  A switching unit 105 is switched by the effective block identification information 253 output from the effective block identification information decoding unit 90, and a macroblock type 313 output from the macroblock type / effective color difference block identification information decoding unit 87 and the encoding method The H.V. An AC coefficient VLD table switching unit that switches an AC coefficient VLD (Variable Length Decoding) table based on H.263 compatible identification information 33.

  107 is an AC coefficient data variable length decoding unit that performs variable length decoding of AC coefficient data and outputs decoded AC coefficient data 112, and 108 is an H.264 output from the coding scheme determination unit 32. A switching unit switched by the H.263 compatible identification information 33, 109 is an AC coefficient data fixed length decoding unit that outputs the decoded AC coefficient data 112, 110 is an AC coefficient data Esc coding decoding unit that outputs the decoded AC coefficient data 112, and 113 is an AC An AC coefficient zero setting unit for setting a coefficient to zero.

Next, the operation of the block data decoding unit 98 will be described.
First, the coded block data is switched by the switching unit 101 according to the macroblock type 313 output from the macroblock type / effective color difference block identification information decoding unit 87. When the macroblock type 313 is other than intra, the switching unit is switched. 105 is output. When the macroblock type 313 is intra, the encoded block data is input to the switching unit 102 and switched by the intra AC / DC prediction mode instruction information 315 set by the MPEG-4 header information setting unit 43.

  MPEG-4 compatible H.264 In the case of the H.263 encoded bit stream 203, since the intra AC / DC prediction mode 315 is set to invalid, the DC coefficient decoding unit 104 does not pass through and is input to the DC coefficient fixed length decoding unit 103. The DC coefficient fixed length decoding unit 103 performs fixed length decoding, outputs the decoded intra DC coefficient 111 to the texture decoding unit 10, and outputs the encoded block data to the switching unit 105. At this time, the fixed length decoded code length is equal to the gradation per pixel (8 bits of default value) 321 set by the MPEG-4 header information setting unit 43, and the gradation 321 per pixel is the default value. Is set to 8 bits. This is the same as the case of the H.263 compatible decoding apparatus.

The switching unit 105 is switched by the effective block identification information 253 decoded by the effective block identification information decoding unit 90. When the block is an invalid block, the AC coefficient zero setting unit 113 determines the decoded AC coefficient data in the block. 112 is set to zero and output to the texture decoding unit 10. If the block is an effective block, the encoded block data is input to the AC coefficient VLD table switching unit 106.
The AC coefficient in the block is scanned by the encoder in the order determined in the block, a flag (LAST) indicating whether the non-zero coefficient is the last in the block, the number of consecutive zeros (RUN) ) And the subsequent non-zero coefficient level (LEVEL) as a set. On the decoding side, a combination of (LAST, RUN, LEVEL) is obtained by variable length decoding of the encoded data, and the AC coefficient in the block can be reproduced based on this. Note that when variable-length coding is performed on the combination of (LAST, RUN, LEVEL), MPEG-4 performs variable-length coding using different VLC (Variable Length Coding) tables depending on the macroblock type. In H.263, variable length coding is performed using the same VLC table regardless of the macroblock type.

  Therefore, in the image decoding apparatus according to this embodiment, the AC coefficient VLD table switching unit 106 outputs the macroblock type 313 output from the macroblock type / effective color difference block identification information decoding unit 87 and the encoding method determination unit 32. H. The AC coefficient VLD table is switched based on the H.263 compatible identification information 33. And H. H.263 compatible identification information 33 is H.264. In the case where H.263 is set, the AC coefficient data variable length decoding unit 107 performs variable length decoding using one VLD table regardless of the macroblock type (intra, inter) 313, and the decoded AC coefficient data 112 is The encoded texture data 6 is output to the texture decoding unit 10.

  The encoding method when the combination of (LAST, RUN, LEVEL) is not in the VLC table is also MPEG-4 and H.264. It is different in H.263. In MPEG-4, if there is no combination of (LAST, RUN, LEVEL) in the VLC table, after encoding the escape code, the value of RUN or LEVEL is corrected and variable length coding is performed or fixed length Encoding is performed. H. In H.263, after the escape code is encoded, the values of LAST, RUN, and LEVEL are fixed-length encoded.

  Therefore, in the image decoding apparatus according to this embodiment, when an escape code is detected from the AC coefficient encoded data by the AC coefficient data variable length decoding unit 107, the encoded bit stream is input to the switching unit. And H. H.263 compatible identification information 33 is H.264. In the case of H.263, the coded bit stream determines the codes relating to the following LAST, RUN, and LEVEL by the AC coefficient data fixed length decoding unit 109 without going through the AC coefficient data Esc coding decoding unit 110, respectively. The fixed length decoding is performed with the obtained code length, and the decoded AC coefficient data 112 is output to the texture decoding unit 10 as the texture encoded data 6.

  Through the above operation, the texture encoded data 6 and the motion vector (texture motion data 7) output from the macroblock layer syntax analysis unit 22 are distributed to the texture decoding unit 10 and the motion compensation unit 8, respectively.

  As described above, the syntax analysis / variable length decoding unit 2 in FIG. 3 decodes and sets the VOP prediction mode. When the VOP prediction mode is inter, the texture motion vector difference vector is decoded. The difference vector of the decoded texture motion vector is the difference vector between the prediction vector obtained from the motion vectors of the three neighboring macroblocks and the actual motion vector, and thus the prediction vector is added to the difference vector of the motion vector. The vector is calculated as a motion vector (texture motion data 8).

The prediction vector is calculated from the motion vectors of three neighboring macroblocks (MV1, MV2, MV3) that have already been decoded as shown in FIG. However, when any of the three neighboring macroblocks is located outside the VOP, the motion vector values of the macroblocks located outside the VOP are shown in FIGS. 14B and 14D. As shown in FIG. 14C, it is set using the motion vector of the macroblock in the VOP. However, the encoding method is H.264. In the case of H.263, when the GOB header is defined, it is necessary to set the prediction vector within the GOB boundary. The setting of the prediction vector is the same as in the case of VOP. Based on the decoded vector, a prediction vector is extracted as predicted texture data 9 and output to the texture decoding unit 10.
On the other hand, when the VOP prediction mode is intra, motion compensation prediction is not performed.
The texture decoding unit 10 receives the texture encoded data 6 and restores the texture data 11.

FIG. 15 is a block diagram illustrating a configuration of the texture decoding unit 10. The inverse quantization unit 114 inversely quantizes the texture encoded data 6.
FIG. 16 is a block diagram showing the configuration of the inverse quantization unit 114.
The switching unit 117 is switched by the macroblock type 313 included in the texture encoded data 6. When the macroblock type 313 of the block to be decoded is in the inter coding mode, the texture coded data 6 is not included in the texture coded data 6, so the texture coded data 6 is input to the AC coefficient inverse quantization unit 120. . When the macroblock type 313 of the decoding target block is the intra coding mode, the texture coded data 6 is input to the switching unit 118.

  The switching unit 118 is H.264. It is switched by the H.263 compatible identification information 33. H. 263 compatible identification information 33 is MPEG-4 compatible H.264. When the H.263 encoded bit stream 203 is indicated, the DC coefficient linear inverse quantization unit 119B performs inverse quantization on the DC coefficient data included in the texture encoded data 6. On the other hand, H. When the H.263 compatible identification information 33 indicates the MPEG-4 encoded bit stream 204, the DC coefficient nonlinear inverse quantization unit 119A performs inverse quantization on the DC coefficient data and outputs a DC coefficient 306. The quantization of the DC coefficient is performed by dividing the DC coefficient by a predetermined value (called a quantization scale) and rounding down the fraction. Therefore, on the decoding side, the DC coefficient 306 can be restored by multiplying the quantized DC coefficient by the quantization scale. The setting of the quantization scale value is different between the DC coefficient linear inverse quantization unit 119B and the DC coefficient nonlinear inverse quantization unit 119A. In the DC coefficient linear inverse quantization unit 119B, the quantization scale is inversely quantized with a fixed value of 8. On the other hand, in the DC coefficient nonlinear inverse quantization unit 119A, the quantization scale value is set nonlinearly according to the range of the quantization step size 319, and the inverse quantization is performed using this quantization scale, and the DC coefficient 306 is output. .

  Next, the AC coefficient inverse quantization unit 120 performs inverse quantization on the AC coefficient data and outputs an AC coefficient 307. The inversely quantized DC coefficient (existing only in the intra coding mode) 306 and the AC coefficient 307 are passed to the inverse DCT unit 115 as the DCT coefficient 308, and after being subjected to inverse DCT, output as a decoded prediction error signal 309. Is done. The addition unit 116 adds the decoded prediction error signal 309 and the predicted texture data 9 obtained from the motion compensation unit 8 and outputs the result as decoded texture data 11. However, in the case of the intra coding mode, the addition of the predicted texture data 9 is not performed.

  The encoded bitstream 1 is H.264. When the H.263 compatible identification information 33 is multiplexed, as shown in FIG. 2A, a sequence end code (EOS) 227 indicating the end of the sequence may be multiplexed. The detection of the sequence end code 227 is performed by the picture start code detection unit 41. When the sequence end code 227 is detected, the decoding operation is ended.

  As described above, according to the first embodiment, the H.264 standard. 263 encoded bit stream 201, VO start code 231, VOL start code, VO identification number 232, and H.264. MPEG-4 compatible H.264 in which H.263 compatible identification information 226 is multiplexed. Since the H.263 encoded bit stream 203 is received and each of these pieces of information is decoded, the H.263 encoded bitstream 203 is decoded. An effect is obtained that an image decoding device compatible with H.263 and MPEG-4 can be obtained.

Embodiment 2. FIG.
FIG. 17 is a block diagram showing the configuration of the image coding apparatus according to the second embodiment, which generates a coded bitstream that can be decoded by the image decoding apparatus described in the first embodiment. In the figure, 121 is an input image signal, 122 is H.264. H.263 encoding unit 123 is an H.263 encoding unit. H.263 encoded bit stream, 124 is an MPEG-4 compatible flag, 125 is a header information multiplexing unit, and 126 is MPEG-4 compatible H.264. H.263 encoded bit stream.

Next, the operation will be described.
First, H. The H.263 method encoding unit 122 converts the input image signal 121 to the H.264 format. H.263 encoding is performed based on the H.263 syntax. A H.263 encoded bit stream 123 is generated. Next, the header information multiplexing unit 125 receives the MPEG-4 compatible flag 124 indicating that a bitstream that can be decoded by the MPEG-4 based decoding device is generated. The VO start code 231, the VO identification number 232, the VOL start code 233 and the H.264 code necessary for decoding by the image decoding apparatus described in the first embodiment before the picture header of the H.263 bit stream. 263 compatible identification information (0 or 1 flag indicating that the bit stream is based on H.263) 226 is multiplexed. MPEG-4 compatible H.264 multiplexed in this way. The contents of the H.263 encoded bit stream 126 are as shown in FIG. 2A of the first embodiment.

  As shown in FIG. When real-time communication is performed between the H.263 encoding device 127 and the MPEG-4 decoding device 128, the H.263 encoding device 127 and the MPEG-4 decoding device 128 are connected to the H.264 encoding device. The MPEG-4 compatible flag 124 is transmitted to the H.263 encoding device 127. When the H.263 encoding device 127 receives the MPEG-4 compatible flag 124, the VO start code 213, the VO identification number 232, the VOL start code 233, and the VO start code 233 necessary for decoding by the image decoding device described in the first embodiment. H. H.263 compatible identification information 226 is changed to H.263. It is also possible to multiplex with the H.263 system bit stream 123.

  As described above, according to the second embodiment, the VO start code 231, the VO identification number 232, the VOL start code 233, and the H.264 H.263 compatible identification information 226 is changed to H.263. Since it is multiplexed with the H.263 encoded bit stream 123, an image encoding apparatus that generates an encoded bit stream that can be decoded by an MPEG-4 compatible image decoding apparatus can be obtained.

Embodiment 3 FIG.
FIG. 19 shows MPEG-4 compatible H.264 in the third embodiment. FIG. 2 is a diagram illustrating the structure of an H.263 encoded bit stream 205, which is the same as the H.263 encoded bit stream 205 shown in FIG. 263 encoded bit stream 201, VO start code 231, VO identification number 232, and H.264 code. A H.263 start code 228 is added. And H. The 263 start code 228 includes the VOL start code 233 multiplexed in the first embodiment and the H.264 start code 228. Both functions of the H.263 compatible identification information 226 are achieved.
Note that the MPEG-4 encoded bit stream 202 is the same as that shown in FIG.

  The image decoding apparatus according to this embodiment differs from the image decoding apparatus described in Embodiment 1 only in the header information analysis unit 21. FIG. 20 is a block diagram showing the configuration of the header information analysis unit 21 in the third embodiment. In FIG. 263 start code / VOL start code detection unit 132 is an encoding method determination unit, VO start code detection unit 30, H.264 H.263 compatible identification information 33, switching unit 34, H.264. H.263 picture header information analysis unit 35, H.264 The H.263 GOB header information analysis unit 36, the VOL header information decoding unit 37, and the VOP header information analysis unit 38 are the same as those in FIG. 5 of the first embodiment.

Next, the operation will be described.
When the VO start code detection unit 30 detects the VO start code 231, the following decoding operation is started. First, H. The H.263 start code / VOL start code detector 131 is MPEG-4 compatible H.264. In the case of the H.263 encoded bit stream 205, H.264 is used. The H.263 start code is detected. In the case of the MPEG-4 encoded bit stream 202, the VOL start code 233 is detected.

  In MPEG-4, the start code of each layer is a code common to all start codes (0000 0000 0000 0000 0000 0001), followed by a start code unique to each layer in a fixed length (5 bits). The common start code is not detected except for the start code in the bitstream. Therefore, H.I. The H.263 start code 228 is also H.264 after the common start code. It is assumed that a fixed-length (5 bits) code that can identify the H.263 encoded bit stream is attached.

  Next, the detected start code is H.264. In the case of the H.263 start code 228, the encoding method determination unit 132 determines whether the H.263 start code 228 is H.263. H.263 compatible identification information 33 is changed to H.263. Set to 263. In the case of the VOL start code 233, the H.264 The H.263 compatible identification information 33 is set to MPEG-4. Subsequent operations are the same as those in the first embodiment.

  As described above, according to the third embodiment, the H.264 standard. 263 encoded bit stream 201, VO start code 231, VO identification number 232, and H.264 code. MPEG-4 compatible H.264 with multiplexed H.263 start code 228. Since the H.263 encoded bitstream 205 is received and each of these pieces of information is decoded, the H.263 encoded bitstream 205 is decoded. An effect is obtained that an image decoding device compatible with H.263 and MPEG-4 can be obtained.

Embodiment 4 FIG.
This embodiment is an image coding apparatus that generates a bitstream that can be decoded by the image decoding apparatus described in the third embodiment, and has the same configuration as that of FIG. 17 of the second embodiment.
Next, the operation will be described.
First, H. The H.263 method encoding unit 122 converts the input image signal 121 to the H.264 format. H.263 encoding is performed based on the H.263 syntax. A H.263 encoded bit stream 123 is generated. Next, the header information multiplexing unit 125 receives the MPEG-4 compatibility flag 124, and Before the picture header of the H.263 encoded bit stream 123, the VO start code 231, the VO identification number 232, and the H.264 code required for decoding by the image decoding apparatus described in the third embodiment. The H.263 start code 228 is multiplexed. MPEG-4 compatible H.264 multiplexed in this way. The contents of the H.263 encoded bit stream 126 are as shown in FIG. 19 of the third embodiment.
Note that the MPEG-4 compatibility flag 124 can also be transferred from the MPEG-4 decoding device 128 as described in FIG. 18 of the second embodiment.

  As described above, according to the fourth embodiment, the VO start code 231, the VO identification number 232, and the H.264 H.263 start code 228 Since multiplexing is performed on the H.263 bit stream 201, an image encoding device that generates an encoded bit stream that can be decoded by an MPEG-4 compatible image decoding device can be obtained.

Embodiment 5 FIG.
In this embodiment, a multiplexing unit that multiplexes header information for MPEG-4 compatibility is provided on a network, for example, independently of the encoding device. FIG. 21 shows an image communication system according to the fifth embodiment. In FIG. H.263 encoding device, 142 is an MPEG-4 decoding device, 143 is an encoded bitstream converting device, The H.263 encoding device 141, the MPEG-4 decoding device 142, and the encoded bit stream conversion device 143 are connected to a network.

Next, the operation will be described.
The encoded bitstream converter 143 receives an MPEG-4 compatible H.264 signal from the MPEG-4 decoder 142 or the user. When the MPEG-4 compatible flag 147 requesting the H.263 encoded bitstream 148 is received, the encoded bitstream conversion apparatus 143 receives the H.263 encoded bitstream 148. From the H.263 encoding device 141, H.263 encoded bit stream 146 is received and, as described in the second embodiment or the fourth embodiment, the header information necessary for decoding by the MPEG-4 decoding apparatus is converted to H.264. It is multiplexed with the H.263 encoded bit stream 146 and transmitted to the MPEG-4 decoding device 142.

  As described above, according to the fifth embodiment, since the encoded bitstream conversion apparatus 143 is provided on the network, An effect is obtained that an image communication system compatible with H.263 and MPEG-4 can be obtained.

Embodiment 6 FIG.
FIG. 22 shows an image communication system according to the sixth embodiment. In FIG. 263 encoding device, 143 is an encoded bit stream converting device, 144 is a server, 145 is a browser with a built-in MPEG-4 decoding device, and these are connected on the network.
Next, the operation will be described.
An MPEG-4 decoding device built-in browser 145 is transmitted over the network. When accessing the H.263 encoded bit stream 146, the MPEG-4 decoding device built-in browser 145 transmits to the server 144 an MPEG-4 compatible flag 147 indicating that decoding is performed by the MPEG-4 decoding device. When the server 144 receives the MPEG-4 compatible flag 147, the H.264 The H.263 encoded bit stream 146 is transmitted to the encoded bit stream converter 143.

  The encoded bitstream converter 143 receives the received H.264. As described in the second or fourth embodiment, the header information is added to the H.263 bit stream 146 so that it can be decoded by the MPEG-4 decoding apparatus. A H.263 encoded bit stream 148 is generated and transmitted to the browser 145 with a built-in MPEG-4 decoding device. In the MPEG-4 decoding device built-in browser 145, MPEG-4 compatible H.264 is used. By receiving the H.263 encoded bitstream 148, It is possible to decode the H.263 encoded bitstream 146 and display the image.

  It is also possible for the MPEG-4 decoding device built-in browser 145 itself to incorporate the coded bitstream conversion device 143. In this case, the MPEG-4 decoding device built-in browser 145 is H.264. 263 encoded bitstream 146 is received from server 144 and MPEG-4 compatible H.264 is received. After being converted into the H.263 encoded bit stream 148, it can be decoded by a built-in MPEG-4 decoding device and displayed.

  As described above, according to the sixth embodiment, since the encoded bit stream conversion device and the server are provided on the network, the An effect is obtained that an image communication system compatible with H.263 and MPEG-4 can be obtained.

Embodiment 7 FIG.
In the image decoding apparatus described in the first embodiment and the third embodiment, H.264 A 263 bitstream or an MPEG-4 bitstream can be identified. However, H. H.264 encoding device generated by the H.263 encoder. It is necessary to multiplex MPEG-4 compatible header information at the head of the H.263 bit stream. The 263 bit stream cannot be received as it is. The seventh embodiment is described in H.264. This is an image decoding apparatus that can receive and decode a 263-bit stream as it is.

  FIG. 23 is a block diagram showing the configuration of the header information analysis unit 21 in the seventh embodiment. In FIG. H.264 encoded in the H.263 encoded bitstream. H.263 detecting the picture start code 221 of H.263. 263 picture start code detection unit, 152 is an encoding method determination unit, 153 is H.264. VOL header information and VOP header information are set based on the picture header information 222 multiplexed in the H.263 encoded bit stream. 263 picture header information analysis unit. Other VO start code detector 30, H.263 compatible identification information 33, switching unit 34, H.264. The H.263 GOB header information analysis unit 36, the VOL header information decoding unit 37, and the VOP header information analysis unit 38 are the same as those in the first embodiment. Further, the configuration other than the header information analysis unit 21 is the same as that of the image decoding apparatus according to the first embodiment.

Next, the operation will be described.
H. The H.263 picture start code detection unit 151 constantly monitors the beginning and end of the encoded bit stream shown in FIGS. H. In the case of the H.263 coded bit stream 201, one picture code from the picture start code 221 to the macro block data 225, and in the case of the MPEG-4 coded bit stream 202, one code from the VO start code 231 to the macro block data 239, respectively. Monitor as a generalized bitstream.

  H. When the H.263 encoded bitstream 201 is received, the H.263 encoded bitstream 201 is received. The H.263 picture start code detection unit 151 detects the picture start code 221 and outputs the result to the encoding method determination unit 152. The encoding method determination unit 152 uses the detected picture start code 221 to determine that the received encoded bit stream is H.264. H.263 encoded bit stream 201 is determined. H.263 compatible identification information 33 is changed to H.263. Set to H.263. On the other hand, when the VO start code detection unit 30 detects the VO start code 231, the encoding method determination unit 152 determines that the received encoded bit stream is the MPEG-4 encoded bit stream 202. The H.263 compatible identification information 33 is set to MPEG-4.

H. In the case of the H.263 encoded bit stream 201, the switching unit 34 performs the H.264 encoding. H.263 encoded bit stream 201 is converted to H.264. This is input to the H.263 picture header information analysis unit 153. H. The H.263 picture header information analysis unit 153 includes the H.263 picture header information analysis unit 153. The picture header information 222 multiplexed in the H.263 encoded bit stream 201 is decoded, and the VOL header information and the VOP header information are set as in the first embodiment. The subsequent operations are the same as those in the first embodiment.
On the other hand, in the case of the MPEG-4 encoded bit stream 202, the switching unit 34 inputs the MPEG-4 encoded bit stream 202 to the VOL header information decoding unit 37. The subsequent operations are the same as those in the first embodiment.

  As described above, according to the seventh embodiment, when the picture start code 221 is detected, H.264 is detected. 263 encoded bit stream 201 is determined and VOL header information and VOP header information are set. An effect is obtained that an image decoding device compatible with H.263 and MPEG-4 can be obtained.

Embodiment 8 FIG.
This embodiment is the same as the H.264 shown in FIG. The present invention relates to an encoded bit stream conversion apparatus that converts an H.263 encoded bit stream 201 into an MPEG-4 encoded bit stream 202 shown in FIG.
FIG. 25 is a block diagram showing an encoded bit stream conversion apparatus according to the eighth embodiment. A syntax analysis unit that separates the H.263 encoded bitstream 201 into a picture header information codeword 401, a GOB header information codeword 402, and a macroblock data codeword 403; 162, a picture header information decoding that decodes the picture header information codeword 401 163, a GOB header information analysis / conversion unit for decoding the GOB header information codeword 402, 164, an MPEG-4 header information setting unit for setting VOL header information 234 and VOP header information 236, and 165, an MPEG-4 code It is a multiplexing unit that outputs the normalized bit stream 202.

Next, the operation will be described.
The syntax analysis unit 161 is an H.264 format. When the picture start code 221 is detected from the H.263 encoded bitstream 201, the subsequent encoded bitstream is separated into a picture header information codeword 401, a GOB header information codeword 402, and a macroblock data codeword 403, and each picture The data is output to the header information decoding unit 162, the GOB header information analysis / conversion unit 163, and the multiplexing unit 165. However, the GOB header information codeword 402 is always H.264. It is not necessarily multiplexed with the H.263 encoded bit stream 201, and is multiplexed only when the GOB start code 223 is detected. When the GOB start code 223 is detected, the GOB header detection information 404 is output to the MPEG-4 header information setting unit 164. The picture header information decoding unit 162 decodes the picture header information codeword 401 in the same manner as in Embodiment 1, and outputs the decoded picture header information 405 to the MPEG-4 header information setting unit 164.

  The MPEG-4 header information setting unit 164 sets the VOL header information 234 and the VOP header information 236 based on the decoded picture header information 405 as in the first embodiment. The header information not described in the first embodiment may be set to any value disclosed in ISO / IEC JTC1 / SC29 / WG11 MPEG-4 Video VM8.0. When the GOB header detection information 405 is received, the error resistant coding instruction mode is set to be valid.

As described in the first embodiment, the H.264 standard. 263 and MPEG-4 are partially different in macroblock data decoding processing procedures. Therefore, on the decoding side, it is necessary to switch the decoding method based on the switching information. Therefore, it is necessary to set the following switching information in the VOL header.
(1) AC coefficient VLC table switching information As described in the first embodiment, when the VLC table used for variable-length encoding of AC coefficient data on the encoding side is different, the AC coefficient data is variable on the decoding side. Information for switching the VLD table used for long decoding.
(2) Esc Coding Switching Information As described in Embodiment 1, when AC coefficient data is variable-length encoded on the encoding side, the code when the AC coefficient data to be encoded is not in the VLC table Information for switching the decoding method on the decoding side when the encoding method is different.
(3) Intra DC coefficient dequantization switching information Information for switching the DC coefficient dequantization method when the intra DC coefficient quantization method is different on the encoding side, as described in the first embodiment.
In addition, the switching information of (1) to (3) is collected and It is also possible to set this as information to be switched between the method employed in H.263 and the others.

The MPEG-4 header information set by the MPEG-4 header information setting unit 164 is variable length encoded and output to the multiplexing unit 165 as an MPEG-4 header information code word 406.
Similar to the first embodiment, the GOB header information analysis / conversion unit 163 decodes the GOB header information codeword 402 and converts the GOB header information 224 into resynchronization information 238 that is an MPEG-4 expression format.

  MPEG-4 resynchronization information 238 is information used to enhance error tolerance, and is multiplexed when error tolerance coding instruction information of VOL header information 236 is valid. When the resynchronization information 238 is decoded on the decoding side, resynchronization with the encoded bitstream is performed, and the prediction vector and the quantization step size used in macroblock decoding are reset. H. In H.263, when the GOB header information 224 is decoded, the prediction vector and the quantization step size are reset. Therefore, by converting the GOB header information 224 with the resynchronization information 238, the GOB header information 224 can be converted into an MPEG-4 representation format.

FIG. 26 shows the structure of GOB header information 224 and resynchronization information 238. The macroblock number 271 in the resynchronization information 238 is a number indicating the position of the macroblock in the VOP. Therefore received H. What is necessary is just to calculate the position in the picture of the macroblock corresponding to 263 macroblock data. Since this is the first macroblock in the GOB, it can be calculated from the GOB number. The quantization scale 272 may set the GOB quantization step size. When the header extension instruction code 273 is “1”, the time reference point 274 and the VOP elapsed time 275 are multiplexed. These pieces of information are used when displaying each VOP. If necessary, the header extension instruction code 273 may be set to “1”, and the time reference point 274 and the VOP elapsed time 275 may be set. The resynchronization information 238 is variable-length encoded, and a resynchronization information codeword 407 to which a resynchronization instruction code, which is a fixed-length unique code for indicating that the resynchronization information 238 is multiplexed, is multiplexed. To 165.
The multiplexing unit 165 multiplexes the MPEG-4 header information codeword 406, the resynchronization information codeword 407, and the macroblock data codeword 403 into the encoded bitstream, and outputs the MPEG-4 encoded bitstream 202.

In this embodiment, the resynchronization information is information that is multiplexed when the error-resistant coding instruction information of the VOL header information 234 is valid. However, the resynchronization information is related to whether the error-resistant coding instruction information is valid or invalid. It is also possible to multiplex them.
H. In the H.263 encoded bit stream 201, when the sequence end code 227 is added after the macroblock data 225, the syntax analysis unit 161 ends the analysis when the sequence end code 227 is detected.

  As described above, according to this embodiment, the H.264 standard. Since the H.263 encoded bit stream 201 is converted into the MPEG-4 encoded bit stream 202, the MPEG-4 image decoding apparatus uses the H.264 encoded bit stream 201. The effect that the H.263 encoded bit stream can be decoded is obtained.

Embodiment 9 FIG.
In the seventh embodiment, in FIG. When the H.263 picture start code detection unit 151 detects the picture start code 221, the encoding method determination unit 152 determines that the H.263 picture start code detection unit 151 detects the picture start code 221. H.263 encoded bit stream 201, and H.263 encoded bit stream 201 is determined. The H.263 picture header information analysis unit 153 sets the VOL header information and the VOP header information. H.263 picture header information analysis unit 153 included in FIG. Based on the picture header information 222 decoded by the H.263 picture header information decoding unit 42, the operation of the macroblock layer syntax analysis unit 22 is switched. At this time, the MPEG-4 header information setting unit 43 is unnecessary. In the seventh embodiment, the GOB start code detection unit 61 in FIG. When the GOB start code 223 is detected in the H.263 encoded bit stream 201, the GOB header information decoding unit 62 decodes the GOB header information 224, and the MPEG-4 header information changing unit 63 includes it in the VOP header information 236. The VOP quantization step size was reset. In the present embodiment, H.264. When decoding the H.263 encoded bitstream 201, the picture quantization step size 304 included in the picture header information 222 may be reset because the macroblock data is decoded using the picture header information 222.

Next, H. Based on the picture header information 222 decoded by the H.263 picture header information decoding unit 42, an operation of the macroblock layer syntax analysis unit 22 when decoding macroblock data will be described.
In the present embodiment, the operations of the switching units 81, 83, 88, and 95, the operation of the addition unit 94, the operation of the motion vector decoding unit 97 of the macroblock layer syntax analysis unit of FIG. 12, and the block of FIG. Since only the operation of the switching unit 102 of the data decoding unit 98 is different, only this part will be described.
The switching unit 81 decodes the MPEG-4 encoded bitstream 202, that is, the H.264 setting set by the encoding method determination unit 152 in FIG. When the H.263 compatible identification information 33 indicates MPEG-4, switching is performed according to the shape information decoded by the VOL header information decoding unit 37. H. When the H.263 encoded bitstream 201 is being decoded, H.263 compatible identification information 33 is H.264. In the case of H.263, the bit stream 1 is input unconditionally to the switching unit 83 without going through the shape encoded data decoding unit 82.

  When the MPEG-4 encoded bitstream 202 is being decoded, the switching unit 83 is switched according to the VOP prediction type decoded by the VOP header information analysis unit 38. H. When the H.263 encoded bitstream 201 is being decoded, the H.263 encoded bitstream 201 is decoded. Switching is performed according to the picture coding type 302 decoded by the H.263 picture header information decoding unit 42. The switching operation is the same as in the first embodiment, and switching may be performed depending on whether the picture encoding type 302 is intra or other.

  When the MPEG-4 encoded bit stream 202 is being decoded, the switching unit 88 is switched based on the intra AC / DC prediction instruction information decoded by the VOL header information decoding unit 37. H. When the H.263 encoded bitstream 201 is being decoded, H.263 compatible identification information 33 is H.264. In the case of H.263, the bit stream 1 is input unconditionally to the effective block identification information decoding unit 90 without going through the AC prediction instruction information decoding unit 89.

  When decoding the MPEG-4 encoded bit stream 202, the adding unit 94 adds the VOP quantization step size of the previously decoded macroblock to the decoded differential quantization step size 254. And output as a quantization step size. H. When the H.263 encoded bitstream 201 is being decoded, the picture quantization step size of the macroblock decoded immediately before is added to the decoded differential quantization step size 254 to obtain the quantization step size. Output as.

  When the MPEG-4 encoded bit stream 202 is being decoded, the switching unit 95 is switched based on the interlace mode instruction information decoded by the VOP header information analysis unit 38. H. When the H.263 encoded bitstream 201 is being decoded, H.263 compatible identification information 33 is H.264. In the case of H.263, the bitstream 1 is unconditionally input to the motion vector decoding unit 97 without going through the interlace information decoding unit 96.

  When the MPEG-4 encoded bit stream 202 is being decoded, the motion vector decoding unit 97 determines a motion vector (texture motion) based on the motion vector search range designation information decoded by the VOP header information analysis unit 38. Decrypt data 7). H. When decoding a H.263 bit stream, H.264 is used. Based on the motion vector search range defined in H.263, a motion vector (texture motion data 7) is decoded.

  When the MPEG-4 encoded bitstream 202 is being decoded, the switching unit 102 of the block data decoding unit 98 is switched by the intra AC / DC prediction mode instruction information decoded by the VOL header information decoding unit 37. . H. When the H.263 encoded bitstream 201 is being decoded, H.263 compatible identification information 33 is H.264. In the case of H.263, the bitstream 1 is input to the DC coefficient fixed length decoding unit 103 unconditionally. The subsequent operations are the same as those in the first embodiment.

  As described above, according to the ninth embodiment, when the picture start code 221 is detected, the Since it is determined as the H.263 encoded bit stream 201, the picture header information 222 is decoded, and the macroblock data is decoded based on the decoded picture header information 222. Therefore, the VOL header information and the VOP header information are not set. H., et al. An effect is obtained that an image encoding device having compatibility between H.263 and MPEG-4 can be obtained.

  As described above, the image decoding apparatus, the image encoding apparatus, the image communication system, and the encoded bit stream conversion apparatus according to the present invention can perform transmission / reception with a simple configuration even with encoded bit streams having different encoding methods.

Conventional H.264. It is a figure which shows the structure of a H.263 encoding bit stream and a MPEG-4 encoding bit stream. It is a figure which shows the structure of the encoding bit stream which the image decoding apparatus by Embodiment 1 of this invention receives. It is a block diagram which shows the structure of the image decoding apparatus by Embodiment 1 of this invention. It is a block diagram which shows the structure of the syntax analysis and variable length decoding part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the header information analysis part by Embodiment 1 of this invention. H. according to Embodiment 1 of the present invention. It is a block diagram which shows the structure of a H.263 picture header information analysis part. H. according to Embodiment 1 of the present invention. It is a block diagram which shows the structure of a H.263 picture header information decoding part. H. according to Embodiment 1 of the present invention. It is a block diagram which shows the structure of a H.263 GOB header information analysis part. It is a figure explaining GOB. It is a block diagram which shows the structure of the GOB header information decoding part by Embodiment 1 of this invention. H. It is a figure which shows the layer structure of 263 macroblock data. It is a block diagram which shows the structure of the macroblock layer syntax analysis part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the block data decoding part by Embodiment 1 of this invention. It is a figure explaining calculation of a prediction vector. It is a block diagram which shows the structure of the texture decoding part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the inverse quantization part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the image coding apparatus by Embodiment 2 and 4 of this invention. According to the second and fourth embodiments of the present invention, It is a figure explaining the relationship between a H.263 encoding apparatus and an MPEG-4 decoding apparatus. MPEG-4 compatible H.264 according to Embodiment 3 of the present invention. It is a figure which shows the content of a H.263 encoding bit stream. It is a block diagram which shows the structure of the header information analysis part by Embodiment 3 of this invention. It is a figure which shows the image communication system by Embodiment 5 of this invention. It is a figure which shows the image communication system by Embodiment 6 of this invention. It is a block diagram which shows the structure of the header information analysis part by Embodiment 7 of this invention. It is a figure explaining the beginning and end of the encoding bit stream by Embodiment 7 of this invention. It is a block diagram which shows the encoding bit stream conversion apparatus by Embodiment 8 of this invention. It is a figure which shows the structure of GOB header information and resynchronization information.

Explanation of symbols

1 encoded bit stream, 2 syntax analysis / variable length decoding unit, 4 shape decoding unit, 8 motion assurance unit, 10 texture decoding unit, 21 header information analysis unit, 22 macroblock layer syntax analysis unit, 30 VO start detection Part, 31VOL start code detection part, 32 encoding method determination part, 34 switching part, 35 H. 263 picture header information analysis unit, 36 H.264. 263 GOB header information analysis unit, 37 VOL header analysis unit, 38 VOP header information analysis unit, 41 H.264 263 picture start code detector, 42 H.264. 263 picture header information decoding unit, 43 MPEG-4 header information setting unit, 61 GOB start code detection unit, 62 GOB header information decoding unit, 63 MPEG-4 header information changing unit, 71 GOB number decoding unit, 72 GOB identification number decoding 73, GOB quantization step size decoding unit, 122 H. H.263 encoding unit, 125 header information multiplexing unit, 203, 205 MPEG-4 compatible H.264 263 encoded bit stream, 204 MPEG-4 encoded bit stream.

Claims (1)

At least H. H.263 encoding method header information and H.264 encoding method. A first encoded bit stream obtained by multiplexing image encoded data encoded by the H.263 encoding method, or an image encoded by the header information of the MPEG-4 encoding method and the MPEG-4 encoding method In an image decoding apparatus that decodes a second encoded bit stream multiplexed with encoded data,
Based on the header information of the first encoded bit stream or the header information of the second encoded bit stream, the received encoded bit stream is converted into the first encoded bit stream or the second encoded bit stream. An encoding method determining means for determining whether the bit stream is present;
When the encoding scheme determination means determines that the first encoded bitstream has been received, H.263 encoding method header information included in the H.263 encoding method header information. H.264 format image encoding information defined based on H.263 format image encoding information is used. When the H.263 encoded bit stream is decoded in accordance with the syntax of the MPEG-4 encoded bit stream, and the encoding scheme determination means determines that the second encoded bit stream has been received, the received second The encoded image data included in the second encoded bit stream is converted to the encoded image data included in the second encoded bit stream using the MPEG-4 encoded image encoded information included in the header information of the encoded MPEG-4 encoding method. In the decoding process of decoding, the quantization scale used for the inverse quantization process of the DC component of the intra DCT coefficient is a fixed value when the determination unit determines that the first encoded bitstream has been received. set, if it is determined that it has received the second coded bit stream quantized determined for each macroblock Setting means for setting a value corresponding to the value of the step size,
An image decoding apparatus comprising: inverse quantization processing means for performing inverse quantization processing of a direct current component of an intra DCT coefficient using a quantization scale determined by the setting means.

Images