JPH1041921A - Decoder and coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent propagation of an undesired correction disable flag by specifying an produced position of a code error in a variable length code. SOLUTION: An error correction decoding means 12 applies error check and correction in the unit of a symbol word length and provides an output of a correction disable flag when correction is disabled because of a code error in excess of correction capability. A variable length decoding means 13 applies variable length decoding to correction data to obtain decoded data so that a correction disable flag in the unit of a symbol word length specifys a code part expressed by a variable length or a level of a fixed length denoted by the code part in the code part and the level part configured by the variable length code and when a correction disable flag is detected in the code part, the decoding disabled flag is added up to final decoding data of a block after the decoded data and provides an output and when the correction disabled flag is detected by the level part only, the decoding disabled flag is added only to the decoding data and the result is outputted. An error correction means 14 applies error correction to the decoded data to obtain correction data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a coding and decoding method for compressing and expanding coefficient data using a statistical property. The present invention relates to an encoding apparatus and a decoding apparatus including means for reducing the quality deterioration of decoded data by accurately specifying and correcting the same.

[0002]

2. Description of the Related Art When recording a large amount of coefficient data represented by image information, the capacity of a recording medium is limited.
From the viewpoint of effective use of a medium, it is desired to perform high-efficiency encoding on image information and record at a lower recording rate. However, performing high-efficiency coding increases the density of image information per code, and increases the effect of a code error on a recording medium on the entire image. In particular, when processing is performed in units of adjacent pixel blocks as a high-efficiency encoding means and variable-length encoding is performed, the codeword cannot be properly decoded by an error of only one bit.
The error propagates until the synchronization of the variable length code is restored again.
Pixel loss due to a code error occurs in blocks over adjacent pixels, causing visually significant image loss and image quality deterioration. Therefore, when performing high-efficiency coding, it is necessary to take measures against code errors in the recording medium by using an error correction code or the like.

However, the correction capability and the redundancy of the error correction code are in a trade-off (reciprocal) relationship, and the correction capability cannot be blindly increased in consideration of the coding efficiency. Further, it is very difficult to completely correct a burst error caused by dropout or head clog (clogging) as in a magnetic recording medium even by using an error correction code. Even from such characteristics of the recording medium, some measures are required to prevent the reproduced image from being broken even when an uncorrectable code error occurs.

Therefore, it is impossible to completely restore a pixel which has been lost due to an uncorrectable code error, but an attempt has been made to correct the occurrence of a code error by a decoding device so as not to be visually detected. . For example, JP-A-7-950
A decoding device disclosed in Japanese Patent Application Publication No. 95 is a decoding device that collects sample values of an input signal into blocks, and decodes information data encoded in block units. And a means for correcting the information data in units of large blocks in which the plurality of blocks are collected according to the output of the detection means. In the conventional decoding device configured as described above, the occurrence of an uncorrectable code error even by the error correction code is detected by the uncorrectable flag, and the image is reproduced only until the variable-length code regains synchronization. By prohibiting data update, the operation is performed so as to replace image data that has been decoded without error in the past.

[0005]

However, in the above-described configuration, updating of a plurality of image blocks included in a sync block after a position where an uncorrectable error is detected regardless of the state of a code error is stopped. Therefore, the effect on code errors in a still image having a high correlation with the past image is recognized, but the still image of a corrected image block in a moving image, especially in a fast-moving image, includes image blocks of different times on the same screen. And the image quality is significantly degraded. In order to improve the quality of corrected images in moving images, it is necessary to limit the propagation range of code errors as much as possible without relying only on past images with low correlation, and to maximize the use of codes that can be decoded without errors It is.

[0006] The ability to detect and correct a code error by an error correction code is proportional to the redundancy of a parity added for error detection and correction. It is generally known that this function works very effectively when the error rate of a codeword is sufficiently smaller than the correction capability limit. However, as the error rate increases, an uncorrectable flag is added to a code that is not erroneous. Erroneous detection of a code error becomes a problem. Further, it is also known that when the correction is performed to the limit of the error correction capability, the probability of the erroneous correction that is erroneously corrected increases. If an erroneous correction occurs, the uncorrectable flag is lost, so that it becomes impossible to perform image correction using the uncorrectable flag.

As a result, the propagation of an uncorrectable flag higher than the actual code error rate due to an erroneous detection of a code error or a code error that cannot be detected due to an erroneous correction causes a significant deterioration in image quality over a wide range.

In view of the foregoing, the present invention limits the propagation range of a code error as much as possible by utilizing the properties of an error correction code and the properties of a variable length code, and maximizes the use of codes that can be decoded without errors. By doing so, it is an object of the present invention to provide an encoding device and a decoding device capable of improving a code error correction capability for decoded data and obtaining stable decoded data.

[0009]

A decoding device according to the present invention encodes a coefficient data sequence to be compressed using a statistical property into a variable length code for each block including a plurality of the coefficient data. Encoding apparatus comprising: a variable-length encoding unit that converts encoded data into encoded data; and an error-correcting encoding unit that performs error-correction encoding on the encoded data in units of a predetermined symbol word length to generate transmission data. A transmission device which receives the transmission data output by the decoding device and obtains the decoded data by performing the reverse processing of the error correction encoding and the variable length encoding,
When an uncorrectable flag is generated due to a code error exceeding the correction capability of the error correction decoding means, the occurrence position of the uncorrectable flag in symbol word length units is determined by the variable portion of the code portion and the level portion constituting the variable length code. It specifies whether it is in a code part represented by a length or in a fixed-length level part indicated by the code part. If an uncorrectable flag is detected in the variable-length code part, an error in the codeword length due to a code error may cause an error to propagate after the variable-length code. After that, a decoding impossible flag is added to the last decoded data of the block and output. On the other hand, when the uncorrectable flag is detected only in the fixed-length level portion, the code error is completed only by the decoded data, and thus the uncorrectable flag is added only to the decoded data and output. Therefore, it acts to limit the range of bit error propagation. Even if an unavoidable code error occurs during the code recording / reproduction or data transmission process, by accurately specifying the position of the occurrence of the code error in the variable length code, unnecessary propagation of the uncorrectable flag is prevented, The decoded data that can be decoded without errors can be used as effectively as possible. Further, even when there is no code error flag such as error correction by an error correction code and it is not possible to detect in advance that the code is incorrect before decoding, the variable length code is used as a kind of error detection code and is not included in the code word table. By combining error detection using one or more variable-length codes, such as error detection using irregular code words, error detection using the number of decoded data, error detection using the end detection bit, and error detection using the decoding code length, A code error can be detected well. Therefore, instead of error correction relying only on past decoded data having a low correlation, it is possible to correct block-complete errors by increasing the number of decoded data that can be correctly decoded even for a block containing an erroneous code. Therefore, the quality of error correction in a time-varying coefficient data sequence such as a moving image can be significantly improved. In addition, since the error detection capability can be enhanced, it is possible to prevent significant information degradation due to decoding of erroneously decoded data as it is.

[0010]

1 is a block diagram showing a configuration of a decoding apparatus according to a first embodiment of the present invention; Input the transmission data that has been subjected to error correction coding in units of a predetermined symbol word length for the coded data, and perform the reverse processing of the error correction coding and the variable length coding. A decoding device for obtaining decoded data, comprising detecting and correcting errors in the transmission data in units of symbol word lengths, outputting corrected data, and adding a non-correctable flag in the case of a code error exceeding the correction capability. And an error correction decoding means for outputting the decoded data in a variable length manner when the corrected data is subjected to variable length decoding and outputting the decoded data. To be If the uncorrectable flag is detected in the code part by specifying whether the code part is in the fixed-length level part indicated by the code part, A variable-length decoding means for adding a non-decodable flag to only the decoded data and outputting it when the uncorrectable flag is detected only in the level portion, and outputting the uncorrectable flag only to the decoded data; Error correcting means for performing error correction on the decoded data to which the non-decoding flag is added using the decoded data and the non-decoding flag to obtain corrected data.

In the decoding apparatus according to the first aspect, when an uncorrectable flag is generated in the code part, the delimitation of the code word becomes unclear due to the code error, and the error is highly likely to propagate to the code and thereafter. Therefore, after that, including the decoded data, a decoding impossible flag is added to the last decoded data of the block and output. If an uncorrectable flag occurs only in the level part, it can be determined that the reliability of the fixed-length data is low but the synchronization of the variable-length code is not lost, so the undecryptable flag is added only to the decoded data. Output. As a result, the uncorrectable flag in symbol word length units is divided / integrated and re-synthesized into an undecryptable flag in decoded data units. Thus, in the variable length decoding means,
Unnecessary code errors can be prevented from propagating by specifying the exact occurrence position in the variable length code of the uncorrectable flag added by the error correction decoding means due to a code error occurring in the symbol during the recording / reproduction or data transmission process. Prevention and error correction image quality can be improved.

[0012] According to a second aspect of the present invention, in the first aspect, the variable length decoding means cannot decode the variable length code because the code portion of the variable length code is not a code word table. If the uncorrectable flag of the code part is not detected from the beginning of the block to the variable length code,
It is characterized in that it is configured to add a non-decodable flag to all decoded data subjected to variable-length decoding in a block and to output the result.

In this case, since it is not possible to specify where the code error has propagated from or the position where the code error has occurred. I do. Therefore,
According to the second aspect, the decoding failure flag is not lost for decoded data in which a code error may have occurred, and as a result, remarkable information deterioration due to reproduction of incorrect decoded data is prevented. it can.

According to a third aspect of the present invention, in the decoding apparatus according to the first aspect, the variable-length decoding means determines that the total number of decoded data in a block unit is different from a predetermined number and that all variable data in the block are variable. When the uncorrectable flag is not detected in the code part of the long code, all the decoded data subjected to the variable length decoding in the block is added with the uncorrectable flag and output. .

[0015] The total number of decoded data decoded by the variable length decoding means should coincide with the total number of coefficient data included in a predetermined block if the decoding is correctly performed without a code error. Therefore, claim 3
In the case where the total number of the decoded data in the block unit is different from the predetermined number and no uncorrectable flag is detected in the code portions of all the variable length codes in the block, all the decoded data decoded in the block By adding an undecodable flag to the output, even if the uncorrectable flag is lost due to erroneous correction in the error correction decoding means, the variable length encoded with various combinations of bit lengths and bit patterns. By using the code as a kind of error detection code, a code error can be detected from the number of data to be decoded by itself and a non-decoding flag can be added. As a result, occurrence of a code error cannot be detected, and remarkable information deterioration due to reproduction of incorrect decoded data can be prevented. According to a fourth aspect of the present invention, in the decoding apparatus according to the first aspect, the variable length decoding means performs variable length decoding within the block when the total number of decoded data in the block unit matches a predetermined number. It is characterized in that it is configured to delete the non-decoding flag of all the decoded data and to output it.

Even in the case where an uncorrectable flag is erroneously added by the error correction decoding means even though the corrected data is not erroneous, the code according to various combinations of lengths and bit patterns is provided. The variable length code is used as a kind of error detection code,
It is possible to detect that there is no code error from the number of its own decoded data and delete the non-decoding flag. This makes it possible to review the erroneously detected code errors and reproduce the decoded data that has been decoded without errors, thereby preventing information deterioration.

According to a fifth aspect of the present invention, there is provided an encoding apparatus, wherein the encoding apparatus converts a plurality of coefficient data into a coefficient data sequence to be compressed by using a statistical property. The end of a specific pattern predetermined according to the number of fractional bits generated when variable length coding is performed for each block and the code length of the variable length code for each block is rounded up to an integral multiple of the symbol word length of the error correction code. A variable-length encoding unit that inserts a detection bit to generate encoded data, and an error-correction encoding unit that performs error correction encoding on the encoded data in units of a predetermined symbol word length to generate transmission data. It is characterized by having.

In the encoding apparatus according to the fifth aspect, the variable length encoding means uses an invalid bit in a gap of valid data, which has not been conventionally used, as an end detection bit of the variable length code. Recording / reproducing or data transmission of additional information for error detection without reducing efficiency. As a result, the operation of the decoding device in claims 6 and 7 described below can be made effective.

According to a sixth aspect of the present invention, there is provided a decoding apparatus which receives the transmission data output by the encoding apparatus according to the fifth aspect and performs the reverse process of the encoding apparatus to obtain decoded data. A decoding device for detecting and correcting errors in the transmission data in units of symbol word lengths, outputting corrected data, and adding an uncorrectable flag to a code error exceeding error correction capability and outputting the corrected data; A correcting / decoding means, when the corrected data is subjected to variable-length decoding to output decoded data, the uncorrectable flag is a code part represented by a variable length among a code part and a level part constituting a variable-length code. Or if it is in the fixed-length level part indicated by the code part, and if an uncorrectable flag is detected in the code part, the decoded data including the decoded data and thereafter are decoded at the end of the block. Conversion If the uncorrectable flag is detected only in the level part while the data is output with the undecodable flag added up to the data, the undecryptable flag is added only to the decoded data and output. Then, the end detection bits are cut out from the remaining corrected data obtained by decoding all the corrected data in the block, (f) the end detection bits have no uncorrectable flag, and (g) the end detection bits have a predetermined identification. Different from the pattern,
(H) When the condition that the uncorrectable flag is not detected in the code portions of all the variable-length codes in the block is simultaneously satisfied, the undecodable flag is added to all the variable-length decoded data in the block. A variable-length decoding unit that outputs the corrected data and an error correction unit that performs error correction on the decoded data to which the non-decoding flag is added using the decoded data and the non-decoding flag to obtain corrected data. It is characterized by having.

In the decoding apparatus of the present invention, the end detection bit is detected by the variable length decoding means and compared with a predetermined specific pattern, whereby the end detection bit and the predetermined specific pattern are distinguished from each other. In contrast, even when the uncorrectable flag is lost due to erroneous correction in the error correction decoding means, a variable length code coded with various combinations of lengths and bit patterns is replaced with a certain type of error detection code. To detect a code error from its own end detection bit and add a non-decodable flag. As a result, it is possible to prevent remarkable information deterioration caused by reproducing erroneous decoded data without detecting occurrence of a code error.

According to a seventh aspect of the present invention, there is provided a decoding apparatus which receives the transmission data output by the encoding apparatus according to the fifth aspect and performs the reverse process of the encoding apparatus to obtain decoded data. A decoding device for detecting and correcting errors in the transmission data in units of symbol word lengths, outputting corrected data, and adding an uncorrectable flag to a code error exceeding error correction capability and outputting the corrected data; A correcting / decoding means, when the corrected data is subjected to variable-length decoding to output decoded data, the uncorrectable flag is a code part represented by a variable length among a code part and a level part constituting a variable-length code. Or if it is in the fixed-length level part indicated by the code part, and if an uncorrectable flag is detected in the code part, the decoded data including the decoded data and thereafter are decoded at the end of the block. Conversion If the uncorrectable flag is detected only in the level part while the data is output with the undecodable flag added up to the data, the undecryptable flag is added only to the decoded data and output. Then, the end detection bits are cut out from the remaining corrected data obtained by decoding all the corrected data in the block, (i) the end detection bits have no uncorrectable flag, and (j) the end detection bits are specified in advance. Matches the pattern,
When the condition is satisfied at the same time, the variable length decoding means for erasing and outputting the non-decoding flag of all the decoded data subjected to variable length decoding in the block, and using the decoded data and the non-decoding flag Error correction means for performing error correction on the decoded data to which the non-decodable flag has been added to generate corrected data.

In the present invention, the end detection bit and the predetermined specific pattern match, and the error correction decoding means erroneously adds an uncorrectable flag even though the correction data is not erroneous. Even in such a case, a variable length code coded with a combination of various lengths and bit patterns is used as a kind of error detection code, and it is detected that there is no code error from its own end detection bit. The non-decryptable flag can be erased. As a result, erroneously detected code errors are reviewed,
By reproducing the decoded data that can be decoded without error, information deterioration can be prevented.

An eighth aspect of the present invention relates to an encoding apparatus, which encodes a plurality of coefficient data for a coefficient data sequence to be compressed by using a statistical property. Variable length coding means for performing variable length coding for each block, and multiplexing the code length of each block as variable length code into variable length code to obtain coded data; and a predetermined symbol word length for the coded data. An error correction encoding unit that performs error correction encoding on a unit basis to generate transmission data.

In the encoding device according to the eighth aspect, the operation of the decoding device according to the ninth and tenth aspects described below can be made effective.

According to a ninth aspect of the present invention, there is provided a decoding apparatus which receives the transmission data output by the encoding apparatus according to the eighth aspect and performs the reverse process of the encoding apparatus to obtain decoded data. A decoding device for detecting and correcting errors in the transmission data in units of symbol word lengths, outputting corrected data, and adding an uncorrectable flag to a code error exceeding error correction capability and outputting the corrected data; A correcting / decoding means, when the corrected data is subjected to variable-length decoding to output decoded data, the uncorrectable flag is a code part represented by a variable length among a code part and a level part constituting a variable-length code. Or if it is in the fixed-length level part indicated by the code part, and if an uncorrectable flag is detected in the code part, the decoded data including the decoded data and thereafter are decoded at the end of the block. Conversion If the uncorrectable flag is detected only in the level part while the data is output with the undecodable flag added up to the data, the undecryptable flag is added only to the decoded data and output. Then, the code length data for each block multiplexed with the correction data is cut out, and (m) the code length data has no uncorrectable flag, and (n) the variable length code decoded from the code length data is decoded. If the condition that the code length is different and (o) the uncorrectable flag is not detected in the code portions of all the variable length codes in the block are simultaneously satisfied, all the decoded data subjected to the variable length decoding in the block A variable-length decoding unit that adds and outputs a non-decoding flag, and corrects an error in the decoded data to which the non-decoding flag is added by using the decoded data and the non-decoding flag. It is characterized by and a error correction means and subjected to correction data.

In the decoding apparatus according to the ninth aspect, since the code length of the block is multiplexed by the variable length coding means, the coding efficiency is slightly reduced. However, by comparing the decoded code length of the block decoded by the variable length decoding means with the code length of the block coded by the coding device, the code length data and the decoded code length are different, and error correction decoding is performed. Even if the uncorrectable flag is lost due to erroneous correction in the conversion means, a variable length code coded with a combination of various lengths and bit patterns is used as a kind of error detection code, and , It is possible to detect a code error from the decoded code length and add a non-decodable flag. As a result, it is possible to prevent remarkable information deterioration due to reproduction of incorrect encoded data without detecting occurrence of a code error.

According to a tenth aspect of the present invention, there is provided a decoding apparatus comprising:
9. A decoding apparatus which receives transmission data output by the encoding apparatus according to claim 8 and obtains decoded data by performing a reverse process of the encoding apparatus, wherein the transmission data has a symbol word length. Error correction decoding means for outputting error correction data by error detection / correction in units, adding an uncorrectable flag to a code error exceeding the error correction capability, and outputting the error correction data by variable length decoding When outputting the encoded data, the uncorrectable flag is in the code part represented by the variable length among the code part and the level part constituting the variable length code, or in the fixed length level part indicated by the code part. If an uncorrectable flag is detected in the code part, the uncorrectable flag is added to the last decoded data of the block after that, including the decoded data, and output. When an uncorrectable flag is detected only in the bell portion, the uncorrectable flag is added to only the decoded data and output, and priority is given to each block multiplexed with the corrected data. When the conditions that (p) the code length data has no uncorrectable flag and (q) the code length data matches the decoded code length of the decoded variable length code are simultaneously satisfied, Variable length decoding means for erasing and outputting a non-decoding flag of all the decoded data subjected to variable length decoding in the block, and a non-decoding flag is added using the decoded data and the non-decoding flag. Error correction means for performing error correction on the decoded data to obtain corrected data.

According to the tenth aspect, when the code length data and the decoded code length match, and the error correction decoding means erroneously adds an uncorrectable flag even though the corrected data is not erroneous, However, even if the variable length code coded with a combination of various lengths and bit patterns is used as a kind of error detection code, decoding is performed by detecting that there is no code error from its own decoding code length. The disabling flag can be deleted. This makes it possible to review the erroneously detected code errors and reproduce the decoded data that has been decoded without errors, thereby preventing information deterioration.

Hereinafter, specific embodiments of the encoding apparatus and the decoding apparatus according to the present invention will be described in detail with reference to the drawings.

[Embodiment 1] FIG. 1 is a block diagram showing a configuration of an encoding apparatus and a decoding apparatus according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 10 denotes a variable-length coding unit that performs variable-length coding of coefficient data in block units to generate encoded data, and 11 denotes an error correction coding unit that performs error-correction coding of the encoded data in units of a fixed symbol word length to obtain transmission data. An error correction decoding means for detecting and correcting an error in symbol word length units and outputting corrected data, and outputting an uncorrectable flag if correction is impossible due to a code error exceeding the correction capability; The data is variable-length decoded to be decoded data, and the uncorrectable flag in symbol word length units is a code part represented by a variable length among a code part and a level part constituting a variable length code. Identify the fixed-length level part indicated by the part,
When an uncorrectable flag is detected in the code part, the uncorrectable flag is added to the last decoded data of the block after the decoded data and output, and when the uncorrectable flag is detected only in the level part, Variable length decoding means for adding a non-decodable flag only to the decoded data and outputting the result;
Reference numeral 4 denotes an error correction unit that performs error correction on the decoded data in accordance with the non-decodable flag to obtain corrected data. In addition, 1
5 is a recording head, 16 is a magnetic recording medium, and 17 is a reproducing head.

JPEG (Joint Photographic Coding Ex)
pers Group), the variable length code includes a code part represented by a variable length and a fixed length level part indicated by the code part. MPEG (Moving Picture Codin
g Experts Group), a quantized DCT (Discrete Cosine Transformer) rearranged in zigzag order.
m: Discrete cosine transform) The coefficients are two-dimensionally variable-length coded by combining the number of leading zero coefficients (runs) and the values and levels of non-zero quantized coefficients. , 113 VCLs (variable length codes) having a high frequency of occurrence are prepared, and pairs of run and level not prepared are represented by fixed length codes.

The operation of the encoding apparatus and the decoding apparatus according to the first embodiment configured as described above will be described below with reference to FIGS.

In an apparatus for encoding and decoding a coefficient data sequence to be compressed using statistical properties for each block composed of a plurality of the coefficient data, the variable-length encoding means 10 includes: Is subjected to variable length coding of the coefficient data. At this time, in order to perform error correction coding on a fixed symbol word length unit at a subsequent stage, the code length of the variable length code block unit is rounded up to an integral multiple of the symbol word length. Therefore, as shown in FIG. 2, a fractional bit 23 occurs after the last code included in the last symbol 21 of the variable length code 20 in block units. This fractional bit number may be from (8-1) = 7 bits to 0 bits, assuming that the symbol word length is 8 bits, for example. Usually, this fractional bit is an invalid bit that is not used for decoding.

Further, the error correction coding means 11 performs error correction coding for a fixed symbol word length unit, and adds the parity necessary for error correction to the transmission data via the recording head 15 to the recording medium 16. Be recorded. Recording medium 16
The transmission data reproduced with a code error through the reproduction head 17 is subjected to detection and correction of a code error by the error correction decoding means 12 as shown in FIG. If a symbol error cannot be corrected due to a code error exceeding the error correction capability, an uncorrectable flag 30 is added to each symbol word length.

As shown in FIG. 3, the variable-length decoding means 13 converts the corrected data 31 obtained by concatenating a plurality of symbols into decoded data 35 from the corrected data 31 and the uncorrectable flag 30 added in symbol word length units. Along with long decoding, whether the uncorrectable flag in symbol word length units is present in the variable length code portion of the variable length code or the fixed length level indicated by the code portion Part, or its occurrence position is specified.

When the uncorrectable flag 32 occurs in the code part, the delimiter of the code word becomes unclear due to the code error, and it can be determined that the error is highly likely to propagate to the code and thereafter. After that, a decoding impossible flag is added to the last decoded data of the block and output (C6, C7,... In FIG. 3).

On the other hand, when an uncorrectable flag is generated only in the level portion, it is determined that the reliability of the fixed-length data is low but the synchronization of the variable-length code is not lost, so that only the decoded data is decoded. An impossible flag is added and output (C2 in FIG. 3).

Accordingly, the uncorrectable flag 30 in symbol word length units is divided and integrated, and recombined into the undecryptable flag 34 in decoded data units.

Further, when decoding the variable length coded data, the variable length decoding means 13 cannot decode (a) because the variable length code is not present in the code word table, and (b) If the total number of decoded data is different from the predetermined number, and (c) the condition that the uncorrectable flag of the code part is not detected before the variable length code in the block, the variable A non-decodable flag is added to all the decoded data that has been subjected to long decoding, and output. Conversely, if the condition is that (d) all variable length codes are in the code word table and can be decoded, and (e) the total number of decoded data in block units matches a predetermined number, Deletes the non-decoding flags of all the decoded data subjected to variable-length decoding and outputs the result.

The error correcting means 14 is the variable length decoding means 13
Is used to correct the decoded data of the block in which the code error has been detected, based on the decoded data 35 decoded and the non-decoding flag 34 added to the decoded data unit to obtain corrected data.

FIG. 4 shows an embodiment of the error correction means 14 in the case where the coefficient data to be recorded is a coefficient data sequence in which coefficients obtained by orthogonally transforming an image signal are arranged in the order from low to high. . The error correction means 14 determines a correction method 45 based on whether the order of the decoded data to which the non-decodable flag 44 of the decoded data unit is added is higher or lower than a certain threshold.

The non-decodable flag 44 is added to the high-order decoded data 42 which is higher than the threshold value and contains only the fine information of the image.
Is added, since no significant deterioration is detected in the reproduced image of the block, although the definition of the image block is reduced, the coefficient is replaced with zero 43 to perform a correction that completes in the own block (C61 → “0”,
C62 → “0”, C63 → “0”).

If the low-order decoded data 40, which is lower than the threshold value and is the basis of the image, is damaged and the non-decodable flag 44 is added, if it is reproduced as it is, a large block-shaped Since the deterioration is detected, the data 41 is corrected by replacing it with the data 41 decoded in the past without error (C3 → C3 ′).

As described above, according to the first embodiment, the variable length decoding means 13 changes the uncorrectable flag added by the error correction decoding means 12 due to a code error occurring in the symbol during the recording / reproducing process. By specifying an accurate occurrence position in a long code, unnecessary code errors can be prevented from being propagated, and thus the image quality of error correction can be improved.

If the variable-length code cannot be decoded because it is not in the code word table, and if the uncorrectable flag of the code part is not detected before the variable-length code in the block,
That is, in the case of the above conditions (a) + (c), it is not possible to specify where the code error has propagated from or the position where the code error has occurred. And outputs the result. Therefore, the decoding failure flag is not lost for the decoded data in which a code error may have occurred, and as a result, remarkable information deterioration due to reproduction of incorrect decoded data can be prevented.

Further, the total number of decoded data decoded by the variable length decoding means 13 should coincide with the total number of coefficient data included in a predefined block if the decoding is correctly performed without a code error. It is. Therefore, by comparing the total number of the decoded data decoded in the block unit with the predetermined number, the total number of the decoded data and the predetermined number are different, and the uncorrectable flag of the code part is not reached by the variable length code in the block. Is not detected, that is, in the case of the above condition (b) + (c), a non-decodable flag is added to all the decoded data decoded in the block and output.
Therefore, even when the uncorrectable flag is lost due to erroneous correction or the like in the error correction decoding means 12, a variable length code coded with a combination of various lengths and bit patterns is used as a kind of error detection code. And
A code error can be detected from the number of data to be decoded by itself and a non-decoding flag can be added. As a result, occurrence of a code error cannot be detected, and remarkable information deterioration due to reproduction of incorrect decoded data can be prevented.

On the other hand, when the total number of the decoded data and the predetermined number match, that is, in the case of the above condition (d) + (e), all the decoded data decoded in the block cannot be decoded. Clear the flag and output. Therefore, even if the error correction flag is erroneously added in the error correction decoding means 12 even though the correction data is not erroneous, the variable length encoded with various combinations of bit lengths and bit patterns. By using the code as a kind of error detection code, it is possible to detect that there is no code error from the number of data to be decoded by itself and delete the non-decoding flag. This makes it possible to review the erroneously detected code errors and reproduce the decoded data that has been decoded without errors, thereby preventing information deterioration.

[Embodiment 2] Next, variable length decoding means of an encoding apparatus and a decoding apparatus according to Embodiment 2 of the present invention will be described with reference to FIGS. Embodiment 2
Are basically the same as those in the first embodiment (FIG. 1), but differ from the first embodiment in the following points. That is, when the variable length coding means 10 according to the second embodiment rounds up the code length of the variable length code 20 in block units to an integral multiple of the symbol word length, it is determined in advance according to the length of the fractional bit 23. The end detection bit of the specific pattern is inserted into encoded data.

On the other hand, the variable length decoding means 13 cuts out the end detection bit from the remaining data obtained by decoding all the corrected data in the block, and (f) there is no uncorrectable flag in the end detection bit, (g) If the condition that the end detection bit is different from the predetermined specific pattern and (h) the uncorrectable flag is not detected in the code part of the variable length code in the block are simultaneously satisfied, the variable length decoding is performed in the block. All the decoded data are output with a non-decodable flag added thereto, and conversely, (i) there is no uncorrectable flag in the end detection bit, and (j) the end detection bit matches a predetermined specific pattern. When the conditions are satisfied at the same time, it is configured to delete the non-decodable flags of all the decoded data subjected to variable length decoding in the block and output the data.

The operation of the encoding apparatus and the decoding apparatus according to the second embodiment configured as described above will be described below.

The variable length coding means 10 performs variable length coding of the coefficient data for each block, and performs error correction coding for each fixed symbol word length in the subsequent stage. Is rounded up to an integral multiple of the symbol word length. Therefore, as shown in FIG. 2, a fractional bit occurs after the final code in block units, as described in the first embodiment. Usually this fractional bit 23
Are invalid bits that are not used for decoding. For example, if the symbol word length is 8 bits, additional information of (8-1) = 7 to 0 bits per block can be recorded and reproduced without lowering the coding efficiency. . In the second embodiment, instead of this invalid bit, a bit string of a specific pattern predetermined according to the number of fractional bits is inserted as end detection bits to obtain encoded data.

For example, if the end detection bit is determined to be a continuation of simple "0", if the decoding is performed up to the final code of the block without any code error, the remaining data is always 7 to 0.
That is, the bits “0” are continuously inserted.
Therefore, the variable-length decoding means 13 performs the decoding shown in the first embodiment, decodes the data until the final code of the block, detects the end detection bit from the remaining data, and determines a predetermined specific pattern. To match. At this time, when the conditions of (f) the end detection bit having no uncorrectable flag and (g) the end detection bit being different from a predetermined specific pattern are satisfied at the same time, a code error occurs somewhere in the variable length code. By having occurred,
Determining that the end position of the variable length code has shifted, and (h)
If the condition that the uncorrectable flag is not detected in the code part of the variable-length code in the block is satisfied, it is difficult to limit the position where the code error first occurred. A non-decodable flag is added to all the decoded data subjected to variable-length decoding in the block, and output.

Even if the conditions (f) and (g) are simultaneously satisfied, the condition (k) is that an uncorrectable flag is detected in the code part of the variable-length code in the block. Determines that the beginning of the code error is the decoded data in which the uncorrectable flag is first detected in the code part, and as shown by 5b in FIG. 5, the decoding in which the uncorrectable flag is first detected in the code part. After that, including the data, a non-decodable flag is added to the final decoded data of the block. This is the same as in the first embodiment. Conversely, (i) there is no uncorrectable flag in the end detection bit,
(J) If the condition that the end detection bit matches a predetermined specific pattern is satisfied at the same time, it is determined that the uncorrectable flag is erroneously added even though the corrected data is not erroneous. As shown by 5c, the decoding disable flags of all the decoded data subjected to the variable length decoding in the block are deleted and output.

With the above configuration, according to the second embodiment, the variable length coding means 10 detects the invalid bits in the gaps of the valid data, which have not been used conventionally, to detect the end of the variable length code. By using them as bits, additional information for error detection is recorded and reproduced without lowering the coding efficiency. By detecting the end detection bit by the variable length decoding means 13 and comparing it with a predetermined specific pattern, the end detection bit is different from the predetermined specific pattern, and erroneous correction in the error correction decoding means 12 is performed. Even if the uncorrectable flag is lost, the variable-length code coded with various combinations of lengths and bit patterns is used as a kind of error detection code. An error can be detected and a non-decodable flag can be added. As a result, it is possible to prevent remarkable information deterioration caused by reproducing erroneous decoded data without detecting occurrence of a code error.

On the contrary, even if the end detection bit and the predetermined specific pattern match, and the error correction decoding means 12 has no error in the correction data, the error correction flag is erroneously added. Even if this is the case, use a variable-length code coded with a combination of various lengths and bit patterns as a kind of error detection code, and detect that there is no code error from its own end detection bit. The non-decodable flag can be deleted.
This makes it possible to review the erroneously detected code errors and reproduce the decoded data that has been decoded without errors, thereby preventing information deterioration.

[Third Embodiment] Next, the variable length decoding means of the encoding device and the decoding device according to the third embodiment of the present invention will be described. The configurations of the encoding device and the decoding device according to the third embodiment are basically the same as those of the first embodiment.
It is the same as FIG. 1 but different from the first embodiment in the following points. That is, the variable-length encoding means 10 according to the third embodiment is configured to multiplex the encoded code length in block units as variable-length code into variable-length code to obtain encoded data.

On the other hand, the variable length decoding means 13 cuts out the code length data multiplexed into the variable length code,
(M) the code length data has no uncorrectable flag; (n) the code length data is different from the decoded code length of the decoded variable length code; (o) the code portion of all the variable length codes in the block is corrected. If the condition that the impossible flag is not detected is satisfied at the same time, a decoding impossible flag is added to all the decoded data subjected to variable length decoding in the block and output, and conversely,
When the conditions of (p) that there is no uncorrectable flag in the code length data and (q) that the code length data matches the decoded code length of the decoded variable length code are simultaneously satisfied, the variable length decoding is performed in the block. It is configured to delete and output the non-decoding flag of all the decoded data.

The operation of the encoding apparatus and the decoding apparatus according to the third embodiment configured as described above will be described below.

The variable length coding means 10 performs variable length coding of the coefficient data for each block, and multiplexes the coded code length in block units as variable length code into variable length code to obtain coded data.

For example, if it is possible to decode up to the last code of the block without any code error, the decoded code length of the decoded block matches the code length data of the block multiplexed by the variable length coding means 10. Therefore, the variable-length decoding means 13 performs the decoding shown in the first embodiment, decodes the code up to the final code of the block, and then cuts out the code-length data multiplexed into the variable-length code.
If the conditions of (m) that there is no uncorrectable flag in the code length data, and (n) that the code length data is different from the decoded code length of the decoded variable length code are satisfied at the same time, the code is generated somewhere in the variable length code. If it is determined that the code lengths do not match due to the occurrence of the error, and (o) the condition that the uncorrectable flag is not detected in the code portions of all the variable length codes in the block is satisfied, the code error is determined. Since it is difficult to limit the position where the error occurs first, a decoding impossible flag is added to all the decoded data subjected to variable length decoding in the block and output in the same way as in the case of 5a in FIG. .

Even if the conditions (m) and (n) are satisfied at the same time, the condition (r) is that the uncorrectable flag is detected in the code part of the variable-length code in the block. It is determined that the head of the code error is the decoded data in which the uncorrectable flag is first detected in the code part, and the uncorrectable flag is first detected in the code part in the same way as in the case of 5b in FIG. Thereafter, a non-decodable flag is added to the final decoded data of the block including the decoded data. This is the same as in the first embodiment. Conversely, (p) there is no uncorrectable flag in the code length data,
(Q) When the condition that the code length data coincides with the decoded code length of the decoded variable length code is simultaneously satisfied, it is determined that the uncorrectable flag is erroneously added even though the corrected data is not erroneous. Then, in the same way as in the case of 5c in FIG. 5, the non-decoding flags of all the decoded data subjected to variable length decoding in the block are deleted and output.

With the above-described configuration, according to the third embodiment, the code length of the block is multiplexed by the variable length coding means 10, so that the coding efficiency is slightly reduced. However, by comparing the decoded code length of the block decoded by the variable length decoding means 13 with the code length of the block coded by the coding device, the code length data and the decoded code length are different, and error correction is performed. Decoding means 1
Even if the uncorrectable flag is lost due to erroneous correction or the like, the variable length code coded with various combinations of length and bit pattern is used as a kind of error detection code, A code error can be detected from the decoded code length and a non-decodable flag can be added. As a result, it is possible to prevent remarkable information deterioration due to reproduction of incorrect encoded data without detecting occurrence of a code error.

On the other hand, in the case where the code length data and the decoded code length match, and the error correction decoding unit 12 erroneously adds an uncorrectable flag even though the corrected data is not erroneous, Even if there is a variable length code that is coded with a combination of various lengths and bit patterns, it cannot be decoded by detecting that there is no code error from its own decoded code length, using a certain type of error detection code. Flags can be cleared. This makes it possible to review the erroneously detected code errors and reproduce the decoded data that has been decoded without errors, thereby preventing information deterioration.

In each of the above embodiments, the recording and reproduction of the coefficient data sequence has been described as an example. However, the present invention is also effective in a configuration in which the recording medium is replaced with a transmission medium such as satellite communication. Also, the symbol word length of the error correction encoding means 11 has been described as being 8 bits, but the symbol word length is not limited to 8 bits.

[0065]

As described above, according to the present invention,
Even if an unavoidable code error occurs during the code recording / reproduction or data transmission process, by accurately specifying the position of the occurrence of the code error in the variable length code, unnecessary propagation of the uncorrectable flag is prevented, The decoded data that can be decoded without errors can be used as effectively as possible. Further, even when there is no code error flag such as error correction by an error correction code and it is not possible to detect in advance that the code is incorrect before decoding, the variable length code is used as a kind of error detection code and is not included in the code word table. Error detection by irregular codeword, error detection by the number of decoded data, error detection by end detection bit, error detection by decoded code length, etc.
By combining error detection using one or a plurality of variable length codes, a code error can be detected with high accuracy. Therefore, instead of error correction relying only on past decoded data having a low correlation, it is possible to correct block-complete errors by increasing the number of decoded data that can be correctly decoded even for a block containing an erroneous code. Therefore, the quality of error correction in a time-varying coefficient data sequence such as a moving image can be significantly improved. In addition, since the error detection capability can be enhanced, it is possible to prevent significant information degradation due to decoding of erroneously decoded data as it is.

In the present invention, a code error is detected with high accuracy by positively utilizing the error propagation characteristic that a variable length code originally has as a weak point, so that a large-scale means is not required for detecting a code error. The practical effect is also large in this respect.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an encoding device and a decoding device according to Embodiments 1, 2, and 3 of the present invention.

FIG. 2 is a diagram showing fractional bits generated when a variable length code is rounded up to an integral multiple of a symbol word length in the first and second embodiments.

FIG. 3 is a diagram showing an operation concept of specifying an occurrence position of an uncorrectable flag in symbol word length units corresponding to a code part and a level part of a variable length code in the first embodiment.

FIG. 4 is a diagram showing one embodiment of an error correction unit in the first embodiment.

FIG. 5 is a diagram showing an example of generating a non-decodable flag based on a result of error detection in the second embodiment.

[Explanation of symbols]

Reference Signs List 10 variable length coding means 11 error correction coding means 12 error correction decoding means 13 variable length decoding means 14 error correction means 15 recording head 16 recording medium 17 ... Reproduction head 20... Variable-length code in block units rounded up to an integral multiple of the symbol word length 21... Last symbol of variable-length code with fractional bits inserted 22... Variable-length code 23. 30: Uncorrectable flag in symbol word length unit 31: Corrected data 32: Code part to which uncorrectable flag is added 33 ... Level part to which uncorrectable flag is added 34: Unable to decode in decoded data unit Flag 35... Decoded data 40... Low-order decoded data to which a non-decodable flag has been added 41... Modified data replaced with data that has been decoded without error in the past 42... Higher-order decoded data to which a non-decodable flag has been added 43... Modified data replaced with zero 44... Non-decodable flag 45 per decoded data unit 45. Judgment criterion 50: decoded data

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁶ Identification code Agency reference number FI Technical display location H04N 1/41 H04N 1/41 Z 5/92 5/92 H 7/24 7/13 A

Claims (10)

[Claims] 1. A method according to claim 1, wherein a predetermined symbol word is added to the coded data which is variable-length coded for each block composed of a plurality of the coefficient data with respect to a coefficient data sequence to be compressed using a statistical property. A decoding device for inputting transmission data on which error correction encoding has been performed in units of length, and performing a reverse process of the error correction encoding and variable length encoding to obtain decoded data, wherein the transmission data Error detection per symbol word length
Error correction decoding means for outputting corrected data after correction and adding an uncorrectable flag in the case of a code error exceeding the correction capability, and outputting the decoded data by variable-length decoding the corrected data. Of the code part and the level part constituting the variable length code, the uncorrectable flag specifies whether the code part is represented by a variable length or a fixed length level part indicated by the code part, If an uncorrectable flag is detected in the code part, the uncorrectable flag is added to the last decoded data of the block after that, including the decoded data, and output. If it is detected, a variable length decoding means for adding a non-decoding flag only to the decoded data and outputting it, and using the decoded data and the non-decoding flag, Grayed decoding apparatus characterized by comprising an error correction means for the correction data by performing error correction with respect to the appended decoded data. 2. The variable-length decoding means cannot decode the variable-length code because the code part of the variable-length code is a code that is not in the codeword table. 2. The apparatus according to claim 1, wherein when an uncorrectable flag is not detected, an undecodable flag is added to all decoded data subjected to variable-length decoding in the block and output. Decryption device. 3. The variable-length decoding means, when the total number of decoded data in a block unit is different from a predetermined number, and when an uncorrectable flag is not detected in a code portion of all variable-length codes in the block, 2. The decoding device according to claim 1, wherein the decoding device is configured to add a non-decoding flag to all the decoded data subjected to variable-length decoding and output the data. 4. The variable length decoding means deletes a non-decoding flag of all variable length decoded data in a block when the total number of decoded data in a block unit matches a predetermined number. 2. The decoding device according to claim 1, wherein the decoding device is configured to output. 5. A coefficient data sequence to be compressed using a statistical property is subjected to variable length coding for each block including a plurality of said coefficient data, and error correction of the code length of the variable length code for each block is performed. A variable-length encoding unit that inserts end detection bits of a predetermined specific pattern according to the number of fractional bits generated when rounding up to an integral multiple of the symbol word length of the code to produce encoded data, An encoding apparatus comprising: an error correction encoding unit that performs error correction encoding on data in units of a predetermined symbol word length to generate transmission data. 6. A decoding device which receives transmission data output from the encoding device according to claim 5 and performs a process reverse to that of the encoding device to obtain decoded data. Error detection in symbol word length units
Error correction decoding means for correcting and outputting corrected data, adding an uncorrectable flag to a code error exceeding the error correction capability, and outputting the corrected data, and outputting the decoded data by variable-length decoding the corrected data At this time, it is specified whether the uncorrectable flag is present in the code part represented by the variable length or the fixed length level part indicated by the code part among the code part and the level part constituting the variable length code. If an uncorrectable flag is detected in the code portion, the uncorrectable flag is added to the last decoded data of the block including the decoded data, and the uncorrectable flag is output only to the level portion. Is detected, a decoding failure flag is added only to the decoded data, and the decoded data is basically output. The end detection bit is cut out from the corrected data of
The condition that there is no uncorrectable flag in the end detection bit, (g) the end detection bit is different from the predetermined specific pattern, and (h) the uncorrectable flag is not detected in the code portions of all variable length codes in the block. Are simultaneously established, a variable-length decoding means for adding a non-decoding flag to all the decoded data subjected to variable-length decoding in the block and outputting the data, and using the decoded data and the non-decoding flag, An error correcting means for performing error correction on the decoded data to which the non-decodable flag has been added to generate corrected data. 7. A decoding device which receives transmission data output from the encoding device according to claim 5 and obtains decoded data by performing a process reverse to that of the encoding device. Error detection in symbol word length units
Error correction decoding means for correcting and outputting corrected data, adding an uncorrectable flag to a code error exceeding the error correction capability, and outputting the corrected data, and outputting the decoded data by variable-length decoding the corrected data At this time, it is specified whether the uncorrectable flag is present in the code part represented by the variable length or the fixed length level part indicated by the code part among the code part and the level part constituting the variable length code. If an uncorrectable flag is detected in the code portion, the uncorrectable flag is added to the last decoded data of the block including the decoded data, and the uncorrectable flag is output only to the level portion. Is detected, a decoding failure flag is added only to the decoded data, and the decoded data is basically output. The end detection bit is cut out from the corrected data, and (i)
If the condition that the end detection bit has no uncorrectable flag and (j) the end detection bit matches a predetermined specific pattern is simultaneously satisfied, all the decoded data of the variable length decoded data in the block are Variable-length decoding means for erasing and outputting a non-decryptable flag, and using the decoded data and the non-decryptable flag to perform error correction on the decoded data to which the non-decryptable flag has been added, A decoding device comprising: 8. A coefficient data sequence to be compressed using a statistical property is subjected to variable length coding for each block composed of a plurality of said coefficient data, and the code length of each block is set to variable length as code length data. A variable-length encoding unit that multiplexes the code into encoded data, and an error-correction encoding unit that performs error correction encoding on the encoded data in units of a predetermined symbol word length to produce transmission data. An encoding device characterized in that: 9. A decoding device which receives transmission data output by the encoding device according to claim 8 and performs a process reverse to that of the encoding device to obtain decoded data. Error detection in symbol word length units
Error correction decoding means for correcting and outputting corrected data, adding an uncorrectable flag to a code error exceeding the error correction capability, and outputting the corrected data, and outputting the decoded data by variable-length decoding the corrected data At this time, it is specified whether the uncorrectable flag is present in the code part represented by the variable length or the fixed length level part indicated by the code part among the code part and the level part constituting the variable length code. If the uncorrectable flag is detected in the code part, the uncorrectable flag is added to the last decoded data of the block, including the decoded data, and output. Is detected, an undecryptable flag is added to only the decoded data, and the data is output. In preference to this, each block multiplexed with the corrected data is (M) There is no uncorrectable flag in the code length data, (n) the code length data is different from the decoded code length of the decoded variable length code, (o) all variables in the block A variable-length decoding means for adding an undecodable flag to all decoded data subjected to variable-length decoding in the block and outputting the decoded data when the condition that an uncorrectable flag is not detected in the code portion of the long code is simultaneously satisfied; And error correction means for performing error correction on the decoded data to which the non-decoding flag has been added using the decoded data and the non-decoding flag to obtain corrected data. Decryption device. 10. A decoding apparatus for receiving transmission data output by the encoding apparatus according to claim 8 and performing decoding in reverse by the encoding apparatus to obtain decoded data. Error detection in symbol word length units
Error correction decoding means for correcting and outputting corrected data, adding an uncorrectable flag to a code error exceeding the error correction capability, and outputting the corrected data, and outputting the decoded data by variable-length decoding the corrected data At this time, it is specified whether the uncorrectable flag is present in the code part represented by the variable length or the fixed length level part indicated by the code part among the code part and the level part constituting the variable length code. If an uncorrectable flag is detected in the code portion, the uncorrectable flag is added to the last decoded data of the block including the decoded data, and the uncorrectable flag is output only to the level portion. Is detected, an undecryptable flag is added to only the decoded data, and the data is output. In preference to this, each block multiplexed with the corrected data is When the conditions that (p) the code length data has no uncorrectable flag and (q) the code length data matches the decoded code length of the decoded variable length code are simultaneously satisfied, A variable-length decoding unit that erases and outputs a non-decodable flag of all decoded data subjected to variable-length decoding in the block; and a non-decoding flag is added using the decoded data and the non-decoding flag. Error correction means for performing error correction on the decoded data to generate corrected data.