JP4851270B2 - Moving picture decoding apparatus and decoding method. - Google Patents

Description

  The present invention relates to a moving image decoding apparatus and a decoding method that take a difference between an original image and a prediction image generated by inter-frame prediction and intra-frame prediction, encode the difference, and transmit the difference to the decoding device side.

  Video coding systems such as MPEG (Moving Picture Experts Group) are widely used in various fields such as digital broadcasting, Internet communication, and mobile phone communication. On the decoding device side in such a moving image coding system, the bit stream data sent from the coding device side is decoded and output. For example, the bit stream data is transmitted by a wireless communication transmission path. In some cases, an error may occur in the bit stream data due to factors such as deterioration of the communication state, and data including the error may be input to the decoding device.

  When such an error is detected on the decoding device side, the most reliable method for error correction is to request the encoding device side to retransmit the encoded data of that part, but in practice communication From various viewpoints such as maintaining speed, a data retransmission request is not a very practical method.

  As a result, when an error is detected on the decoding device side, it is necessary to perform processing on the decoding device side so that the influence of the error is minimized. Established in 2003 by ITU-T (International Telecommunication Union-Telecommunication Standardization Sector). In the H.264 video compression encoding method, a redundant slice is used as one countermeasure against such a transmission error, and when an error occurs, the content is used to restore the decoded image for the macroblock in which the error has occurred. be able to.

  When such a method cannot be used, an error concealment process may be performed on the decoding device side in order to reduce the influence of errors as much as possible, and various error concealment methods have been proposed. Such error concealment processing will be described using a conventional example of a video decoding device.

  FIG. 16 is a block diagram showing the configuration of a conventional example of a moving picture decoding apparatus. In the figure, the decoding apparatus includes a variable length decoding processing unit 100 that performs decoding processing on input stream data, an intra-frame prediction unit 101 that generates an intra-frame prediction image, an inter-frame prediction processing unit 102 that generates an inter-frame prediction image, The inverse quantization unit 103 and the inverse transform unit 104 for the output of the variable length decoding processing unit 100, the switch 105 for selecting the outputs of the intra-frame prediction unit 101 and the inter-frame prediction processing unit 102, and the selection result of the switch 105 and the inverse transform An addition unit 106 that adds the output of the unit 104, a deblocking filter processing unit 107 that outputs a decoded image by removing block distortion generated at a block boundary in the block decoding process from the output of the addition unit 106, and stores the decoded image And a frame memory 108 to be supplied to the inter-frame prediction processing unit 102. During prediction processing unit 102 includes a vector reproducing section 109, and the inter-frame prediction image generating unit 110 for detecting a motion vector.

  There is a tendency that the size of an image to which a moving image prediction method such as MPEG is applied tends to be large, and the decoding apparatus as described with reference to FIG. 16 also requires a high processing capacity to process large image data. It has become. In such a case, processing for improving the overall processing capability of the system without increasing the processing speed of each functional block by performing processing on different image blocks at the same time as each functional block as a configuration of the decoding device A method such as a pipeline processing method is used.

  FIG. 17 is a timing chart of a decoding operation example by pipeline processing in units of macroblocks (MB) in the conventional example of FIG. In the first stage of the figure, vector reproduction processing by the vector reproduction unit 109 is performed as part of the inter-frame prediction process for the macroblock with the identifier n + 1, and the preceding macroblock in the same stage, that is, the macro with the identifier n Inter-frame prediction image generation processing by the inter-frame prediction image generation unit 110 is performed on the block.

  In the next stage, an inter-frame prediction image generation process is performed on the macroblock with the identifier n + 1, and an inverse quantization / inverse transform process is performed on the macroblock with the identifier n by the inverse quantization unit 103 and the inverse transform unit 104.

  Further, in the next stage, inverse quantization / inverse transform processing is performed on the macro block with the identifier n + 1, and thereby the inter-frame prediction processing by the three-stage pipeline for the n + 1 macro block is completed. In the same stage, vector reproduction processing is performed on the macroblock with the identifier n + 3, and thereafter the same operation is performed.

  For the macroblock with the identifier n + 2, for example, intraframe prediction by the intraframe prediction unit 101 is selected in accordance with the determination result of intraframe / interframe by the variable length decoding processing unit 100, and the intraframe for this macroblock is selected. The prediction image generation process and the inverse quantization / inverse transform process are completed in one stage.

  Thus, the process for inter-frame prediction requires, for example, a three-stage pipeline. That is, the image data used in inter-frame prediction is already decoded image data before the image currently being decoded, and is used in processes such as intra-frame prediction, inverse transform, and inverse quantization. The amount is much larger than that of FIG. 16 and is accumulated in the frame memory 108 of FIG. 16, and it takes time to acquire data from the frame memory 108. In FIG. 17, the data processing unit in the inter-frame prediction in the macro block may differ from the data processing unit in the intra-frame prediction, inverse quantization, and inverse transform, so that the block rearrangement in the macro block. In response to this need, the processing stage in the pipeline is set as a separate stage in, for example, inter-frame prediction image generation and inverse quantization / inverse transform processing.

  In FIG. 17, for example, it is assumed that an error is detected in the process of the inverse quantization unit 103 or the inverse transform unit 104 for the macroblock with the identifier n. In general, in a video prediction method, a variable-length code method is often used, and when a decoding error is detected at a certain point in time, even if the bit after the error bit is correctly transferred, The possibility of correct decoding is very low. For this reason, it is provided inside the stream, for example, for a section up to the return position in the stream where a correct decoding result can be obtained by decoding from that position, such as a picture boundary or a slice boundary. In general, a process in which all the macroblocks corresponding to the section are subjected to error concealment is performed without using the stream data as it is in the decoding process.

  As an example of this error concealment process, here, either error concealment using inter-frame prediction or error concealment using intra-frame prediction is used. In error concealment based on inter-frame prediction, information corresponding to motion vector information included in the bitstream is determined by some method, and a part of adjacent images in the display order is cut out using the determined vector information. Is subjected to a filter to make a compensation image for the error concealment target macroblock. As a simple implementation example, the absolute value of the vector is set to “0”, and an image decoded immediately before can be used as an image for cutting out a supplementary image.

  On the other hand, in error concealment based on intra-frame prediction, a method for generating an intra-frame predicted image or a method close thereto is defined using correct image data that has already been decoded around the macro block subject to error concealment. A supplementary image is generated using. However, it is generally known that error concealment processing using inter-frame prediction can provide a higher quality error concealment result than using intra-frame prediction.

  FIG. 18 is an explanatory diagram of a pipeline reset example prior to error concealment processing in the conventional example. As described with reference to FIG. 17, it is assumed that an error has been detected for the macroblock with the identifier n preceding the macroblock with the identifier n + 1 for which interframe prediction is selected. As described with reference to FIG. 17, the inter-frame prediction scheme has already been applied to the n + 1 macroblock, and an error is detected for the macroblock with the identifier n. The result of the vector reproduction processing for the macroblock of, and the inter-frame prediction image generation processing in the next stage becomes unreliable, and in order to apply the inter-frame prediction processing to the macroblock with the identifier n + 1, reset the pipeline once However, it is necessary to repeat the processing from the vector reproduction processing for this macroblock. As a result, the interframe prediction processing as error concealment can be completed only after two stages, and processing is performed by resetting the pipeline. As time goes on, pipes There is a problem that the control mechanism be newly required for in reset.

  Patent Document 1 as a prior art relating to error concealment in such an image decoding apparatus reads out image data in a compensation region at a position corresponding to an error macroblock from a previous frame without providing a memory for storing motion vectors. An error image interpolation method capable of interpolating a lost image of error macroblocks with a more natural image is disclosed.

  Similarly, Patent Document 2 as a conventional technique searches for adjacent blocks that are inter macro blocks in the order of left, top, top right, and top left of an error macro block, and if all are intra macro blocks, the previous one. A reference frame determination method using a decoded frame as a reference frame is disclosed.

JP-A-2001-309388 “Error Image Interpolation Method in Image Decoding Device” Japanese Patent Application Laid-Open No. 2005-252549 “Error Concealment Reference Frame Determination Method and Decoder”

  In the prior art described above, immediately after an error for a certain macroblock is detected, it is necessary to perform pipeline reset in the case of inter-frame prediction processing as error concealment processing, which extends the overall processing time. In addition, there is a problem that a new control mechanism is required to redo the pipeline reset and the inter-frame prediction process.

A moving picture decoding apparatus according to the present invention is an apparatus that generates a prediction image by dividing an image, which is stream data, into blocks, and decodes moving image data, and a frame that generates an intra-frame prediction image for a processing target block After the start of the inter-frame prediction means, the inter-frame prediction means for generating the inter-frame prediction image for the processing target block, and the inter-frame prediction processing for one block, an error is detected for the block temporally preceding the one block. The intra-frame prediction process as error concealment processing for the one block is executed by the intra-frame prediction means , and the start code in the stream data from the block temporally following the one block In the previous block, the error concealment process is performed. Blocks in which the inter-frame prediction process is performed within a predetermined period by causing the intra-frame prediction means to execute the intra-frame prediction process for blocks that do not fit within the predetermined period. Control means for causing the inter-frame prediction means to execute inter-frame prediction processing.

  According to the present invention, a pipeline reset is performed by causing an intra-frame prediction process to be executed as an error concealment process for a block immediately after a block in which an error is detected and for which an inter-frame prediction process has already been started. In addition, a control mechanism for re-execution of inter-frame prediction is unnecessary, and it is possible to avoid an increase in overall processing time.

  The present invention is described in, for example, H.D. Related to error concealment processing to be performed on the decoding device side when an error is detected in decoding the encoded data sent from the encoding device using the H.264 moving image compression coding method To do. Various schemes are conceivable as a scheme for sequentially decoding the continuous frames in the stream input on the decoding apparatus side. Here, in order to efficiently execute the decoding process, the conventional example shown in FIG. The embodiment of the invention will be described on the assumption that pipeline processing is performed in the same manner as in the description, but the present invention is not limited to a pipeline decoding apparatus.

  FIG. 1 is a block diagram showing the principle configuration of a moving picture decoding apparatus according to the present invention. The video decoding device 1 is a device that generates a predicted image by dividing an image into blocks and decodes the moving image, and includes an intra-frame prediction unit 2 that generates an intra-frame predicted image for at least a processing target block, and a process Inter-frame prediction means 3 for generating an inter-frame prediction image for the target block, and when an error is detected for a block temporally preceding the one block after the start of inter-frame prediction processing for one block, the first And control means 4 for executing intra-frame prediction processing as error concealment processing for one block.

  In the embodiment of the invention, the result of the variable length decoding process is inversely quantized and inversely transformed and added to the intra-frame prediction image or the inter-frame prediction image, and is output as a basic decoded image. In order to perform inter-frame prediction as an error concealment process for a block for which inter-frame prediction is to be performed when an error is detected for a block temporally preceding the block to which inter-frame prediction is to be applied For example, instead of taking a pipeline reset method, error concealment processing is performed, and intra-frame prediction is applied instead of inter-frame prediction.

  FIG. 2 is a block diagram showing the overall configuration of the moving picture decoding apparatus according to the first embodiment of the present invention. In the first embodiment, it is assumed that the frame image is decoded by three-stage pipeline control. 2 is basically different from FIG. 16 of the conventional example in that a control unit 10 for controlling the pipeline control and the error concealment process is added.

  FIG. 3 is a configuration block diagram of the control unit 10 in the first embodiment. In the figure, the control unit 10 corresponds to an error notification signal sent from the inverse quantization unit 15 or the inverse transform unit 16, an error information holding unit 25 that holds the content thereof, the stage 1 of the three-stage pipeline, A pipeline stage 1 control unit 26, a pipeline stage 2 control unit 27, and a pipeline stage 3 control unit 28 are provided to control the stage 2 and the stage 3, respectively.

  The pipeline stage 1 corresponds to vector reproduction processing, and a vector reproduction start signal is given to the vector reproduction unit 21 from the pipeline stage 1 control unit 26. Stage 2 corresponds to an inter-frame prediction image generation process, and an inter-frame prediction image generation activation signal is given to the inter-frame prediction image generation unit 22 from the pipeline stage 2 control unit 27. Stage 3 includes intra-frame prediction, which corresponds to the process of adding the prediction result of intra-frame prediction or inter-frame prediction and the conversion result of the inverse conversion unit 16 by the adder 17, and from the pipeline stage 3 control unit 28 An intra-frame prediction activation signal is provided to the intra-frame prediction unit 14 and an addition unit input selection control signal for the switch 18 that selects an input signal to the addition unit 17 is provided.

  Although not shown, from the variable-length decoding processing unit 11 in FIG. 2, either the macroblock to which inter-frame prediction is applied or the macroblock to which intra-frame prediction is applied is the current macroblock to be processed in the input stream. A signal as a result of the determination is given to, for example, the pipeline stage 1 control unit 26, and the information is also transmitted to the pipeline stage 2 control unit 27 and the pipeline stage 3 control unit 28.

  The processing by the control unit in FIG. 3 will be described with reference to the flowcharts in FIGS. FIG. 4 is a flowchart of processing by the pipeline stage 1 control unit 26. In the figure, in step S1, it is determined whether or not there is an error in this macro block, that is, the macro block prior to the macro block currently being processed. If there is an error, an error concealment operation is performed in step S2. The vector playback unit is activated. If it is determined in step S1 that there is no error, it is determined in step S3 whether or not the prediction method for this macroblock is interframe prediction. If it is not interframe prediction, stage 1 is performed without performing any processing. When the process is completed and the inter-frame prediction is performed, the vector reproduction unit is activated as a normal operation in step S4. In the inter-frame prediction process as the error concealment operation performed in step S2, as described above, for example, the value of the motion vector is set to “0”, and the image data decoded immediately before is used.

  FIG. 5 is a flowchart of processing by the pipeline stage 2 control unit 27. In stage 2, first, in step S6, it is determined whether or not the current macroblock to be processed has been processed as an error concealment operation target in stage 1 of the pipeline. The image generation unit is activated. If it is not the target of the error concealment operation, it is determined in step S8 whether or not the macroblock is the target of inter-frame prediction, and if it is not the target, that is, the target of intra-frame prediction. Stage 2 is terminated without performing any processing, and if it is the target of inter-frame prediction, the process proceeds to step S7.

  FIG. 6 is a flowchart of processing by the pipeline stage 3 control unit 28. In stage 3, first, in step S10, it is determined whether or not the processing target macroblock is the target of the error concealment operation in stage 1, and if so, the adding unit 17 performs inter-frame prediction processing in step S11. It is connected to the unit 13 side, and inverse quantization / inverse transformation processing is performed in step S12.

  If it is determined in step S10 that the processing target macroblock is not an error concealment operation target in stage 1, it is determined in step S13 whether or not the prediction method for the macroblock is interframe prediction. In the case of prediction, the processing after step S11 is performed. If the prediction is not inter-frame prediction, the adding unit 17 is connected to the intra-frame prediction unit 14 in step S14, and it is determined in step S15 whether or not there is an error in a macroblock prior to the current macroblock to be processed. If there is, the intra-frame prediction unit is activated as an error concealment operation in step S16, and if there is no error, the intra-frame prediction unit is activated as a normal operation in step S17.

  FIG. 7 is an explanatory diagram of macroblocks that are subject to error concealment when an error is detected in the stream. In the stream, a start code is embedded as a position where correct decoding is possible by decoding from that position. In general, errors are detected in this start code, that is, the macro block up to the return position. It becomes the target of error concealment processing starting from the macro block next to the macro block. This start code is provided, for example, at a picture boundary or a slice boundary.

  8 and 9 are time charts of an example of image data decoding processing in the first embodiment. FIG. 8 is a time chart of a normal operation in which no error is detected, and FIG. 9 is a time chart in a case where an error is detected and the error concealment process is performed up to the return position.

  In these time charts, processing is performed in three stages of the pipeline, that is, three stages, for the macroblock for inter-frame prediction, whereas for the macroblock for intra-frame prediction in FIG. As described, processing is performed using only stage 3 as the last stage of the three-stage pipeline.

  In FIG. 8, in the first stage, for the macroblock with the identifier n, the inter-frame prediction image generation unit 22 is activated to generate the inter-frame prediction image, and for the macro block with the identifier n + 1, vector reproduction is performed. A vector reproduction process is performed in response to the process activation instruction. Here, inter-frame prediction is applied to macroblocks with identifiers n and n + 1.

  In the next stage, for the macroblock with the identifier n + 1, an inter-frame prediction image generation unit activation instruction is issued to generate an inter-frame prediction image, and for the macroblock with the identifier n, inverse quantization / Inverse conversion processing is performed. At this time, the adding unit 17 in FIG. 2 selects the inter-frame prediction image generated by the inter-frame prediction processing unit 13.

  In the next stage, inverse quantization / inverse transform processing is performed on the macroblock with the identifier n + 1. At this time, the adding unit 17 selects an inter-frame prediction image. In the next two stages, intra-frame prediction image generation and inverse quantization / inverse transform processing are performed on macroblocks with identifiers n + 2 and n + 3. This is because intra-frame prediction is applied to these two macroblocks. At this time, the addition unit 17 naturally selects the intra-frame prediction image output from the intra-frame prediction unit 14.

  However, in these two stages, since inter-frame prediction is applied to the two macroblocks of the next identifiers n + 4 and n + 5, the processing of stage 1 of the three-stage pipeline for inter-frame prediction is started. The That is, simultaneously with processing such as intra-frame prediction image generation for the macro block with the identifier n + 2, vector reproduction processing for the macro block with the identifier n + 4 is performed. In the next stage, inter-frame prediction image generation for the macroblock and vector reproduction processing for the next n + 5 macroblock are performed.

  Hereinafter, the same processing is continued in accordance with whether inter-frame prediction or intra-frame prediction is applied to the macroblock. In FIG. 8, although not related to the decoding process of the macroblock, the return position is between the macroblocks with the identifiers n + 6 and n + 7 as described in FIG.

  In FIG. 9, it is assumed that an error has been detected in the inverse quantization / inverse transformation process for the macroblock with the identifier n in the second stage from the beginning of the figure. As described above, as error concealment processing, it is possible to obtain a higher-quality concealment result by inter-frame prediction than by intra-frame prediction. For the macroblocks up to the identifier n + 6, a three-stage pipeline process for performing inter-frame prediction is performed as long as time permits.

  That is, in the next stage, vector reproduction processing is performed on the macroblock with the identifier n + 3. Here, the macro block with the identifier n + 1 is a macro block to be subjected to the inter-frame prediction as described in FIG. 8, and the inter-frame prediction process is performed again on this macro block as described in FIG. Incurs longer processing time. For this macroblock, intra-frame prediction image generation and inverse quantization / inverse transform processing are performed at this stage. In error concealment based on intra-frame prediction, pixel information of macroblocks adjacent to the left, top, and top left is used. Since this information has a smaller amount of data than the decoded image, it does not require a memory that requires a large amount of data, such as a frame memory, but takes time to acquire data. For example, it can be stored inside the intra-frame prediction unit. Is possible. For this reason, the inter-frame prediction unit can perform intra-frame prediction from pixels of adjacent macroblocks stored immediately after an error concealment instruction by intra-frame prediction. In addition, there is no pipeline reset and no associated processing time extension.

  In the next stage, inter-frame prediction image generation for the macroblock with the identifier n + 3 and vector reproduction processing for the macroblock with n + 4 are performed, and the macroblock of n + 2 is originally applied as described in FIG. Error concealment processing is performed using power intra-frame prediction.

  In the same manner, error concealment processing using inter-frame prediction is performed for the macroblock with the identifier n + 6. As described with reference to FIG. 8, the n + 3 and n + 6 macroblocks are originally macroblocks to which intra-frame prediction is applied. However, up to the return position, that is, the macroblock with the identifier n + 6, has a large concealment effect as an error concealment process. Error concealment is performed by inter-frame prediction. In this way, after error detection, it is possible to obtain a decoded image as a high-quality error concealment result, for example, compared to performing inter-frame prediction using bitstream data as it is.

  Next, a second embodiment will be described. In the second embodiment, the number of pipeline stages is two, and the vector reproduction process and the inter-frame prediction image generation process in the inter-frame prediction are executed in one stage.

  FIG. 10 is a block diagram showing the overall configuration of the moving picture decoding apparatus according to the second embodiment. In the second embodiment, in order to perform the vector reproduction process and the inter-frame predicted image generation process in one stage, the inter-frame configured by the vector reproduction unit 21 and the inter-frame predicted image generation unit 22 of FIG. The point that an inter-frame prediction unit 31 is provided in place of the prediction processing unit 13 is basically different, and the configuration of the control unit 30 is also different from the control unit 10 of FIG. .

  FIG. 11 is a configuration block diagram of the control unit 30 in FIG. In the figure, the control unit 30 includes an error information holding unit 32, a pipeline stage 1 control unit 33, and a pipeline stage 2 control unit 34. The error information holding unit 32 receives the error notification signal from the inverse quantization unit 15 and the inverse transform unit 16 as in FIG. 3, and the pipeline stage 1 control unit 33 and the pipeline stage 2 control unit 34 according to the contents thereof. To control. The pipeline stage 1 control unit outputs an inter-frame prediction activation signal to the inter-frame prediction unit 31 in FIG. 10, and the pipeline stage 2 control unit 34 outputs an intra-frame prediction activation signal to the intra-frame prediction unit 14. In addition, a control signal for selecting an input to the adder 17 is output to the switch 18. Note that the intra-frame / inter-frame prediction determination signal from the variable-length decoding processing unit 11 is given to, for example, the pipeline stage 1 control unit 33 and transmitted to the pipeline stage 2 control unit 34.

  12 and 13 are flowcharts of pipeline control processing in the second embodiment. FIG. 12 is a flowchart of processing performed by the pipeline stage 1 control unit 33. In this stage, first, in step S21, it is determined whether or not an error has been detected in a block preceding the current macro block to be processed. If this is the case, the inter-frame prediction unit is activated as an error concealment operation in step S22. On the other hand, if no error is detected, it is determined in step S23 whether or not the macroblock is an application target of inter-frame prediction. If it is not the application target, no processing is performed. The stage 1 process is terminated. If the target is an inter-frame prediction target, the inter-frame prediction unit is activated as a normal operation in step S24.

  FIG. 13 is a flowchart of a processing example performed by the pipeline stage 2 control unit 34. In the second embodiment, the processing of stage 1 and stage 2 in the first embodiment is substantially performed only in stage 1, and the processing of stage 3 in the first embodiment is performed in the stage in the second embodiment. This corresponds to the second processing. Therefore, the processing from step S30 to step S37 in FIG. 13 is substantially the same as the processing from step S10 to step S17 in FIG. 6, and detailed description thereof will be omitted.

  14 and 15 are time charts of the moving picture decoding process in the second embodiment. Similarly to FIGS. 8 and 9, FIG. 14 is a time chart of a normal operation when no error is detected, and FIG. 15 is a time chart when an error is detected and error concealment processing is performed.

  In the first stage of FIG. 14, an inter-frame prediction unit activation instruction is issued for the macroblock with the identifier n, and an inter-frame prediction image generation process is performed. In the next stage, an inverse quantization / inverse transform process is performed on the macroblock with the identifier n, and an inter-frame prediction image generation process is performed on the n + 1 macroblock. At this time, the adding unit 17 selects an inter-frame prediction image. Hereinafter, the same processing is performed for each macroblock, but the processing is different from FIG. 8 only in that processing for inter-frame prediction is performed in only two stages, Detailed description thereof is omitted.

  In FIG. 15, as in FIG. 9, an error is detected in the inverse quantization / inverse conversion process for the macroblock with the identifier n, and thereafter the error concealment process is performed until the macroblock with the identifier n + 6 is reached. In FIG. 9, since the pipeline processing for inter-frame prediction is performed over three stages, the error concealment processing in the inter-frame prediction is performed not only for the macro block of the identifier n + 1 but also for the macro block of n + 2. In time, the error concealment process in the inter-frame prediction is performed from the macro block with the identifier n + 3, but in FIG. 15, the identifier is detected at the stage next to the stage where the error is detected for the macro block with the identifier n. The inter-frame prediction image generation process for the n + 2 macro block is possible, and the macro block that is the target of the error concealment process in the intra-frame prediction is only the macro block with the identifier n + 1.

  In the above description, the case where an error is detected for a macroblock that is the target of inter-frame prediction processing has been described as an example of the time charts in the two embodiments. Naturally, the present invention can be applied even when an error is detected with respect to a macroblock that has become defective, and all macroblocks up to the return position in the stream after error detection are subject to error concealment. On the other hand, error concealment processing by inter-frame prediction is performed for macroblocks that meet the intra-frame prediction, and error concealment processing by intra-frame prediction processing is performed on macroblocks that are not in time.

  In the above description, H.C. The embodiment has been described using the H.264 system as one example of the moving image compression coding system. The present invention is not limited to H.264, and the present invention can be applied to all encoding methods including a method that will be standardized in the future. H. This is a method in which the image quality at the same bit rate as that of the H.264 method is the same, and the compression method of Microsoft Corp. is standardized (SMPTE-421M) by the SMPTE (American Film and Television Engineers Association) VC-1 method However, it is naturally applicable.

It is a principle structure block diagram of the moving image decoding apparatus of this invention. 1 is an overall configuration block diagram of a video decoding device in a first embodiment. FIG. 3 is a configuration block diagram of a control unit in FIG. 2. It is a process flowchart of the pipeline stage 1 in a 1st Example. It is a process flowchart of the pipeline stage 2 in a 1st Example. It is a process flowchart of the pipeline stage 3 in a 1st Example. It is explanatory drawing of the error concealment object macroblock to the return position in a stream. It is a time chart of the normal decoding operation example in a 1st Example. It is a time chart of the decoding operation example at the time of error detection in the first embodiment. It is a whole block diagram of the moving image decoding apparatus in a 2nd Example. It is a block diagram of the configuration of the control unit of FIG. It is a processing flowchart of pipeline stage 1 in the 2nd example. It is a process flowchart of the pipeline stage 2 in a 2nd Example. It is a time chart of the example of normal decoding operation | movement in a 2nd Example. It is a time chart of the example of decoding operation at the time of error detection in the 2nd example. It is a structure block diagram of the prior art example of a moving image decoding apparatus. It is a time chart of the prior art example of a pipeline control decoding operation | movement. It is explanatory drawing of the pipeline reset in a prior art example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Moving image decoding apparatus 2 Intra-frame prediction means 3 Inter-frame prediction means 4 Control means 10, 30 Control part 11 Variable length decoding process part 12 Frame memory 13 Inter-frame prediction process part 14 In-frame prediction part 15 Inverse quantization part 16 Inverse Conversion unit 17 Addition unit 18 Switch 19 Deblocking filter processing unit 21 Vector reproduction unit 22 Inter-frame prediction image generation unit 25, 32 Error information holding unit 26, 33 Pipeline stage 1 control unit 27, 34 Pipeline stage 2 control unit 28 Pipeline stage 3 control unit 31 Interframe prediction unit

Claims (5)

An apparatus that generates a predicted image by dividing an image, which is stream data, into blocks, and decodes moving image data,
Intra-frame prediction means for generating an intra-frame prediction image for the processing target block;
Inter-frame prediction means for generating an inter-frame prediction image for the processing target block;
When an error is detected for a block temporally preceding the one block after the start of the inter-frame prediction process for one block, the intra-frame prediction process as an error concealment process is performed on the one block. The inter-frame prediction processing as the error concealment processing is performed within a predetermined period in a block from the block that continues in time to the one block and immediately before the start code in the stream data. The intra-frame prediction unit executes the intra-frame prediction process for a block that does not fit in the frame, and the inter-frame prediction unit performs a frame for the block in which the inter-frame prediction process as the error concealment process falls within a predetermined period. and control means for executing between prediction process
Video decoding apparatus comprising: a.   The moving picture decoding apparatus according to claim 1, wherein the control means causes the inter-frame prediction means to perform pipeline processing. The pipeline processing is performed in three stages,
3. The moving picture decoding apparatus according to claim 2, wherein the inter-frame prediction unit generates an inter-frame prediction image using the first stage and the second stage of the pipeline. The pipeline processing is performed in two stages,
3. The moving picture decoding apparatus according to claim 2, wherein the inter-frame prediction unit generates an inter-frame prediction image using the first stage of the pipeline. A method of generating a prediction image by dividing an image, which is stream data, into blocks, and decoding a moving image,
After an inter-frame prediction process for one block is started, when an error is detected for a block temporally preceding the one block, an intra-frame prediction process is executed for the one block as an error concealment process In addition, with respect to a block in which the inter-frame prediction process as the error concealment process does not fall within a predetermined period in a block from a block temporally following the one block to a block immediately before the start code in the stream data. A video decoding method comprising: executing an intra-frame prediction process, and performing an inter-frame prediction process on a block in which the inter-frame prediction process as the error concealment process falls within a predetermined period.