JP2006262406A - Encoded data generator - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect and correct syntax and semantics errors of encoded data. <P>SOLUTION: Detecting sections 100 and 101 detect syntax and semantics errors existing in encoded data and outputs at least one of error existence information, first positional information indicative of the position where an error exists, first range information indicative of a range where an error exists, and first type information indicative of encoding type at the position where an error exists as error detection information 20. Substitution information generating sections 200 and 201 select one or more substitution code sequences among one or more code sequences prepared previously based on the error detection information, and generate second positional information to be substituted of the encoded data and second range information to be substituted of the encoded data based on the substitution code sequence and the error detection information, and then generate substitution information 30 by adding the substitution code sequence to the second positional information and the second range information. A bit stream substituting section 300 substitutes an error corrected bit stream 50 for the encoded data based on the substitution information. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an encoded data generation apparatus that generates encoded data by correcting an error in encoded data encoded using motion compensated prediction.

  When transmitting or storing a moving image signal, it is often compressed and encoded into a bit stream for efficiency. MPEG (Moving Picture Experts Group, hereinafter referred to as MPEG) system has been standardized and widely used as a moving picture compression encoding system. Therefore, it is very important for the efficient use of a transmission system or a storage system to compress a moving image signal by a method such as MPEG and handle it as a bit stream.

  By the way, if a syntax (encoded data string rule) error occurs in compressed encoded data that has been compression-encoded according to an encoding standard with compression as described above, compressed encoded data in which a syntax error has occurred. Compressed encoded data that follows can not be decoded by a decoder that does not have a function capable of avoiding errors. Here, the syntax error is an error in which parsing (decomposing syntax elements) of compression-encoded data cannot be continued. For example, in MPEG and the like, a syntax error is detected when a value that is not in the VLC (variable length coding) table is detected in the parsing process, or when the number of blocks in the macroblock does not match a specified value, Detected when the number of macroblocks exceeds the specified value. A syntax error is corrected by an error correction process such as Reed-Solomon when an error is inherent in a compression encoding device, encoding conversion device, or multiplexing device, or an error that occurs during the transmission or storage of compressed encoded data This may occur when it is not possible or when packet loss occurs.

  When a syntax error is detected, the compression encoding apparatus performs the following operation in order to continue decoding. Since the synchronization code is inserted in the encoded data with general compression, the operation is such that the data block in which the syntax error is detected is discarded, and the next data block is acquired and decoded. Do. As a processing method when a syntax error occurs as described above, a process called error concealment is performed.

  In addition, in the case of an error in semantics (content semantic content description tool) in compressed encoded data that has been compression-encoded in accordance with an encoding standard with compression as described above, the encoded encoded data cannot be encoded correctly. Generally not possible. In addition, depending on the semantic error, a fatal failure may occur in the decoding device. Semantic error is not an error that occurs in parsing compression-encoded data, but an error that occurs because the value obtained by parsing violates the restriction defined in the standard. For example, in MPEG, which is one of the moving image compression coding standards, there is a semantic error that a certain syntax element value is defined as a fixed value but not a fixed value. In MPEG-4 AVC, which is the latest moving image compression coding standard, restrictions are defined in detail by profile and level. For this reason, there is a semantic error in which compressed encoded data that conforms to the standard in the baseline profile level 2 is in violation of the standard in the baseline profile level 3.

  Semantic errors may occur when encoded by an incorrect interpretation in the compression encoding device and encoding conversion device, and may occur in the process of storing, transmitting, and acquiring compressed encoded data. In addition, when users who are not familiar with profiles and levels of video compression coding standards perform data interoperability among many models, there is a high possibility that semantic errors will occur due to misuse. . In addition, although it is not a semantic error, there are cases where it is necessary to limit the encoding standard in the operation of the system. An error that does not satisfy the predetermined encoding condition on the system cannot be handled by the corresponding system. As described above, a process called error concealment is performed as a processing method when a semantic error occurs.

  As a conventional system using such an error concealment technique, the system described in Patent Document 1 below can perform appropriate packet discard compensation according to the nature of an image and improve the image quality of a decoded image. The moving image transmission is realized. In other words, this is a system that detects packet discard of an encoded video packet, generates a decoded signal by concealment by replacing an error signal in the discarded packet with an already decoded image signal. Therefore, in the system described in Patent Document 1, the input is a bit stream that may include an error, and the output is a decoded image signal that has been corrected for the error when a bit stream that includes an error is input.

  As a conventional system using another error concealment technique, the device described in Patent Document 2 below does not require a separate disk device for error correction, and is a disk array device that takes advantage of the characteristics of video data. The video transmission provided is realized. That is, the input video data is divided and stored as a bit stream, and when an error occurs in the input video data bit stream, a concealment bit stream that is highly compressed in advance from the input video data is stored in advance. In this case, the error bit stream of the input video data is replaced by decoding and re-encoding the data bit stream. Therefore, in the apparatus of Patent Document 2, an input is a bit stream in which an error may occur, and an output is a bit stream in which the error is corrected when an error occurs.

Further, as a conventional system using the error concealment technique, the circuit described in Patent Document 3 below uses a normal line immediately after encoding in a facsimile apparatus that requires encoding conversion in real time. When a code error is detected by a code buffer that can temporarily store the encoded information for one minute, concealment is performed by outputting the encoded information for one normal line in the buffer.
Japanese Unexamined Patent Publication No. 7-111654 (Abstract) Japanese Patent No. 3482971 (Claims) JP 63-292873 A (Claims)

  However, the system or apparatus using the error concealment technique described above has the following problems. In the moving image transmission system in Patent Document 1, since the output after error concealment is a decoded image signal, an error remains in the bitstream itself, and as a result, what is stored in a storage medium or the like is transferred to another medium. There is a problem that the error propagates.

  In the video transmission device in Patent Document 2, the output is a bit stream of input video data that has been error concealed, but a bit stream for concealment is prepared in a storage medium for each bit stream of the input video data. It is necessary to increase the bit stream of the input video data and increase the data of the concealment bit stream, and it cannot be said that the storage medium is effectively utilized. In the error replacement circuit of a facsimile in Patent Document 3, the previous one normal line is stored as coding information for concealment. However, in the compression-coded bit stream of a moving image using motion compensation predictive coding, If the concealment is performed only with the encoded information, the quality of the reproduced image is greatly deteriorated.

  In view of the problems of the conventional example, the present invention provides an encoded data generation apparatus capable of detecting and correcting syntax and semantic errors in encoded data encoded using motion compensated prediction. With the goal.

In order to achieve the above object, the present invention receives encoded data encoded using motion compensated prediction, and detects whether there is a syntax error in the input encoded data. , Detecting whether there is a semantic error in the encoded data, the presence information of the error, the first position information indicating the position where the error exists, the first indicating the range where the error exists An error detection unit that outputs at least one of the range information and the first type information indicating the encoding type at the position where the error exists, as error detection information;
Based on the error detection information, one or more replacement code strings are selected from one or more code strings prepared in advance, and the encoded data is replaced based on the selected replacement code string and the error detection information. Creating second position information indicating a position to be replaced and second range information indicating a range to be replaced of the encoded data, and selecting the selected replacement in the generated second position information and second range information A replacement information generation unit for generating replacement information with the addition of a code string;
A replacement unit that replaces the encoded data with encoded data in which the detected error is corrected based on the replacement information of the replacement information generation unit;

  Further, the error detection unit detects whether there is an error that does not satisfy a predetermined encoding condition in encoded data in the error detection of the semantics, and the presence information of the error that does not satisfy the predetermined encoding condition, Position information indicating a position where an error not satisfying the predetermined encoding condition exists, range information indicating a range where an error not satisfying the predetermined encoding condition exists, and an error not satisfying the predetermined encoding condition exist At least one or more of classification information indicating an encoding type at the position to be output is output as semantic error detection information.

  According to the present invention, even when there is a syntax and / or semantic error in the received encoded bitstream, the error is verified, corrected, and stored in the storage medium. It is possible to copy and edit an error-free bitstream and reproduce it on a client-side decoder without an error concealment function without propagation of an error in transmission. In addition, it is not necessary to prepare error correction information from the input bit stream in advance, and the storage system can perform error correction without requiring data storage for concealment.

<First Embodiment>
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a transmission system in which a server is used as the encoded data generation apparatus according to the first embodiment of the present invention, and FIG. 2 is a block diagram showing in detail the corrector of FIG.

  The transmission system in FIG. 1 shows a connection example of a bit stream generation device 500, the encoded data server 800 according to the first embodiment of the present invention, and a playback device 700 that is a client. The bit stream generation device 500 converts the image information in the storage medium 510 into an encoded bit stream by the encoder 520 and transmits it to the transmission line 1001 by the transmitter 530. The encoded data server 800 receives the bit stream once at the receiver 810, corrects the bit stream if there is an error by the corrector 820, and stores the corrected bit stream in the storage medium 350. The encoded data server 800 transmits the bit stream stored in the storage medium 350 to the client-side playback device 700 via the transmission path 1002 by the transmitter 840 as necessary. The playback apparatus 700 on the receiving side receives the bit stream by the receiver 710 and decodes it by the decoder 720.

  In this path, assuming that the bit stream transmitted from the bit stream generation device 500 has some error, the encoded data server 800 according to the present invention verifies and corrects the error of the bit stream by the corrector 820. By storing in the storage medium 350, it is possible to prevent an illegal bit stream from being recorded on the storage medium 350 due to a transmission path error occurring in the transmission path 1001 between the bit stream generation device 500 and the encoded data server 800. Is possible. Therefore, it is possible to prevent propagation of an error when transmitting a bit stream to the playback apparatus 700.

<Syntax error correction section>
The corrector 820 includes a syntax error correction unit and a semantic error correction unit. FIG. 2 shows a block diagram of the syntax error correction unit in the first embodiment. As shown in FIG. 2, the syntax error correction unit includes a syntax error detection unit 100, a replacement information generation unit 200, and a bitstream replacement unit 300. However, hereinafter, the compression-encoded bit stream is simply referred to as a bit stream.

  The overall operation of the syntax error correction unit of the corrector 820 will be described with reference to FIG. First, the bit stream 10 is input to the syntax error detection unit 100 as input. The syntax error detection unit 100 detects whether a syntax error is included. The presence / absence of a syntax error and the occurrence position of the syntax error are output as syntax error detection information (hereinafter simply referred to as detection information) 20. The replacement information generation unit 200 selects a replacement code string for correcting a syntax error of the input bitstream 10 from several code strings prepared in advance based on the detection information 20. From the selected replacement code string and the detection information 20, the position, range, and replacement code string of the input bitstream 10 to be replaced are determined and output as replacement information 30. However, the replacement information 30 may be obtained by some external input 40. The bitstream replacement unit 300 outputs a bitstream 50 in which a syntax error is concealed using the input bitstream 10 and the replacement information 30, and records the bitstream 50 in the storage medium 350.

  The operations of the syntax error detection unit 100 and the replacement information generation unit 200 will be described in detail. First, the flowchart of the syntax error detection unit 100 is shown in FIG. 3, and the operation will be described (steps S10 to S15). The syntax error detection unit 100 first parses the input bitstream 10 (step S10), and detects a syntax error of the input bitstream while parsing (step S11).

  In the syntax error detection determination (step S12), when the input bitstream 10 includes a syntax error, a flag indicating that the syntax error exists, and position information of the bitstream 10 in which the syntax error has occurred When processing that can parse the bitstream from the reverse, such as Reversible VLC, is used, range information in which a syntax error has occurred, for example, in the MPEG-4 AVC standard, access unit unit, slice unit, macroblock Type information indicating the occurrence of a syntax error in units and the coding type such as intra, inter, etc. is generated (step S13) and output as detection information 20 (step S15). On the other hand, if the input bitstream 10 does not include a syntax error, a flag indicating that there is no syntax error is generated (step S14) and output as detection information 20 (step S15).

  In steps S11 and S12, for example, when several macroblocks are continuously missing, the number of macroblocks relative to the picture size is fixed. If they do not match, it is determined that there is a syntax error. Therefore, the detection information 20 in this case is type information indicating that an error has occurred in the macro block, and position information where the macro block is missing. That is, the address of the location where the error occurred, a bit stream such as Reversible VLC, etc. If processing that can be parsed from the opposite is used, the address range information of the missing macroblock is obtained.

  The replacement information generation unit 200 will be described. The flowchart of the replacement information generation unit 200 is shown in FIG. 4 and the operation will be described (steps S20 to S21). From the detection information 20, the replacement information generation unit 200 selects a replacement code string used for replacement from the code strings prepared in advance from the type and position of the syntax error occurring in the input bitstream 10 (step S20). . Based on the selected replacement code string and the detection information 20, the position and range where the input bit stream 10 is to be replaced are determined, and are generated as replacement information 30 together with the selected replacement code string (step S21).

  For example, the type information of the detection information 20 is a macro block, some macro blocks are missing, the macro block type before and after the missing macro block is Not Coded, and dmv (differential motion vector) is 0. In this case, the macroblock type of the missing part is similarly determined as Not Coded and dmv is 0, and becomes a replacement code string of the macroblock in which the flag indicating Not Coded of the prepared macroblock is missing. In this case, the number of macroblocks with missing replacement code strings is generated as replacement information 30.

  For example, when the type information of the detection information 20 is an access unit and some access units are missing, a replacement code string of an access unit lacking a NoMC / Not Coded flag is obtained. In this case, the number of missing access units in the replacement code string is generated as replacement information 30. In addition, in order to match the code amount of the input bitstream 10, a stuffing byte can be added to the replacement code string of the replacement information 30. Here, in the selection of the replacement code string, the replacement code string may be selected by the external input 40. By controlling with the external input 40 in this way, it is possible to prevent an unintended reproduced image from being output.

  The bitstream replacement unit 300 will be described. The bitstream replacement unit 300 uses the replacement information 30 generated as described above to replace a portion to be replaced in the input bitstream 10 with a replacement code string of the replacement information 30, thereby replacing a replacement bit without a syntax error. A stream 50 is generated. The generated bit stream 50 is recorded on the storage medium 350.

  As described above, in the present embodiment, when there is a syntax error in a motion-coded video stream that has been predicted for motion compensation, an error concealment is prepared by preparing a code string for replacing an error in advance. Compared with the case where the decoded image is re-encoded for the purpose of the image processing, a high-speed error concealment process can be realized. In addition, compared with the case where the previous normal code string is simply used as the code string for concealment, the quality of the reproduced image can be improved in the bit stream after the error concealment. In addition, since errors are corrected before recording on the storage medium, it is possible to handle streams without error propagation in copying and editing bitstreams.

<Reference example of syntax error correction section>
FIG. 5 shows a block diagram of a reference example of the syntax error correction unit. As shown in FIG. 5, the syntax error correction unit of the reference example includes a syntax error detection unit 100, a decoding unit 400, a replacement information generation unit 200, and a bitstream replacement unit 300. Hereinafter, the compression-encoded bit stream is simply referred to as a bit stream. In FIG. 5, first, the bit stream 10 is input to the syntax error detection unit 100 as an input. The syntax error detection unit 100 detects whether a syntax error is included, and outputs the presence / absence of the syntax error, the occurrence position of the syntax error, and the like as detection information 20.

  The decoding unit 400 decodes the input bitstream 10. The header information generated at the time of decoding is output as the encoding parameter 60, and the decoded image is output as the decoded image 70. The replacement information generation unit 200 selects a replacement code string for correcting a syntax error of the input bitstream 10 from several code strings prepared in advance based on the detection information 20. From the selected replacement code string and the detection information 20, the position, range, and replacement code string of the input bitstream 10 to be replaced are determined and output as replacement information 30. However, the replacement information 30 may be obtained by some external input 40. The bit stream replacement unit 300 outputs a bit stream 50 without a syntax error using the input bit stream 10 and the replacement information 30. The generated bit stream is recorded on the storage medium 350.

  Next, each component will be described. The syntax error detection unit 100 will be described. The operation of the syntax error detection unit 100 is the same as that of the first embodiment, except that the detection information 20 is sent to the decoding unit 400. That is, for example, when several macroblocks are continuously missing, the number of macroblocks with respect to the picture size is fixed, so when the number of macroblocks does not match when one picture is parsed. It is determined that it is a syntax error. Therefore, the detection information 20 in this case is type information indicating that an error has occurred in the macroblock, and position information where the macroblock is missing, that is, the address of the location where the error has occurred and a bit stream such as Reversible VLC. When processing that can be parsed from the opposite is used, the address range information of the missing macroblock is obtained.

  The flowchart of the decoding unit 400 is shown in FIG. 6 and the operation will be described (steps S40 to S44). The decoding unit 400 decodes the input bitstream 10 (step S40). At this time, it is determined whether or not the decoded image 70 can be generated from the error occurrence position information, the range information, and the type information of the detection information 20 (step S41). If the decoded image 70 cannot be generated due to the occurrence of a syntax error, error concealment is performed on the location where the error has occurred (step S44). Error concealment is performed by interpolating the missing pixel by using the pixel values of the surrounding image. For example, if a macroblock is missing, it is interpolated from the pixel values around the macroblock, or if it is missing in units of access units, it is interpolated with a duplicate image from the previous image, etc. Perform error concealment using the method. An image subjected to error concealment is generated as a decoded image 70 (step S43). If a decoded image can be generated, the decoded image 70 is output.

  Also, the decoding unit 400 can generate not only the decoded image 70 but also an encoding parameter 60 generated at the time of decoding as additional information in a portion where there is no syntax error (step S42). Examples of the encoding parameter 60 generated at the time of decoding include information that can be used for simplification of re-encoding, such as sequence and picture header information, motion vectors, and macroblock types.

  Subsequently, the flowchart of the replacement information generation unit 200 is shown in FIG. 7 and the operation will be described (steps S50 to S52). Based on the detection information 20, the replacement information generation unit 200 performs re-encoding using the decoded image 70 when the above-described encoding parameter 60 is not used (step S51). When the encoding parameter 60 is used, a re-encoding parameter is generated based on the encoding parameter 60 (step S50). The re-encoding parameter here is a position or range of the decoded image 70 to be re-encoded. Further, in order to simplify the re-encoding of the decoded image 70 having no syntax error, when the encoding parameter 60 such as header information and motion vector is used, it is based on the encoding parameter 60 output by the decoding unit 400. Thus, the re-encoding parameter can be obtained and used. The decoded image 70 of the input bitstream 10 is re-encoded using the re-encoding parameter (step S51). Here, the re-encoding parameter may be determined by the external input 40. As described above, the re-encoded bit stream is used as the replacement code string, and the replacement information 30 including the position information and the range information of the input bit stream 10 to be replaced is generated from the replacement code string and the detection information 20 ( Step S52). The operation of the bitstream replacement unit 300 is the same as that in the first embodiment.

  As described above, when there is a syntax error in the motion-compensated predicted moving image encoded bitstream, re-encoding is performed using the decoded image. Therefore, the motion-compensated prediction of the portion where the syntax error has occurred is performed. Recalculation is possible. Therefore, compared to a method in which a replacement code string is prepared in advance or a method in which error concealment is performed by simply copying the previous normal code string, the quality of the reproduced image of the bitstream after error concealment is improved. It is possible to improve. In addition, since errors are corrected before they are recorded on the storage medium, the streams can be handled without error propagation in bitstream copying and editing. As described above, according to the reference example of the syntax error correction unit illustrated in FIG. 5, even when there is a syntax error in the received encoded bitstream, the error is verified, corrected, and stored in the storage medium 350. Therefore, an error does not propagate in transmission from the storage medium 350 to another medium, and an error-free bitstream can be copied and edited and reproduced by a decoder that does not have an error concealment function. In addition, it is not necessary to prepare error correction information from the input bitstream 10 in advance, and the error correction is possible in the storage system without needing to store data for concealment.

<Semantics error correction section>
FIG. 8 is a block diagram of the semantic error correction unit of the corrector 820 according to the first embodiment. The semantic error correction unit includes a semantic error detection unit 101, a replacement information generation unit 201, and a bitstream replacement unit 300. Hereinafter, the compression-encoded bit stream is simply referred to as a bit stream. The overall operation will be described with reference to FIG. First, the bit stream 10 is input to the semantic error detection unit 101 as an input. The semantic error detection unit 101 detects whether or not a semantic error is included, and outputs the presence / absence of the semantic error, the occurrence position of the semantic error, etc. as semantic error detection information (hereinafter simply referred to as detection information) 21.

  The replacement information generation unit 201 selects a replacement code string for correcting a semantic error of the input bitstream 10 from several code strings prepared in advance based on the detection information 21. The position, range, and replacement code string of the input bit stream 10 to be replaced are determined from the selected replacement code string and the detection information 21 and output as replacement information 31. However, the replacement information 31 may be obtained by some external input 41. The bitstream replacement unit 300 outputs the bitstream 50 in which the semantic error is concealed using the input bitstream 10 and the replacement information 31. The generated bit stream 50 is recorded on the storage medium 350.

  The semantic error detection unit 101 will be described. The flowchart of the semantic error detection unit 101 is shown in FIG. 9, and the operation will be described (steps S10 to S15). The semantic error detection unit 101 first parses the input bitstream 10 (step S10), and detects a semantic error of the input bitstream 10 while parsing (step S11a). In the detection determination of the semantic error (step S12), when the input bitstream 10 includes a semantic error, a flag indicating that there is a semantic error, the position information of the bitstream 10 in which the semantic error has occurred, the Reversible VLC When processing that can parse the bitstream from the reverse side is used as described above, range information where a semantic error has occurred, for example, in the MPEG-4 AVC standard, a semantic error occurs in units of access units, slices, and macroblocks. In addition, type information indicating an encoding type such as intra or inter is generated (step S13), and is output as detection information 21 (step S15). When the input bitstream 10 does not include a semantic error, a flag indicating that there is no semantic error is generated (step S14) and output as detection information 21 (step S15).

  For example, in the MPEG-4 AVC encoding method, if the profile value of the sequence parameter set of the stream including B-Slice is a value indicating Baseline Profile, a semantic error occurs, and the unit of the sequence parameter set is the type information. The address where the sequence parameter set exists is position information.

  The replacement information generation unit 201 will be described. The flowchart of the replacement information generation unit 201 is shown in FIG. 10, and the operation will be described (steps S20 to S21). From the detection information 21, the replacement information generation unit 201 selects a code string used for replacement from among the code strings prepared in advance from the type and position of the semantic error occurring in the input bitstream 10 (step S20). Based on the selected code string and the detection information 20, a position and a range to be replaced in the input bit stream 10 are determined, and are generated as replacement information 31 together with the selected code string (step S21).

  For example, in the MPEG-4 AVC encoding method, when the profile value of the sequence parameter set of a stream including B-Slice is a value indicating Baseline Profile, a value indicating Main Profile in the profile value as a replacement code string Is selected as the replacement code string of the replacement information 31. Further, in order to match the code amount of the input bit stream 10, a stuffing byte can be added to the replacement code string of the replacement information 31. Here, when selecting a replacement code string, the replacement code string may be selected by the external input 41.

  Also, a method may be considered in which the semantic error detection unit 101 does not detect a semantic error, and the replacement information generation unit 201 always replaces the same syntax of the input bitstream 10 using the same code string. For example, it is assumed that the decoder on the receiving side has a restriction that a bit stream having a level value higher than a certain level is not reproduced by looking at the level value in the header of the bit stream. However, it is assumed that the decoder has a capability of reproducing an image size that can be originally reproduced only at a level value equal to or higher than a certain level in the coding standard. At this time, if only the level value is always rewritten below a certain level, the decoder can reproduce the bit stream. Therefore, when the semantic error detection unit 101 does not detect a semantic error, the replacement information generation unit 201 prepares a level value supported by this decoder as a replacement code string, and when a bitstream is input. The bit stream can be reproduced by the decoder simply by replacing the level value of the header of the bit stream with the code string of the prepared level value.

  The bitstream replacement unit 300 uses the replacement information 31 generated as described above to replace the place to be replaced in the input bitstream 10 with the replacement code string of the replacement information 31, thereby preventing the bitstream 50 without a semantic error. Can be generated. The generated bit stream 50 is recorded on the storage medium 350.

  As described above, according to the present embodiment, when there is a semantic error in the motion compensated predicted moving image encoded bitstream, an error concealment is prepared by preparing a code string for replacing an error in advance. Compared with the case where the decoded image is re-encoded for the purpose of the image processing, a high-speed error concealment process can be realized. In addition, compared with the case where the previous normal code string is simply used as the code string for concealment, the quality of the reproduced image can be improved in the bit stream after the error concealment. Further, since the error is corrected before recording in the storage medium 350, it is possible to handle the stream without propagation of the error in copying or editing the bit stream.

<Reference example of the semantic error correction section>
FIG. 11 shows a block diagram of a reference example of the semantic error correction unit. The semantic error correction unit of the reference example includes a semantic error detection unit 101, a decoding unit 401, a replacement information generation unit 201, and a bitstream replacement unit 300. Hereinafter, the compression-encoded bit stream is simply referred to as a bit stream. The overall operation will be described with reference to FIG. First, the bit stream 10 is input to the semantic error detection unit 101 as an input. The semantic error detection unit 101 detects whether or not a semantic error is included, and outputs the presence / absence of the semantic error, the occurrence position of the semantic error, and the like as detection information 21.

  The decoding unit 401 decodes the input bitstream 10. The header information generated at the time of decoding is output as the encoding parameter 60, and the decoded image is output as the decoded image 70. The replacement information generation unit 201 selects a replacement code string for correcting a semantic error of the input bitstream 10 from several code strings prepared in advance based on the detection information 21. The position and range replacement code string of the input bit stream 10 to be replaced are determined from the selected replacement code string and the detection information 21 and output as replacement information 31. However, the replacement information 31 may be obtained by some external input 41. The bit stream replacement unit 300 outputs a bit stream 50 without a semantic error using the input bit stream 10 and the replacement information 31. The generated bit stream 50 is recorded on the storage medium 350.

  Each component will be described. The operation of the semantic error detection unit 101 is the same as that of the first embodiment. For example, when the motion vector exceeds a value restricted by the encoding method level, the unit of the motion vector becomes type information, and the macro block address where the motion vector exists becomes position information. The operation of the decoding unit 401 is almost the same as that of the decoding unit 400 shown in FIG. 5, and the operations of the replacement information generating unit 201 and the bitstream replacement unit 300 are the same as those in the first embodiment.

  As described above, when there is a semantic error in the motion-compensated motion-coded bitstream, re-encoding is performed using the decoded image. Is possible. Therefore, compared to a method in which a replacement code string is prepared in advance or a method in which error concealment is performed by simply copying the previous normal code string, the quality of the reproduced image of the bitstream after error concealment is improved. It is possible to improve. Further, since the error is corrected before recording in the storage medium 350, the stream can be handled without error propagation in the copying or editing of the bit stream 50.

<Second Embodiment>
In the second embodiment, the semantic error detection in the first embodiment is applied, for example, when a restriction is imposed on the coding conditions of the standardized coding standard in system operation. For example, the semantic error detection unit 101 operates as a restriction semantic error detection unit, and detects whether a certain system restriction is satisfied for an input bitstream. If not satisfied, the same detection information 20 as in the first embodiment is output. As for the following operations, semantic errors are read as errors that do not satisfy a predetermined encoding condition, and operations similar to those in the first embodiment are performed.

  In this embodiment, even if the standardized encoding method is satisfied, a bitstream having an error that does not meet the standard is corrected by adding a more detailed restriction to the restriction of the encoding method for operational reasons. can do. For example, a bit stream that satisfies the MPEG-2 standard but does not satisfy the reserved bit value of the MPEG-2 bit stream restricted by a certain system cannot be handled by the corresponding system. Since the reserved bit value in the bit stream is corrected so as to satisfy the restrictions of the corresponding system, it is possible to generate a bit stream that can be handled by the corresponding system.

  Further, for example, it is assumed that there is a bit stream encoded by Open GOP that satisfies the MPEG-2 standard. Suppose that there is a restriction that an MPEG-2 bitstream that is restricted in some systems must be encoded with a Closed GOP. In this case, it cannot be reproduced by the decoder of the corresponding system, but according to the present embodiment, the MPEG-2 GOP is encoded in the I-picture in the encoding order so as to satisfy the restrictions of the corresponding system. The first B-picture immediately after is changed from bidirectional prediction to unidirectional prediction and re-encoded. Thus, since the generated bit stream satisfies the Closed GOP, it is possible to generate a bit stream that can be reproduced by the decoder of the corresponding system. Of course, two error detectors, a syntax error detector and a semantic error detector, may be provided so that the two errors can be corrected simultaneously.

1 is a block diagram illustrating a transmission system in which a server is used as an encoded data generation apparatus according to a first embodiment of this invention. It is a block diagram which shows the syntax error correction part of the corrector of FIG. 1 in detail. It is a flowchart which shows the process of the syntax error detection part of FIG. It is a flowchart which shows the process of the replacement information generation part of FIG. FIG. 3 is a block diagram illustrating a reference example of the syntax error correction unit in FIG. 2. It is a flowchart which shows the process of the decoding part of FIG. It is a flowchart which shows the process of the replacement information generation part of FIG. It is a block diagram which shows the semantic error correction part of the corrector of FIG. It is a flowchart which shows the process of the semantic error detection part of FIG. It is a flowchart which shows the process of the replacement information generation part of FIG. It is a block diagram which shows the reference example of the semantic error correction part of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Syntax error detection part 101 Semantic error detection part 200, 201 Replacement information generation part 300 Bit stream replacement part 350, 510 Storage medium 400, 401 Decoding part 500 Bit stream generation apparatus 520 Encoder 530, 840 Transmitter 700 Playback apparatus 710, 810 Receiver 720 Decoder 800 Encoded data server 820 Modifier 1001, 1002 Transmission path

Claims (2)

Encoded data encoded using motion compensated prediction is input, and it is detected whether there is a syntax error in the input encoded data, and there is a semantic error in the encoded data. Whether the error exists, first position information indicating a position where the error exists, first range information indicating a range where the error exists, and a position where the error exists An error detection unit that outputs at least one or more of the first type information indicating the encoding type in the error detection information;
Based on the error detection information, one or more replacement code strings are selected from one or more code strings prepared in advance, and the encoded data is replaced based on the selected replacement code string and the error detection information. Creating second position information indicating a position to be replaced and second range information indicating a range to be replaced of the encoded data, and selecting the selected replacement in the generated second position information and second range information A replacement information generation unit for generating replacement information with the addition of a code string;
A replacement unit that replaces the encoded data with encoded data in which the detected error is corrected based on the replacement information of the replacement information generation unit;
An encoded data generation apparatus provided. In the error detection of the semantics, the error detection unit detects whether there is an error that does not satisfy the predetermined encoding condition in the encoded data, and the presence information of the error that does not satisfy the predetermined encoding condition, Position information indicating a position where an error not satisfying the encoding condition exists, range information indicating a range where an error not satisfying the predetermined encoding condition exists, and a position where an error not satisfying the predetermined encoding condition exists The encoded data generation apparatus according to claim 1, wherein at least one or more of classification information indicating an encoding type is output as semantic error detection information.

Images