JPH08251597A - Moving image encoding and decoding device - Google Patents

Abstract

PURPOSE: To provide a moving image encoding device which can perform refreshing operation efficiently by suppressing the influence of an error and preventing the error from being propagated between frames even when the error rate of a transmission line or storage medium is high. CONSTITUTION: This device has a block dividing circuit 102 which divides an input moving image signal 101 into plural blocks consisting of plural pixels, a mode selecting circuit 105 which selects an encoding mode by the divided blocks, 1st encoding circuits 108, 109, and 112 which encodes the respective blocks of the input moving image signal 101 in the selected encoding mode, and a 2nd encoding circuit 113 which encodes mode information showing the selected encoding mode; and the mode selecting circuit 105 is constituted so as to selectively set plural kinds of screen including a 1st screen where the intra-encoding of blocks that are larger in difference between two adjacent screens more than a specific threshold value is performed and a 2nd screen where inter-prediction encoding is performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding / decoding device.

[0002]

2. Description of the Related Art Video signals are transmitted / received in TV telephones, TV conference systems, personal digital assistants, digital video disc systems, digital TV broadcasting systems, etc.
In a storage system, a moving image signal is compression-encoded into the smallest possible amount of information, the obtained compression-encoded signal is transmitted / stored in a transmission line / storage medium, and the transmitted / stored code sequence is stored. The original moving image signal is reproduced by decoding.

In such a system, in general, if there is an error in the transmission line / storage medium between the moving picture coding apparatus and the moving picture decoding apparatus, the decoding apparatus cannot obtain correct information. There is a problem that a decoded image with high image quality cannot be obtained. There are various measures to be taken when there is an error in the transmission path / storage medium. Among them, an effective method is to use the information of the previous frame periodically within one frame on the encoding device side. There is a method of performing encoding, that is, inserting a pixel area for performing intra-frame encoding, and refreshing a screen disturbed by the influence of an error.

Regarding refresh, for example, MPEG
2 (ISO / IEC CD13818
-2), 16 pixels x 16 called a macroblock
Intra-frame coding and inter-frame coding can be switched on a pixel-by-pixel basis.
There is shown a method called intra slice, in which macroblocks of horizontal 2 rows of 1 frame composed of 44 horizontal macroblocks are intra-coded and 2 rows are slid every 1 frame time to form 1 cycle for 15 frames. .
In this method, as shown in FIG. 7, when inter-frame coding is performed in the area 2, by limiting the search range of the motion vector not to start from the area 1 that has not been refreshed, the worst two cycles are completed. Guaranteed to be refreshed.

In addition, H.264 of the old CCITT. DC in 261
1 macroblock to eliminate T mismatch
It is recommended to use intra macroblocks once during 32 times of encoding (edited by Yasuda: "International standard for multimedia encoding", p.89, Maruzen). Furthermore, in order to eliminate the image quality deterioration due to an error in the transmission line, 1992
Image Coding Symposium (PCSJ'92), 6-
1, a frame (intra frame) or a slice (intra slice) for intra-frame coding is periodically inserted, as also used in the "ATM image coding method having cell discard resistance".

A problem with such a periodic refresh system is that when the error rate of the transmission line / storage medium is high, the effect of a new error spreads on the decoded image before the screen disturbed by the error is refreshed. In order to
One of the reasons is that the refreshing effect is not fully exerted. In order to enhance the effect of refresh, the refresh cycle may be shortened, but in that case, the ratio of intra-frame coding, which is lower in coding efficiency than that of inter-frame coding, is high, so that the overall code However, there is a problem that the efficiency of image conversion is reduced and the image quality is deteriorated.

A problem with the refresh method itself is that it is possible to prevent the influence of an error from propagating between frames, but if there is an error in the intraframe coded information itself in the refreshed area, However, it is not possible to suppress the effect of error itself. Since the intra-frame coded information for refresh has a relatively large code amount as compared with the coded information at other places,
The probability of error in the storage medium is relatively high. Further, since the information of the previous frame is not used in the refresh area, the coding efficiency is poor, and the ratio of the information in the refresh area is large, especially at a low bit rate, and the coding performance is deteriorated.

When refresh is realized by periodically inserting intra slices as shown in FIG. 8, if an error occurs immediately after the refresh, one cycle of error is propagated in the area 1. In this case, in the moving area of the screen, image quality deterioration due to an error is more likely to remain than in the still area, but the periodic refresh does not consider whether it is the moving area or the still area.

On the other hand, as another measure against the error in the transmission line / storage medium, the variable length code output from the coding device is divided into a fixed length, for example, 239 bits, and 16 bits of error correction code is added to this. Then, there is a method in which the moving image signal is reproduced after being output to the transmission line / storage medium and subjected to error correction in the decoding device. This method can sufficiently correct the error when the error rate is sufficiently low, but cannot be sufficiently corrected only by the error correction code when the error rate becomes high. If an error remains in the variable-length code, the length of the variable-length code is unknown, and thereafter, random decoding is performed, so that the decoded image is greatly deteriorated.

Therefore, an error occurs when a variable length code that does not exist in the variable length code table is found in the variable length decoding in the decoding device, or when there is no decoded value. Therefore, the decoding process is stopped, the detection of the next synchronization code is started, all the data in the meantime is discarded, and the data is not displayed on the screen. However, since the variable-length code table is originally created so that the redundancy is small, the number of codes that do not exist in the table is small and the probability of error detection is low. Moreover, the probability that the decoded value is impossible is low. Therefore, it takes a considerable amount of time from the time when an error actually occurs until the error is detected, an erroneous decoding result up to that time is displayed on the screen, and the decoded image deteriorates.

Further, as another countermeasure against an error in the transmission path / storage medium, 1 is added to the prediction signal in the motion compensation prediction.
Leak predictions that multiply by smaller coefficients are also known,
When this leak prediction is performed, the DC component of the image signal changes, which causes a problem that the brightness of the image changes.

A synchronization code is used as a method for recovering synchronization in a decoding device when an error occurs in a transmission line / storage medium. All the data from the detection of the error to the detection of the synchronization code is discarded, so in order to reduce the amount of data to be discarded, it is sufficient to insert as many synchronization codes as possible. However, the more the sync code is inserted, the more the amount of information in the image must be reduced, which causes a problem that the coding performance is deteriorated.

The synchronization code will be specifically described. For example, H.264 of the old CCITT is used. In FIG. 261, a synchronization code is added as shown in FIG. 21A for each GOB (group of blocks) composed of 11 MB × 3 lines shown in FIG. FIG. 21A shows an array of data for one frame, and the data described here as GOB # 1 is the encoded data of the first GOB. The same applies to GOB # 3 and the like. As the synchronization code PS of each picture, a code in which 4-bit "0" is added to a 16-bit unique word (UW) is used as shown below.

[0014] The sync code GS of each GOB is 1 as shown below.
A code in which a GOB number represented by 4 bits is added to a 6-bit UW is used.

[0015] If the GOB number is known, the position of each MB on the screen can be specified. In addition, in ISO MPEG1 of FIG.
One or more MBs consecutive in the scanning order of the image shown in
A synchronization code is added as shown in FIG. 22A for each slice constituted by. FIG. 22A shows an array of data for one frame, and the data described as slice # 1 here is the encoded data of the first slice. sl
The same applies to ice # 2 and the like. Sync code PS for each picture
Is a 24-bit UW with an 8-bit 0
A code added with is used.

[0016] Further, as the synchronization code SS of each slice, a code is used in which the value of the SVP to which the head MB of the slice represented by 8 bits is added to the 24-bit UW as shown below.

[0017] The SVP value is from 1 to 175. Each M
The position of B in the screen can be specified from the SVP and the address of the first MB of the slice. The details of the structure of the synchronous code described above are described in "International Standard of Multimedia Coding", edited by Yasuda, Maruzen.

Since the synchronization code uses a unique word having a long code length, when encoding at a low rate, if the synchronization code is frequently inserted to recover the loss of synchronization due to an error, the total code amount is increased. Since the ratio of the synchronization code to the total becomes large, the coding efficiency is significantly reduced. Further, in MPEG1, two pieces of information, that is, the SVP and the address information of the head MB, are required to specify the position of each MB in the screen, and thus are susceptible to errors.

On the other hand, as another method of compressing and coding a moving image signal, conditional pixel supplement (CR: Conditional Replen
ishment) method is known. This method is disclosed in, for example, Nikkei Electronics 1984.4.23, pp. 197
As described in the article entitled "Band compression of color TV signals for video conferences and transmission over digital telephone lines" described in No. 203 to No. 203, rewriting the previous frame in the reference frame memory in the encoding device. The data of the completed frames is accumulated, the error (the degree of dissimilarity) between the current frame and the reference frame is calculated for each block of 8 × 8 pixels, and the block in which this error exceeds the threshold changes greatly. Therefore, the contents of the reference frame memory are updated, and the information regarding the block is encoded and transmitted / stored. In the decoding device, the corresponding block in the frame reproduced using the transmitted / stored encoded information is rewritten, and the block in which the error is below the threshold value does not change, so no processing is performed. I will not rewrite it. The threshold value is determined by the rate control circuit so that the output buffer provided in the output unit of the encoder will not become empty or full.

This conditional pixel replenishment method has an advantage that the image quality deterioration of important information is immediately recovered because the reference frame memory is frequently rewritten for a block having a large change (a block having a large amount of information). The coding efficiency is lower than that of the method of coding the motion compensation prediction error described above. Also, when the amount of change between the current frame and the reference frame is calculated between the original signals, the amount of change in the still region is almost 0, so even if there is a margin in the output buffer,
The contents of the reference frame memory in that area cannot be rewritten. Therefore, there is a problem that image quality deterioration due to quantization error or error remains forever.

[0021]

As described above, the conventional moving picture coding / decoding apparatus has the following problems.

In the periodic refresh method, which is one of the countermeasures against errors in the transmission line / storage medium, when the error rate is high, the influence of a new error affects the decoded image before the screen disturbed by the error is refreshed. Therefore, if the refresh cycle is shortened in order to improve the refresh effect, the rate of intra-frame coding, which reduces coding efficiency, increases, so that the overall coding efficiency Deteriorates and the image quality deteriorates. Further, when the intra-frame coded information itself in the refreshed area has an error, there is a problem that the influence of the error cannot be suppressed.

Further, as another example of the error countermeasure in the transmission line / storage medium, the variable length code is divided into a certain length, an error correction code is added and output to the transmission line / storage medium. The method of reproducing the moving image signal after performing error correction is
When the error rate becomes higher, the error-correcting code alone cannot correct it sufficiently, and when the error remains in the variable-length code, the length of the variable-length code becomes unknown. As a result, the random decoding is performed thereafter. Therefore, the decoded image is greatly deteriorated.

In order to solve this problem, when a variable-length code that does not exist in the variable-length code table is found during variable-length decoding, or when a decoded value is impossible, it is determined that an error has occurred. With the method of stopping the conversion process and starting the detection of the next sync code, and discarding all the data in the meantime and not displaying it on the screen, the variable length code table was originally created to reduce redundancy and exists in the table. The number of codes that are not used is small, the probability that an error will be detected is low, and the probability that a decoded value will be impossible is low, so it takes a considerable time from the time when an error actually occurs until the error is detected. There is a problem that the erroneous decoding result up to is displayed on the screen and the decoded image deteriorates. In addition, as another error countermeasure, the method of performing the leak prediction has a problem that the brightness of the image changes because the DC component of the image signal changes.

When the synchronization code is used for the synchronization recovery in the decoding device when an error occurs in the transmission line / storage medium, all the data from the detection of the error to the detection of the synchronization code are discarded. Therefore, if many synchronization codes are inserted to reduce the amount of data to be discarded as much as possible, the information amount of the image must be reduced, which causes a problem that the encoding performance deteriorates.

Further, since the synchronization code uses a unique word having a long code length, when encoding at a low rate, if the synchronization code is frequently inserted in order to recover from the loss of synchronization due to an error, the entire code amount is obtained. Since the ratio of the synchronization code to the total becomes large, the coding efficiency is significantly reduced. Further, since two pieces of information are required to specify the position of each macroblock on the screen, there is a problem that it is easily affected by an error.

On the other hand, the conditional pixel replenishment method has a high error resistance, but on the other hand, a block having a large change has a low coding efficiency, and the image quality deterioration due to a quantization error or an error remains in the still region of the image. There is.

The present invention has been made to solve the above problems, and provides a moving picture coding / decoding apparatus having high error resistance and excellent coding performance.
That is, the first object of the present invention is to suppress the influence of an error even when the error rate in a transmission line or a storage medium is high, prevent the error from propagating between frames, and perform refresh efficiently. An object of the present invention is to provide a moving picture coding / decoding device capable of performing the above.

A second object of the present invention is to improve the detection accuracy for errors in the transmission line and the storage medium and prevent the output of erroneous decoding results, thereby suppressing deterioration of image quality due to errors. It is to provide an encryption / decoding device.

A third object of the present invention is to provide a moving picture coding / decoding apparatus which suppresses an increase in the code amount due to periodic refresh and has a high coding efficiency.

A fourth object of the present invention is to provide a moving picture coding / decoding apparatus capable of refreshing image quality deterioration occurring in a still area of an image faster than periodic refreshing.

A fifth object of the present invention is to provide a moving picture code which has a high ability to recover errors in a transmission line or a storage medium and which can reduce a redundant code amount for the purpose of recovery and reduce the influence of errors. It is to provide an encryption / decoding device.

A sixth object of the present invention is to provide a moving picture coding / decoding apparatus which is resistant to errors in a transmission line and a storage medium and which can improve the coding efficiency of the conditional pixel supplement system. It is in.

[0034]

A moving picture coding apparatus according to a first aspect of the present invention divides an input moving picture signal into a plurality of blocks composed of a plurality of pixels, and a block dividing means for dividing the moving picture signal. Mode selecting means for selecting an encoding mode for each block, first encoding means for encoding each block of the input moving image signal in the encoding mode selected by the mode selecting means, and the mode selecting Second encoding means for encoding mode information indicating the encoding mode selected by the means, wherein the mode selecting means is an intra code for a block in which a difference between two adjacent screens is equal to or more than a predetermined threshold value. It is characterized by selectively setting a plurality of types of screens including a first screen for performing encoding and a second screen for performing inter prediction encoding.

Here, the mode selection means performs inter-prediction coding on a block in which the difference between the two adjacent screens is equal to or greater than a predetermined threshold, and intra-codes at least a part of the block whose difference is less than the threshold. A third screen for encoding is further included as one of the plurality of types of screens.

A moving picture decoding apparatus according to a first aspect of the invention decodes an original moving picture signal by decoding a code string obtained by coding a moving picture signal in a coding mode selected for each block composed of a plurality of pixels. A first decoding means for reproducing, a second decoding means for decoding a code string obtained by encoding mode information indicating the encoding mode, and a mode decoded by the second decoding means. Mode selecting means for selecting a decoding mode of the second decoding means based on information, wherein the mode selecting means is intra-coded for a block in which a difference between two adjacent screens is a predetermined threshold value or more. First to do
It is characterized in that the decoding mode is selectively set in correspondence with a plurality of types of screens including the screen of (1) and a second screen for performing inter prediction coding.

A moving picture coding apparatus according to a second aspect of the present invention is a variable length coding means for variable length coding of coded data obtained by compression coding an input moving picture signal, and a code output from this coding means. A checksum generating unit that divides the sequence into a predetermined number of bits for each encoding unit including a plurality of variable-length codes, adds the codes of the respective divisions, and generates a value of a lower predetermined number of bits of the added value as a checksum. And a checksum adding means for adding the checksum generated by the checksum generating means to each of the coding units and outputting the added checksum.

The moving picture decoding apparatus according to the second aspect of the invention is a decoding means for decoding the coded string obtained by variable-length coding the coded data obtained by compression-coding the moving picture signal to reproduce the original moving picture signal. And dividing the code string into a predetermined number of bits for each coding unit consisting of a plurality of variable length codes, adding the codes of the respective divisions, and adding the value of the lower predetermined number of bits of the added value to each of the coding units. And an error detecting means for detecting an error in the code string by comparing with the checksum added to.

A moving picture coding apparatus according to a third aspect of the invention divides an input moving picture signal and a motion compensated prediction signal obtained by performing motion compensation on the input picture signal into a plurality of subbands each having a different frequency band. The first and second sub-band dividing means and the lowest frequency sub-band of the plurality of sub-bands are divided into a plurality of areas on the screen at least once within a predetermined period for each area. First coding means for intra-coding the input moving image signal of the lowest frequency sub-band from the sub-band dividing means, and sub-bands other than the lowest frequency among the plurality of sub-bands, the second coding means The leak prediction signal obtained by multiplying the motion-compensated prediction signal of the sub-band other than the lowest frequency from the sub-band splitting means by a coefficient smaller than 1 and the first sub-band splitting means. And having a second encoding means for encoding the prediction error signal of the moving image signal of the outermost low-frequency other subbands.

Further, the moving picture coding apparatus according to the second aspect of the present invention is provided with means for increasing / decreasing the area to be refreshed by the intra coding and the coefficient of leak prediction according to the error rate in the transmission path / storage medium. Is characterized by.

Further, it is characterized in that it is provided with means for increasing or decreasing the number of sync codes to be inserted in one screen of a moving image signal according to the error rate in the transmission line / storage medium.

A moving picture decoding apparatus according to a third aspect of the invention comprises a subband dividing means for dividing a motion compensated prediction signal obtained by performing motion compensation on a reproduced moving picture signal into a plurality of subbands having different frequency bands. , The lowest frequency sub-band of the plurality of sub-bands, the motion compensation prediction signal of the lowest frequency from the sub-band dividing means at least once within a predetermined period for each region obtained by dividing the screen into a plurality of regions And a prediction error signal are added to decode the reproduced moving image signal, and the sub-bands other than the lowest frequency among the plurality of sub-bands are output from the sub-band dividing unit. A leak prediction signal obtained by multiplying a motion compensation prediction signal of a subband other than the lowest frequency by a coefficient smaller than 1 is added to the playback video signal to generate the playback video signal. And having a second decoding means.

A moving picture coding apparatus according to a fourth aspect of the present invention is a block dividing means for dividing an input moving picture signal into a plurality of blocks composed of a plurality of pixels, and the input moving picture signal is intra-coded for each block. And coding means for selectively switching to inter coding and coding, counting means for counting the number of screens in which each of the blocks of the input video signal is not continuously intra coded, and this counting means. And a control unit that performs control for forcibly intra-coding the block having a larger count value by the coding unit.

A moving picture coding apparatus according to a fifth aspect of the present invention is a block including a sync code added to each frame and a plurality of block lines when the frame is divided into a plurality of rectangular blocks composed of a plurality of pixels. It is characterized in that a code string of a syntax having a sync code added to the block group excluding the block group at the top of the screen is output to the group.

In the moving picture coding apparatus according to the sixth aspect of the present invention, the input moving picture signal is divided into a plurality of blocks composed of a plurality of pixels, and the input is performed for each block divided by the block dividing means. Coding means for coding a moving picture signal, local decoding means for decoding the coded information obtained by this coding means to generate a reproduced picture signal, and at least a reference picture signal for one screen are accumulated. Storage means, first calculating means for calculating a first error evaluation value indicating an error of each block of the reproduced image signal with respect to the input moving image signal, and each block of the reference image signal with respect to the input moving image signal And a second calculation means for calculating a second error evaluation value indicating the error, and a value obtained by adding a non-negative value to the first error evaluation value as a threshold value. Shi Reference value stored in the storage means for a block for which the second error evaluation value is larger than the threshold value, based on the determination result of the determination means. The image signal is rewritten with the reproduced image signal, and the encoding information of the block and rewriting information indicating that the block has been rewritten are output.

A moving picture coding apparatus according to a seventh aspect of the present invention includes a block dividing means for dividing an input moving picture signal into a plurality of blocks composed of a plurality of pixels, and the input for each block divided by the block dividing means. Coding means for coding a moving picture signal, local decoding means for decoding the coded information obtained by this coding means to generate a reproduced picture signal, and at least a reference picture signal for one screen are accumulated. Accumulating means, motion parameter detecting means for detecting and outputting a motion parameter between the input moving image signal and the reference image signal, and moving the reference image signal accumulated in the accumulating means based on the motion parameter. A motion-compensated prediction unit for compensating to create a motion-compensated predicted image signal, and a reference image signal stored in the storage unit by the motion-compensated predicted image signal are written. Changing means, first calculating means for calculating a first error evaluation value indicating an error for each block of the reproduced image signal with respect to the input moving image signal, and the motion compensation prediction image signal with respect to the input moving image signal. Second calculation means for calculating a second error evaluation value indicating an error for each block; and a value obtained by adding a non-negative value to the first error evaluation value as a threshold value, and the second error evaluation value. Is determined to be greater than the threshold value, and based on the determination result of this determination means, a block in which the second error evaluation value is greater than the threshold value is accumulated in the accumulation means. A reference image signal being rewritten by the reproduced image signal, and outputting the coding information of the block and rewriting information indicating that the block has been rewritten. And butterflies.

A moving picture decoding apparatus according to an eighth aspect of the invention is a decoding apparatus which decodes coding information obtained by coding a moving picture signal for each of a plurality of blocks composed of a plurality of pixels to generate a reproduced picture signal. Converting means, accumulating means for accumulating at least one screen of reference image signal, and motion compensating the reference image signal accumulated in the accumulating means based on a motion parameter between the moving image signal and the reference image signal. A motion-compensated prediction means for creating a motion-compensated predicted image signal, and a block for which rewriting information has been sent among the plurality of blocks, outputs the reproduced image signal of the block, and is stored in the storage means. A means for rewriting an image signal with the reproduced image signal, and outputting the motion compensation predicted image signal for a block to which the rewriting information is not sent. To.

Further, the decoding means divides the encoded information into a plurality of different components, performs different post-processing on each of these components, synthesizes the components, and outputs the synthesized image signal. And

The moving picture coding apparatus according to the ninth invention is such that the picture signal of the current screen is divided into a plurality of blocks composed of a plurality of pixels, and an error evaluation value from the reference picture is calculated for each block. 1) and a second step of detecting a motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value using at least the block whose error evaluation value exceeds the threshold value. Accordingly, in the third and subsequent steps, the motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value is detected by using the block whose error evaluation value exceeds the threshold value by the previous steps. Is characterized by.

A moving picture coding apparatus according to the tenth invention is
Block dividing means for dividing the input moving image signal into a plurality of blocks composed of a plurality of pixels, encoding means for encoding the input moving image signal for each block divided by the block dividing means, and this encoding means. A local decoding means for decoding the coded information obtained by generating a reproduced image signal, a storage means for storing a reference image signal for at least one screen, and the input moving image signal and the reference image signal. Motion parameter detecting means for detecting and outputting a motion parameter between them, and motion compensation predicting means for motion compensating the reference image signal accumulated in the accumulating means on the basis of the motion parameter to create a motion compensated predicted image signal. Means for rewriting the reference image signal stored in the storage means by the motion-compensated predicted image signal, and the moving image for the input moving image signal. A judgment means for judging whether or not an error evaluation value indicating an error for each block of the compensation prediction image signal is larger than a predetermined threshold value, and the error evaluation value is based on the judgment result of the judgment means. The reference image signal accumulated in the accumulating unit for a block larger than the value is rewritten with the reproduced image signal, and the encoding information of the block and rewriting information indicating that the rewriting has been performed for the block are output. Means, the encoding means detects an edge in each block of the input moving image signal by template matching,
The edge pattern index that maximizes the absolute value of the edge gain of the edge, the edge gain, and the average value of the pixel values in the block of the input moving image signal are encoded and output as the encoded information. And

The moving picture decoding apparatus according to the tenth invention is
Decoding means for decoding a coded information obtained by coding a moving image signal for each of a plurality of blocks made up of a plurality of pixels to generate a reproduced image signal, and a storage for storing a reference image signal for at least one screen Means for compensating the reference image signal stored in the storage means based on a motion parameter between the moving image signal and the reference image signal to create a motion-compensated prediction image signal; Regarding the block to which the rewriting information is sent among the plurality of blocks, the reproduction image signal of the block is output and the reference image signal accumulated in the accumulating unit is rewritten with the reproduction image signal, and the rewriting information is transmitted. And a decoding unit that outputs the motion-compensated prediction image signal, and the decoding unit outputs each block of the moving image signal as the encoding information. Input is the index of the edge pattern that maximizes the absolute value of the edge gain of the edge in the clock, the information obtained by encoding the edge gain and the average value of the pixel values in the block of the moving image signal, and input to the index. The reproduction image signal is generated by multiplying each element of the corresponding edge pattern by the edge gain and then adding the average value.

In the moving picture decoding apparatus according to the ninth or tenth aspect of the invention, a judging means for judging whether or not the coding information is normally decoded by the decoding means, and this judging means. According to the present invention, it further comprises means for prohibiting the rewriting regardless of the presence or absence of the rewriting information with respect to the block in which the encoded information is not normally decoded.

The moving picture coding / decoding apparatus according to the eleventh invention is characterized in that the motion parameter information is fixed-length coded and arranged in the picture header.

[0054]

In the first aspect of the present invention, a plurality of screens including a screen for intra-coding and a screen for inter-coding can be selectively set for a block whose difference between two adjacent screens is equal to or greater than a threshold value. As a result, even when the error rate of the transmission line / storage medium is higher than in the conventional case, the influence of the error can be suppressed as small as possible, and the influence of the error can be refreshed efficiently.

In the second aspect of the invention, when the encoded data obtained by compression-encoding the moving image signal is variable-length encoded and transmitted / stored, a checksum is added to the variable-length encoded code string to transmit / store. By performing the above, and detecting an error in the code string using the checksum in the moving picture decoding device, it is possible to improve the error detection accuracy in the transmission path / storage medium.

Further, by not outputting the decoding result in which an error has been detected, that is, by not displaying it on the screen, it is possible to suppress image quality deterioration due to the error.

In the third aspect of the invention, the input moving image signal is divided into a plurality of subbands, and only the lowest frequency subband is subjected to periodic refreshing by intra coding by several blocks, and leak prediction is performed for the other subbands. By performing the inter coding using the prediction signal obtained by performing the above, it is possible to suppress an increase in the code amount due to the periodic refresh.

Further, according to the error rate in the transmission line / storage medium, the area to be refreshed and the coefficient of leak prediction are increased or decreased by intra-encoding, and the number of synchronization codes to be inserted in one screen of a moving image signal is set. By increasing or decreasing, it is possible to improve the coding efficiency when the error rate is low, and avoid the image quality deterioration due to the error when the error rate is high.

According to the fourth aspect of the invention, the input moving image signal is divided into a plurality of blocks each having a plurality of pixels, and each block is provided with a coding circuit capable of selectively switching between intra coding and inter coding. , Each of the blocks of the input moving image signal continuously counts the number of screens that are not intra-coded, and blocks with a larger count value are forcibly ordered in order from a block with a longer period that has not been refreshed by intra-coding. By performing refresh automatically, it is possible to quickly recover the image quality deterioration in the still region.

According to the fifth aspect of the invention, when the frame is divided into a synchronous code in which the structure of the code string output from the moving picture coding device is added to each frame and the frame is divided into a plurality of rectangular blocks composed of a plurality of pixels. For a block group including a plurality of block lines of, the syntax has a sync code added to the block group excluding the block group at the top of the screen, and omits the sync code added to the block group at the top of the screen. Therefore, when the same synchronization recovery capability as the conventional code string syntax is provided, the total code amount of the synchronization code, that is, the redundant code amount for the synchronization recovery can be reduced,
Further, when the syntax is the same as the conventional code string syntax, the ability of the synchronization recovery of the redundant code amount for the synchronization recovery is further improved.

In the sixth aspect of the invention, when the conditional pixel replenishment is performed, the reference frame memory is rewritten with the reproduced image signal (local decoded signal), which is different from the conventional method in which the rewriting is performed with the input moving image signal. Even in the still region, when the coding distortion of the reference image signal stored in the reference frame memory is large, conditional pixel supplementation is performed if the output buffer has a margin, so that a still scene with little motion continues. In this case, the image quality in the still area can be improved and the error resilience when an error occurs in the still area can be improved.

Further, a value obtained by adding a non-negative value to the error evaluation value indicating the error of each block of the reproduced image signal with respect to the input moving image signal is used as a threshold value, and the error of each block of the reference image signal with respect to the input moving image signal is set. When the error evaluation value indicating is larger than the threshold value, the contents of the block of the reference image signal stored in the reference frame memory are rewritten and the block is rewritten to perform conditional pixel replenishment. By outputting the rewriting information indicating that the conditional pixel replenishment may not be performed when the conditional pixel replenishment is better, or conversely, the conditional pixel replenishment may be performed when the conditional pixel replenishment is better not performed. Therefore, appropriate conditional pixel replenishment can always be performed.

In the seventh invention, by combining the motion compensation prediction with the conditional pixel replenishment method according to the sixth invention, the same effect as the sixth invention can be obtained, and the motion compensation prediction cannot be performed. Since it suffices to perform conditional pixel replenishment only for the blocks, the coding efficiency is improved particularly in a scene including motion, as compared with the conventional conditional pixel supplement method.

In the eighth aspect of the invention, in the moving picture decoding apparatus, when the conditional pixel replenishment is performed, the reference frame memory is rewritten with the reproduced picture signal obtained by the local decoding. Different from the conventional method of rewriting with the reproduced image signal, conditional pixel supplementation is performed when the coding distortion of the reference image signal stored in the reference frame memo is large even in the still region. Therefore, when a still scene with a small amount of motion continues, the image quality in the still area is improved, and the error resilience when an error occurs in the still area is improved.

In the ninth invention, the first step of dividing the image signal of the current screen into a plurality of blocks composed of a plurality of pixels and calculating an error evaluation value from the reference image for each block, at least The method has a second step of detecting a motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value by using a block whose error evaluation value exceeds the threshold value, and a third step if required. After that, by using the block whose error evaluation value exceeds the threshold value by the previous step, the motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value is detected. Even if the global motion parameter of is included, when extracting a plurality of motion parameters from the screen, there is another motion parameter when extracting the parameter of each area. Remove the influence of the region, each can extract appropriate motion parameters.

In the tenth aspect of the invention, an edge in each block of the input moving image signal is detected by template matching in the moving image encoding device, and an edge pattern index that maximizes the absolute value of the edge gain of the edge is used. The edge gain and the average value of the pixel values in the block of the input moving image signal are encoded and output as encoded information, the encoded information is input in a moving image decoding device, and an edge pattern corresponding to the index is input. By multiplying each element of (1) by the edge gain and adding the average value to generate a reproduced image signal, a reproduced image of good quality without edge blurring can be obtained.

In the eleventh invention, the motion parameter information is fixed-length coded and arranged in the picture header, so that the influence of the error on the transmission line / storage medium can be reduced.

[0068]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) FIGS. 1 and 2 are block diagrams showing an embodiment of a moving picture coding apparatus and a moving picture decoding apparatus according to the first invention.

First, the moving picture coding apparatus shown in FIG. 1 will be described. In FIG. 1, an input moving image signal 101 input in units of frames is first divided into a plurality of blocks (macroblocks in this example) including a plurality of pixels by a blocking circuit 102. The input moving image signal of each divided macroblock is input to the motion compensation prediction circuit 103,
Motion compensation prediction is performed based on the motion vector with respect to the reference image signal stored in the frame memory 104. The motion compensation prediction circuit 103 uses the motion vector information MV.
And the prediction information P is output.

The mode selection circuit 105 is based on the prediction information P from the motion compensation prediction circuit 103 and the coding control information C from the coding control circuit 106, and for each block of the input moving image signal 101, as shown in FIG. ) Four coding modes M shown in
One mode is selected from ODE0, MODE1, MODE2, and MODE3, and mode information M indicating the selected coding mode is output to the coding control circuit 106.

Further, the mode selection circuit 105, as will be described later, includes a picture for which intra-frame coding (generally, intra-coding) and a picture for which intra-frame coding is performed for a block whose inter-frame difference is a threshold value or more. , A picture for which inter-frame predictive coding (generally, inter-coding) is selectively set for a block whose inter-frame difference is equal to or more than a threshold value. In addition,
The picture here corresponds to a screen, and corresponds to a frame in this example.

In the mode selection circuit 105, the first mode MO
When DE0 (Not Coded), that is, a mode in which no coding is performed, is selected, the mode information M indicating the mode MODE0 is input from the coding control circuit 106 to the variable length coding circuit 113 and variable length coding is performed. .

When the mode selection circuit 105 selects the second mode MODE1 (MC Not Coded), that is, the mode in which only the motion compensation of the reference image signal is performed by the motion vector and the coding is not performed, The mode information M indicating the mode MODE1 and the motion vector information MV are supplied from the coding control circuit 106 to the variable length coding circuit 11
3 and variable length coded.

Further, in the mode selection circuit 105, the third mode MODE2 (INTRA), that is, the intraframe coding mode, and the fourth mode MODE3 (INTRA + D).
QUANT), that is, in the intra-frame coding mode and the mode for changing the quantization step size is selected, the input moving image signal of each block divided by the blocking circuit 101 is the coding control circuit. Under the control of 106, it is input to a DCT (discrete cosine transform) circuit 108 via a switch 107 which is turned on only in MODE 2 or MODE 3, and is subjected to discrete cosine transform.
The DCT coefficient information obtained by the DCT circuit 108 is quantized by the quantization circuit 109.

The quantized DCT coefficient information is branched into two, and one of them is variable length coded by the variable length coding circuit 112. On the other hand, the other of the quantized DCT coefficient information is quantized by the inverse quantization circuit 110 and the IDCT (inverse discrete cosine transform) circuit 111, and the quantization circuit 109 and the DCT circuit 10 are quantized.
After sequentially performing the process reverse to the process of 8, the locally decoded signal is generated. This locally decoded signal is stored in the frame memory 104 as a reference image signal for encoding the input moving image signal of the next frame. This frame memory 104
The reference image signal from is input to the motion compensation prediction circuit 103.

The motion compensation prediction circuit 103 calculates and outputs the prediction information P, and when the mode selected by the mode selection circuit 105 is MODE1, the input moving image signal 10
1 and a reference image signal from the frame memory 104 (a local decoded signal one frame before) are obtained, and motion vector information MV is output.

The multiplexer 114 multiplexes the code string variably coded by the variable length coding circuits 112 and 113 based on the syntax shown in FIG. 6 described later. The multiplexed code string is sent to the output buffer 115 and smoothed to a constant rate. From the output buffer 115, buffer amount information B indicating the buffer amount is sent to the encoding control circuit 106. The encoding control circuit 106 performs encoding control based on the buffer amount information B so that the buffer amount is constant, specifically, the quantization step size in the quantization circuit 109 and the dequantization circuit 110. . The code string 116 smoothed to a constant rate by the output buffer 115 is output to a transmission line or a storage medium (hereinafter referred to as a transmission line / storage medium) via an output unit (not shown).

Next, the moving picture decoding apparatus shown in FIG. 2 will be described. In FIG. 2, a code string 121 output from the moving image coding apparatus of FIG. 1 and input via a transmission line / storage medium (not shown) is received by an input unit (not shown) and is temporarily stored in the input buffer 122. The demultiplexer 123 separates the code string stored in the input buffer 122 for each frame based on the syntax shown in FIG. 5, and outputs it to the variable length decoding circuits 124 and 125. The variable length decoding circuit 124 decodes the quantized DCT coefficient information encoded by the variable length encoding circuit 112 of FIG. The variable length decoding circuit 125 decodes the mode information M and the motion vector information MV coded by the variable length coding circuit 113 of FIG.

The mode information M decoded by the variable length decoding circuit 125 is input to the mode determination circuit 126. In the mode determination circuit 126, the mode information M indicates the first MODE.
When 0 is displayed, the reproduced image signal stored in the frame memory 131 as the reference image signal is switched to the switch 13
When the mode information M indicates the second mode MODE1, the motion vector information MV indicates one of the reproduced image signals stored in the frame memory 131. Select the image signal of the area in which the switch 130 is present and output the moving image signal 1
It is output as 32.

Further, the mode decision circuit 126 indicates that the mode information M is the third mode MODE2 or the fourth mode MO.
When DE3 is indicated, the switch 127 is turned on, the quantized DCT coefficient information decoded by the variable length decoding circuit 124 is inversely quantized by the inverse quantization circuit 128, and the inverse DC is obtained.
A reproduced image signal is generated by performing an inverse discrete cosine transform process by a T (inverse discrete cosine transform) circuit 129. This reproduced image signal is output as the output moving image signal 132 via the switch 130 or is stored in the frame memory 131 as a reference image signal.

FIG. 4 is a flow chart showing the operation of the mode selection circuit 105 every time the input moving image signal 101 in this embodiment is input for one macroblock. First,
By the picture type (PICTURE TYPE) described in the picture header (PCT) of FIG. 3B,
As shown in FIG. 5B, a picture INTRA that is intraframe-encoded for the first frame, and a picture CR that intraframe-encodes a block whose inter-frame difference is a threshold value or more,
The discrimination of the picture MC in which the inter-frame difference is greater than or equal to a threshold value is inter-frame predictive-coded is determined (step S
101). In this embodiment, as shown in FIG.
In the R picture, the difference between the input moving image signal 101 and the locally decoded image signal is a threshold value (in this example, the threshold value is "2").
Intra-frame coding of the above macroblocks, and MC
The picture employs a method of intraframe coding a macroblock that becomes Not Coded and periodically refreshing it.

If the result of determination in step S101 is a CR picture, the prediction information P is compared with the first threshold th1 (step S102). If P is greater than or equal to th1, the process proceeds to the next determination (step S103), and if not, "0" is assigned to MODE (step S10).
8) The process ends. In step S103, the prediction information P is compared with the second threshold value th2. If P is greater than or equal to th2, the process proceeds to the next determination (step S104). If not, "1" is assigned to MODE (step S103). S10
7) and the process ends. In step S104, the quantization step size change information C. dqua
nt is checked, if it is ON, “4” is assigned to MODE (step S105), if it is OFF, “3” is assigned to MODE (step S106), and the process is ended.

If the result of the determination in step S101 is an MC picture, the prediction information P is compared with the first threshold th1 (step S109). If P is smaller than th1, the process proceeds to the next determination (step S110), and if not, "1" is substituted into MODE (step S1).
15) and ends. In step S110, Frame
_No mod MB_Column value, that is, the vertical address i of the macroblock and the frame counter value F
The frame_No is compared with the remainder obtained by dividing the macroblock number MB_Column in the vertical direction of one picture, and if they are equal, the process proceeds to the next determination (step S111), and if they are not equal, “0” is substituted into MODE (step S114),
finish. In step S111, the quantization step size change information C. Check dquant, if it is ON, assign "4" to MODE (step S112), and if it is OFF, "3" to MODE
Is substituted (step S113), and the process ends.

Here, the prediction information P is the average of the absolute value difference between the input moving image signal and the locally decoded signal for the luminance signal for each macroblock, and is calculated by the following equation.

[0086] Also, the relationship of th1 <th2 is established.

FIG. 5 shows an example of a combination of CR picture and MC picture. In FIG. 5A, the CR picture and the MC picture are repeated for each frame.
With such a combination of pictures, refresh can be efficiently performed according to the movement of the screen. As a combination method, as shown in FIG. 5 (b), a method of using two MC pictures for one CR picture, or as shown in FIG. 5 (c), a CR picture and an MC picture are selected according to the image. It is also possible to adopt a method of adaptively switching the image according to the movement of the image.

Next, the syntax of the coded information in this embodiment shown in FIG. 6 will be described. The picture layer includes a picture header, an upper layer, and a lower layer, as shown in FIG. In the picture header, a synchronization code and a parameter for each picture are described. Of the coded information, what is considered to be important information is described in the upper layer, and normal coded information is described in the lower layer. This is because the upper layer is described immediately after the sync code in the picture header, and thus the probability of being affected by an error is smaller than that in the lower layer.

More specifically, in this embodiment, the picture header describes the sync code SYNC, the display order TR, the picture type PCT, and the quantization step size QUANT, as shown in FIG. 6B. In the upper layer, as shown in FIG. 6C, the mode information MODE of the macroblock and the D of the macroblock for intra-frame coding are set.
The C coefficient DC INTRA is described. Further, as shown in FIG. 6D, the motion vector MV, the quantization step size change information DQUANT, and the DCT coefficient TCOF and EOB are described in the lower layer.

As another transmission method, it is conceivable to prioritize the error correction code or the error detection code on the coded information of the upper layer to strengthen the tolerance against the error as compared with the lower layer.

As described above, according to the present embodiment, the picture (intra-coding screen) for which intra-frame coding is performed on the block whose inter-frame difference (difference between two adjacent screens) is greater than or equal to the threshold, and the inter-frame coding The error rate of the transmission path / accumulation medium is higher than that of the conventional one by enabling the setting of the picture for which the inter-frame predictive coding (the screen for inter-coding) is selectively set for the block whose difference is equal to or more than the threshold. Even in such a case, there is an advantage that the influence of the error can be suppressed to a minimum and the influence of the error can be refreshed efficiently.

In this embodiment, the example of the moving picture coding / decoding apparatus in units of frames has been shown, but it is also applicable when the input moving picture signal is an interlaced signal and in units of fields. Needless to say.

(Embodiment 2) FIG. 9 is a block diagram showing an embodiment of a moving picture coding / decoding apparatus according to the second invention.

In the moving picture coding apparatus 1, the input moving picture signal 201 is input to the moving picture coding unit 202 for performing so-called signal source coding. Video coding unit 202
Is a motion-compensated prediction error signal which is a difference signal between the prediction signal obtained by motion compensation of the locally decoded signal of the previous frame and the input moving image signal 201, and is subjected to the discrete cosine transform (DCT). , Quantizes the obtained DCT coefficient information and outputs it. The output of the moving picture coding unit 202 is
The variable length coding unit 203 performs variable length coding for each predetermined coding unit and outputs the code string. Here, the coding unit is, for example, a group of a plurality of motion vector information,
Alternatively, it is a collection of a plurality of pieces of DCT coefficient information.

The code string output from the variable length coding unit 203 is input to the checksum calculation unit 204, and the checksum is calculated for each processing unit including a plurality of variable length codes. Specifically, the code string of the processing unit is divided into a predetermined number of bits, for example, 4 bits each, and the values of the codes of the respective divisions are added, and the value of the lower predetermined number of bits of the added value, for example, the lower 4 The value of the bit is generated as a checksum.

The checksum thus generated by the checksum calculation unit 204 is input to the checksum addition unit 205, where it is added after each coding unit of the code string output from the variable length coding unit 203. , To the transmission path / storage medium 206.

On the other hand, in the moving picture decoding apparatus 2, the code string input via the transmission path / accumulation medium 206 is variable length decoded by the variable length decoding unit 207, and at the same time,
The checksum check / error detection unit 208 calculates a checksum and compares it with the checksum added to the input code string to detect an error.

In the checksum checking / error detecting section 208, when the calculated checksum matches the checksum added to the input code string, the variable length decoding section 20
The data decoded in 7 is directly used as the moving picture decoding unit 20.
9, the reproduced image signal 210 is output by being decoded and being decoded. The moving picture decoding unit 209 performs the reverse processing of the moving picture coding unit 202. For example, the quantized DCT coefficient information from the variable length decoding unit 207 is dequantized, and further the inverse discrete cosine transform (inverse DCT) is performed. ) Is performed to generate a motion-compensated prediction error signal, and this is added to a prediction signal obtained by motion-compensating the reproduced image signal of the previous frame to generate a reproduced image signal 210.

Even if the moving picture coding apparatus 1 adds a checksum to the code string of the variable-length code, further adds and outputs an error correction code, and the moving picture decoding apparatus 2 performs error correction. Good.

A simple example of checksum generation in the checksum calculation unit 204 and check / error detection in the checksum check / error detection unit 208 will be described below with reference to FIGS. 10 and 11. FIG. 10 shows examples of variable length codes for various input values. FIG. 11A shows a code string obtained by variable-length coding the input values 1, 3, 5, 0, 2, 3, 1. Checksum bit number is 4
If it is a bit, the checksum calculation unit 204 in FIG.
The code string of (a) is divided into 4 bits from the beginning, and
As shown in (b), the value of the code of each delimiter is added. If the number of delimiter bits is insufficient, the rest is regarded as 0 and added. The lower 4 bits of the added value thus obtained are added by the checksum adding unit 205 as a checksum after the code string of the variable length code and output.

Now, assuming that an error occurs in the transmission path / storage medium at the 8th bit of the code string to which the checksum is added as shown in FIG. 11A, this code string is transmitted to the moving picture decoding apparatus 2. The result obtained by the variable length decoding in the variable length decoding unit 207 is 1, 3, 1, as shown in FIG.
It becomes 1,1,0,2, and a completely random decoding result is obtained after the position where the error occurs. Further, since the length of the variable length code is also erroneously recognized due to an error, the position of the checksum checked by the checksum checking / error detecting unit 208 also differs. However, when the checksum calculated by the checksum check / error detection unit 208 and the value which is recognized as the checksum added by the checksum addition unit 205 in the moving picture coding apparatus 1 are compared with each other, it is found that they match in this case. Since it does not, it can be seen that an error occurred on the way.

An error can be detected with a high probability by checking the checksum, but if the code following the checksum is known, by checking it,
Further, the accuracy of error detection can be improved. For example,
When the synchronization code is added for each coding unit, the synchronization code always comes after the checksum. Therefore, if the checksums match and the synchronization code follows them, the variable-length code is correctly decoded both in terms of length and value.

As described above, according to this embodiment, when the moving picture coding apparatus compresses and codes the moving picture signal, the variable length coded data is transmitted / stored. By adding a checksum for transmission / accumulation and detecting a code string error in the video decoding device using the checksum, it is possible to improve the accuracy of error detection on the transmission path / accumulation medium. .

(Embodiment 3) FIG. 12 is a block diagram showing another embodiment of the moving picture coding / decoding apparatus according to the second invention, and further has the configuration of Embodiment 2 shown in FIG. The feature is that an error control unit 211 is added. The error control unit 211 has a checksum check / error detection unit 208.
The error detection signal from is monitored to obtain an approximate error rate.If the error rate is higher than a certain threshold, (a) increase the number of checksum bits or numbers, (b) increase the number of synchronization codes, Error control is performed by (c) decreasing the leak prediction coefficient and (d) increasing the area for periodic refreshing.
In this case, the error rate threshold may be set manually. Further, if this threshold value is added to the output data of the moving picture coding apparatus and transmitted / stored, the error control can be matched between the moving picture coding apparatus and the moving picture decoding apparatus.

As described above, according to the present embodiment, the accuracy of error detection on the transmission line / storage medium can be improved as in the second embodiment, and the decoding result in which an error is detected is not output. That is, by not displaying the image on the screen, it is possible to suppress image quality deterioration due to an error.

(Embodiment 4) FIG. 13 is a block diagram showing an embodiment of a moving picture coding apparatus according to the third invention.

The input moving image signal 301 is input to the first subband dividing section 302 and divided into a plurality of subbands having different frequency bands. On the other hand, the input motion image signal 301 which is the image signal of the current frame and the reference image signal 316 which is the image signal of the previous frame are input to the motion compensation unit 302, and the result of analysis of the motion of the input moving image signal 301 with respect to the reference image signal A motion compensation prediction signal is output by performing motion compensation on the reference image signal based on (motion parameter, motion vector).

The motion-compensated prediction signal is input to the second subband division unit 304, and like the input moving image signal 301,
It is divided into a plurality of subbands having different frequency bands. Then, in the motion compensation prediction signal divided into these subbands,
In the leak prediction unit 305, a coefficient smaller than 1 (leak prediction coefficient) is multiplied. Here, in order to prevent the direct current component of the reproduced image signal from deviating, it is important that the motion compensation prediction signal of the lowest frequency subband is not multiplied by the leak prediction coefficient. For the lowest frequency sub-band, the screen is divided into a plurality of areas (may be blocks), and the switch 306 is turned off at a constant cycle for each area, so that the prediction signal 307 supplied to the subtractor 308 is " It is set to 0 ", the intra-frame coding process that does not use the information of the previous frame is performed, and the periodic refresh is performed.

That is, the first subband division unit 302
The input moving image signal of the current frame divided into subbands is input to the subtractor 308, where the prediction signal 307 is input.
A prediction error signal 309, which is an error between and, is obtained. The prediction error signal 309 is quantized by the quantizer 310 and output as moving image coded data 311 to a transmission line / storage medium (not shown). Further, the prediction error signal quantized by the quantizer 310 is inversely quantized by the inverse quantizer 312, and is added to the prediction signal 307 by the adder 313.
It is stored in the frame memory 313 as a subband local decoded signal. Here, with respect to the lowest frequency sub-band, the switch 30 has a constant cycle for each area of the screen as described above.
When 6 is turned off and the prediction signal 307 is set to "0", the intraframe coding process is performed, and the periodic refresh is performed.

FIG. 14 is a block diagram showing the structure of a moving picture decoding apparatus corresponding to the moving picture coding apparatus of FIG. The moving picture coded data 321 input from the moving picture coding apparatus of FIG. 13 via the transmission path / storage medium is dequantized by the dequantizer 322 and added to the prediction signal 324 by the adder 323. , And is written in the frame memory 325 as a subband decoded signal. Frame memory 325
The subband decoded signal read from the subband combining unit 326 combines the decoded signals of all subbands,
The reproduced image signal 327 is output.

The reproduced image signal 327 is transmitted to the motion compensation unit 3
Motion compensation is performed at 28, sub-band division is performed at sub-band division unit 329, and a leak prediction coefficient is multiplied at leak prediction unit 330. The leak prediction unit 330 does not multiply the lowest frequency subband by the leak prediction coefficient, like the leak prediction unit 305 of FIG. 13. Then, for the lowest frequency sub-band, the switch 331 is turned off and the prediction signal 324 supplied to the adder 323 is set to "0", thereby performing periodic refreshing.

As described above, according to this embodiment, the input moving image signal is divided into a plurality of subbands, and only the lowest frequency subband is subjected to periodic refreshing by intra-frame coding processing (intra coding) by several blocks. For other subbands, interframe coding (inter coding) is performed using a prediction signal obtained by performing leak prediction, so that an increase in code amount due to periodic refresh can be suppressed.

FIG. 15 is a block diagram showing the moving picture coding / coding according to the present embodiment.
S / N of decoded image with respect to original image when decoding is performed
Is shown. As a countermeasure against errors, a 4-bit checksum is added to each coding unit, and a 20-bit synchronization code is added to each significant subband divided into 10 bands.
Further, this is an example in which 4% of the screen is refreshed per frame and leak prediction with a leak prediction coefficient of 0.8 is performed. The total code amount of the checksum and the synchronous code is about 5%.

By taking such an error countermeasure, the average S / N is lowered by 0.4 dB. If the error rate is 10 ^-4, the average S / N will drop by 8 dB if no error countermeasure is taken,
By taking this error countermeasure, the decrease of 3 dB can be suppressed. This 3 dB S / N reduction is hardly perceptible visually. That is, the error rate of the transmission line / storage medium is 10 ⁻⁴ by implementing the error countermeasure as in this embodiment.
However, it is possible to suppress the deterioration of the image quality of the decoded image due to an error to the extent that it is almost imperceptible.

(Embodiment 5) FIG. 16 is a block diagram showing an embodiment of a moving picture coding apparatus according to the fourth invention. The input moving image signal 401 divided into a plurality of blocks including a plurality of pixels is supplied to the motion vector detection circuit 402, the subtractor 403, the mode determination circuit 404, and the selector 408.

In the motion vector detection circuit 402, the motion vector between the input moving image signal 401 and the reference frame (reference image signal) stored in the reference frame memory 405 having the variable delay function for motion compensation is 16 × 16 pixels. It is detected for each configured macroblock (MB).

The subtractor 403 uses the reference frame memory 40.
A prediction error signal 406, which is the difference between the motion-compensated prediction signal 416 and the input moving image signal 401, is obtained for each macroblock, and the selector 408 and the mode determination circuit 4 are used.
Supply to 04.

The mode determination circuit 404 calculates the power of the prediction error signal 406 input from the subtractor 403 for each block as described later, and compares it with the result of calculating the power of the AC component of the input moving image signal 401. Then, it is determined whether the block is intra-frame encoded or inter-frame encoded. Mode information 407, which is output from the mode determination circuit 404 and indicates whether to perform intra-frame coding), intra-frame coding, or inter-frame coding (inter coding), is supplied to selectors 408 and 409.

The selector 408 selects and outputs the input moving image signal 401 in the case of intra-frame coding, and selects and outputs the prediction error signal 406 in the case of inter-frame coding. The signal output from the selector 408 is the DCT
The circuit 410 performs the discrete cosine transform and the DCT coefficient, and then is input to the quantization circuit 411 and the quantization step size information 42 given from the control circuit 424.
It is quantized with a quantization step size according to 5. On the other hand, the selector 409 outputs “0” in the case of intra-frame coding, and the motion compensation prediction signal 4 supplied from the reference frame memory 405 in the case of inter-frame coding.
16 is output.

The DCT coefficient 418 quantized by the quantization circuit 411 is inversely quantized by the inverse quantization circuit 412, and further reproduced by the inverse DCT circuit 413.
The signal supplied from the selector 409 is added by the addition circuit 414 to form a local decoded signal 415, which is stored in the reference frame memory 405 as a reference image signal.

The mode information 417 indicating the prediction mode from the mode determination circuit 404, the DCT coefficient 418 quantized by the quantization circuit 411, and the motion vector detection circuit 40.
The motion vector information 419 detected in 2 is variable-length coded in the variable-length coding circuits 420, 421, and 422, respectively, and then multiplexed in the multiplexer 423, and further transmitted to the transmission path / storage medium via the buffer 427. Sent out.

Further, in the multiplexer 423, the control circuit 42 is used, as described in the seventh embodiment described later.
The sync code from the sync adder circuit 426 controlled by 4 is added.

FIG. 17 is a block diagram showing an internal structure of the mode determination circuit 404 in FIG. In FIG. 17, the intra-frame / inter-frame determination circuit 431 includes an input moving image signal 401 and a prediction error signal 4 from the subtractor 403.
06 is input. In-frame / inter-frame determination circuit 4
Reference numeral 31 compares the activities of the input moving image signal 401 and the prediction error signal 406 for each block, and determines whether the block is intra-frame encoded or inter-frame encoded. As the activity, for example, the AC power of the input moving image signal 401 and the signal power of the prediction error signal 406 are used. The intra-frame / inter-frame determination circuit 431 outputs a result indicating whether intra-frame encoding or inter-frame encoding is performed, and the result is input to one input end of the 2-input logical sum circuit 437.

The counter group 432 is composed of the same number of counters as the number of macroblocks (MB) in one frame, and the MB address 433 indicating the address information of the macroblock is inputted to the MB group 43.
The counter corresponding to 3 counts up. When the processing for one frame is completed, the count value of each counter (MB count value) in the counter group 432 is supplied to the refresh MB selection circuit 434.

The refresh MB selection circuit 434 selects a predetermined number of MBs from the MBs having a large count value as refresh MBs and supplies the addresses of the refresh MBs to the refresh MB address memory 435. The refresh MB selection circuit 434 refreshes M
When selecting B, if the count values are the same, priority is given in raster order from the upper left of the screen. In the next frame, the address stored in the refresh MB address memory 435 is compared with the MB address 433 in the MB address comparison circuit 436. If they match each other, it means that the MB is a refresh MB. A signal indicating that intraframe coding is to be performed is supplied to the input terminal.

When the signal indicating that the MB is intra-frame encoded is supplied from either the intra-frame / inter-frame determination circuit 431 or the MB address comparison circuit 436, the logical sum circuit 437 receives the counter group 432. To the selectors 407 and 409 shown in FIG. 16 while resetting the counter corresponding to the MB and resetting the counter corresponding to the MB.

The operation of this embodiment will be described with reference to FIG.
18, (a) shows the value of each counter of the counter group 432 in association with the frame number of the input moving image signal 401, and (b) shows the state of the image of each frame. In FIG. 18B, the shaded MB at the upper right is the intra-frame / inter-frame determination circuit 4 of FIG.
The MB determined to perform intra-frame coding (intra-coding) in step 31 is the MB indicated by the upper left diagonal line,
This is an MB that has been determined to be a refresh MB by the refresh MB selection circuit 434. Here, 2 per frame
2 MB is represented as being forcibly refreshed.

As described above, according to this embodiment, the input moving image signal is divided into a plurality of blocks made up of a plurality of pixels, and each block can be selectively switched between intra coding and inter coding. Control for counting the number of frames (the number of screens) in which each block of the input moving image signal 401 is not continuously intra-coded, and forcibly performing intra-coding on the block having a larger count value. I do. That is, by forcibly performing refresh in order from a block having a long period that has not been refreshed by intra-encoding, the effect of quickly recovering the image quality deterioration in the still region can be obtained.

(Sixth Embodiment) FIG. 19 is a block diagram showing another embodiment of the moving picture coding apparatus according to the fourth invention. In FIG. 16, the inter-frame prediction error signal is encoded, whereas in the present embodiment, the inter-frame prediction signal 416 from the reference frame memory 405 and the input moving image signal 40.
1 is switched via the switch 429 between the DCT circuit 410 and the quantizing circuit 411, and the signal obtained by intra-frame coding is switched to the reproduced signal 428 reproduced through the dequantizing circuit 412 and the inverse DCT circuit 413. The image signal is obtained and stored in the reference frame memory 405.

In this case, in the mode determination circuit 404, the prediction signal 416 is input instead of the prediction error signal 406 in the intra-frame / inter-frame determination circuit 431 in FIG. 17, and the prediction signal 416 and the input moving image signal 401 are input. The error evaluation value of is calculated and compared with a predetermined threshold value. If the error evaluation value is larger than the threshold value, the intraframe coding mode is set.

Also in this embodiment, the same effect as that of the fifth embodiment can be obtained.

(Embodiment 7) An embodiment according to the fifth invention will be described with reference to FIGS. The configuration of the moving picture coding apparatus according to the present embodiment may be basically the same as that of the fifth embodiment shown in FIG. FIG. 20 shows the syntax of the code string in this embodiment. In this embodiment, FIG.
As shown in FIG. 20B, a synchronization code is added as shown in FIG. 20A to each slice composed of one or a plurality of MB lines. The synchronization code is added to the code string output from the multiplexer 423 by the synchronization code addition circuit 426 of FIG.

FIG. 20A shows an arrangement of data of a code string for one frame, which is followed by a frame synchronization code PS for obtaining the head frame synchronization, followed by header information P of a picture.
H, header information SH of the first slice, encoded data slice # 1 of the first slice, synchronization code SS for obtaining synchronization of the second slice, header information SH of the second slice, second slice Encoded data slice # 2
Data is sequentially assigned as shown in FIG. The same applies to the third and fourth slices as the second slice.

In the example of FIG. 20B, the number of slices is 1
You can change up to ~ 9. As the sync code of each picture, a code in which M bit “0” is added to N bit UW is used as shown below.

[0135] Further, as the synchronization code of each slice, a code obtained by adding a value of SVP (slice vertical position information) to which the head MB of the slice expressed in M bits is added to UW (unique word) of N bits as shown below. Has been. For example,
In the case of the second slice of FIG. 20 (b), it becomes as follows.

[0136] It should be noted that 0 ... 0 is an arbitrary 0 run, and its length (run length) may be determined in consideration of other codes.

In this embodiment, since the slice is composed of one or a plurality of MB lines, the position of each MB on the screen can be specified from the SVP of the slice. 1M
The number of MBs included in the B line is determined in a layer higher than the picture. Here, as can be seen from FIG. 20, since the SVP of slice # 1 is always “0”, the position of the MB can be specified without sending the SVP value again. Therefore, the sync code of slice # 1 is omitted.

As described above, in this embodiment, the synchronous code (PS) in which the structure of the code string output from the moving picture coding apparatus is added to each frame, and the plurality of rectangular blocks each including a plurality of pixels in the frame. A block group including a plurality of block lines when divided by (in this example, slice # 1,
, etc.) is added to the block group (slice # 2, ...) Excluding the block group (slice # 1, in this example) at the top of the screen, and the syntax has the synchronization code SS. . That is, the sync code added to the block group at the top of the screen is omitted.

Therefore, in the case of having the same synchronization recovery capability as the conventional code string syntax shown in FIG. 21 or 22, the total code amount of the synchronization code, that is, the redundant code amount for the synchronization recovery is reduced. When the syntax is the same as that of the conventional code string, there is an advantage that the redundancy code amount for the synchronization recovery is further improved in the synchronization recovery capability.

(Embodiment 8) FIG. 23 is a block diagram showing an embodiment of a moving picture coding apparatus according to the sixth invention. The input moving image signal 501 is input to the encoding circuit 502. The encoding circuit 502 encodes the input moving image signal 501 for each block composed of a plurality of pixels, and supplies the obtained code string (bit stream) to the output buffer 508. The output buffer 508 transfers the code string 510 to the transmission line /
The data is output to the storage medium at a predetermined rate and the buffer amount (retention amount) is supplied to the rate control circuit 509. The rate control circuit 509 supplies the control circuit 502 with a control signal for preventing the output buffer 508 from becoming empty or full. The coding circuit 502 includes a block selection circuit 503, a reference frame memory 504, a conversion circuit 505, a reproduction circuit 506, and a block coding circuit 5.
07.

FIG. 24 shows the configuration of the block selection circuit 503. The block selection circuit 503 divides the input moving image signal 501 into a plurality of blocks composed of a plurality of pixels, and the input moving image signal 501 of the current frame is divided into blocks.
And the reproduction image signal from the reproduction circuit 506 has the first degree of similarity.
Is calculated in the similarity calculation circuit 511. As the similarity, for example, an error signal power (error _n ) between the input moving image signal 501 and the reproduced image signal is calculated, and if the error signal power is small, it is determined that the similarity is large. The error signal power calculated in the similarity calculation circuit 511 is supplied to the adder circuit 513, and the non-negative (0 or positive value) value N supplied from the rate control circuit 509 is added to result in error _n + N. The threshold value th is supplied to one input terminal of the comparison circuit 514.

In the second similarity calculation circuit 512, the similarity between the input moving image signal 501 of the current frame and the reference image signal from the reference frame memory 504 can be calculated in block units in the same manner as in the error signal power between the two signals. (Error
_p )), and the error signal power is calculated as
4 is supplied to the other input terminal.

In the comparison circuit 514, the error signal power (error _p ) supplied from the similarity calculation circuit 512.
Is the threshold value th = er supplied from the adding circuit 513.
ror _n + N. Based on the comparison result of the comparison circuit 514, the error signal power is larger than the threshold value by the selector 515 (error _p > error _n +).
N), that is, when the conditional pixel supplement is performed, the reproduced image signal is selected, and when the error signal power is less than or equal to the threshold value (error _p ≤ error _n + N), that is, when the conditional pixel supplement is not performed. The reference image signal is selected and output to the reference frame memory 504, and the reference frame memory 504 is updated accordingly.

In the conversion circuit 505, the block selection circuit 50
The input moving image signal blocked in 3 is supplied, and for example, as shown in FIG.
After the signal is converted into a DCT coefficient by, the DCT coefficient is quantized by the quantization circuit 522 according to the quantization step size set by the rate control circuit 509 and output. The quantized signal is supplied to the reproduction circuit 5060 and the block encoding circuit. In the reproduction circuit 506, the reproduction signal of the block is obtained through the reverse process of the conversion circuit 505. That is, FIG.
As shown in (b), the quantized DCT coefficient from the conversion circuit 505 is the inverse quantization circuit 524 and the inverse DCT circuit 5.
The signal is reproduced via 25 and the reproduced signal is supplied to the block selection circuit 503.

In the block coding circuit 507, the quantized DCT coefficient (coding information), the block address, and the error signal power (error _p ) are set to the threshold value (th =
If it is larger than error _n + N), side information such as rewriting information indicating that the reference image signal of the block stored in the reference frame memory 504 has been rewritten is multiplexed and then output to the output buffer 508.

FIGS. 26 (a) and 26 (b) are block diagrams showing another configuration example of the conversion circuit 505 and the reproduction circuit 506 in FIG. The conversion circuit 505 and the reproduction circuit 506 shown in this example are, for example, BTC (Block Truncation Co).
ding: Kishimoto et al., "Block Coding Method for Still Images", Theoretical Theory (B), J62-B, 1, pp. 17 to 24), which is a method for approximating and coding an image signal.

That is, first, in the conversion circuit 505 shown in FIG. 26A, the approximate image parameter detection circuit 53
In step 1, the block-type input moving image signal supplied from the block selection circuit 503 of FIG. 23 is classified into a plurality of clusters, and each pixel is rewritten with the classified cluster representative value to obtain an approximate image signal. The quantization circuit 532 quantizes the parameters such as the representative value of each cluster and the cluster classification information supplied from the approximate image parameter detection circuit 531.

On the other hand, the reproducing circuit 506 shown in FIG. 26 (b).
In the above, the inverse quantization circuit 534 restores the representative value of each cluster and the cluster classification information, and the approximate image generation circuit 535 reconstructs the approximate image signal based on these.

As described above, in this embodiment, when the conditional pixel replenishment is performed, the reference frame memory 504 is set to the reproducing circuit 5.
The reconstructed image signal (locally decoded signal) obtained from No. 06 is used for rewriting. Therefore, unlike the conventional method of rewriting with the input moving image signal, the reference image signal stored in the reference frame memory 504 even in the still region. If the coding distortion of (1) is large, conditional pixel supplementation is performed if the output buffer 508 has a margin. Therefore, when a still scene with a small amount of motion continues, there are advantages that the image quality in the still area is improved and the error resilience is improved when an error occurs in the still area.

Further, in the present embodiment, error _p > error is set as a condition for judging whether or not conditional pixel replenishment is performed.
_In the case of _n + N, that is, a value obtained by adding a non-negative value N to the error evaluation value (error _n ) indicating the error for each block of the reproduced image signal with respect to the input moving image signal calculated by the first similarity calculation circuit 511. The threshold value th is set, and an error evaluation value (erro) indicating the error of each block of the reference image signal with respect to the input moving image signal calculated by the second similarity calculation circuit 512.
When r _p ) is larger than the threshold value th, the contents of the block of the reference image signal accumulated in the reference frame memory 504 are rewritten and the rewriting of the block is performed to perform conditional pixel replenishment. The rewriting information shown is output.

In the conventional conditional pixel replenishment method, conditional pixel replenishment is performed when error _p > th '(th' is a non-negative fixed value). Therefore, conditional pixel replenishment is preferable. If good (error _p ＞ er
If there is a case where conditional pixel replenishment is not performed on (or r _n )) or conversely it is better not to perform conditional pixel supplementation (er
Although there is a case where conditional pixel replenishment is performed for error _p <error _n ), according to the present embodiment, such a problem is solved, and appropriate conditional pixel replenishment can always be performed.

(Embodiment 9) FIG. 27 is a block diagram showing an embodiment of a moving picture coding apparatus according to the seventh invention. The present embodiment is a combination of the conditional pixel supplement method shown in the eighth embodiment and the motion compensation prediction, and the portions corresponding to those in FIG.

The input moving image signal 501 is encoded by the encoding circuit 502.
Is input to The encoding circuit 502 encodes the input moving image signal 501 for each block composed of a plurality of pixels and outputs a code string (bit stream). In the motion compensation loop 540, the original image signals of the current frame and the reference frame are stored in the frame memory 541,
The motion parameter detection circuit 542 detects a motion parameter (motion vector) between the current frame and the reference frame between the original image signals. This motion parameter may be calculated for each block, or may be calculated as a so-called global motion parameter of the entire screen. The motion compensation prediction circuit 544 uses the motion parameter detection circuit 542.
Based on the motion parameter supplied from the reference frame memory 545, the motion compensation prediction value is generated using the reproduced image signal of the reference frame stored in the reference frame memory 545.

The block selection circuit 543 is configured as shown in FIG. 24 similarly to the block selection circuit 503 of FIG. 23, and the input moving image signal 501 of the current frame is block by block.
Is calculated in the similarity calculation circuit 511, and when the error signal power is larger than the threshold th, the reproduced image signal from the reproduction circuit 506 has the error signal power smaller than the threshold th. If is smaller, the motion-compensated predicted image signal is output via the selector 515, and the non-reference area of the reference frame memory 545 is updated via the motion-compensated prediction circuit 544. The block encoding circuit 507 multiplexes DCT coefficients and block addresses, and side information such as rewriting information and motion parameters, and outputs the multiplexed information.

As described above, in this embodiment, the same effect as that of the eighth embodiment can be obtained by combining the motion compensation prediction with the conditional pixel supplement method described in the eighth embodiment.
Since it is necessary to perform conditional pixel supplementation only on blocks that cannot be predicted by motion compensation prediction, there is an advantage that the coding efficiency is improved especially in a scene including motion, as compared with the conventional conditional pixel supplement method.

(Embodiment 10) In this embodiment, a motion compensation loop 550 shown in FIG. 28 is used instead of the motion compensation loop 540 shown in FIG. In the motion compensation loop 550 shown in FIG. 28, the original image signal of the current frame is stored in the frame memory 551, and in the motion parameter detection circuit 552, the original image signal of the current frame and the reference frame memory 555 are stored. Parameter (motion vector) between the reproduced image signals of the reference frame
Is detected. This motion parameter may be calculated for each block, or may be calculated as a global motion parameter for the entire screen. The motion compensation prediction circuit 554 generates a motion compensation prediction value using the reproduced image signal of the reference frame stored in the reference frame memory 555 based on the motion parameter supplied from the motion parameter detection circuit 552.

Here, the detection of motion parameters and the motion compensation based on them may be global motion parameter detection and global motion compensation for the entire image, not for each block. Also, the block size for motion compensation and the block size for encoding may be different.

(Embodiment 11) FIG. 29 is a block diagram showing an embodiment of a moving picture decoding apparatus according to the eighth invention.
This corresponds to the moving picture coding apparatus according to the ninth embodiment shown in FIG. The code string (bit stream) output from the moving picture coding apparatus in FIG. 27 is input to the decoding circuit 562 via the input buffer 561. Decoding circuit 562
In the block decoding circuit 563, the multiplexed bitstream is first separated, and then the side information including the motion parameter and the DCT coefficient information of the signal in the block are decoded and output. The reproducing circuit 564 is configured as shown in FIG. 25 (b) or FIG. 26 (b) similarly to the reproducing circuit 505 used in the moving picture coding apparatus shown in FIG. 23, and reproduces the moving picture signal in the block. After that, it is supplied to the block update circuit 565. The motion compensation prediction circuit 566 generates a motion compensation prediction value using the reproduced image signal of the reference frame stored in the reference frame memory 567 based on the motion parameter supplied from the block decoding circuit 563.

FIG. 30 shows the configuration of the block update circuit 565. In the block update circuit 565, the reproduced image signal of the block from the reproduction circuit 564 and the motion compensation predicted image signal from the motion compensation prediction circuit 566 are switched by the selector 569 according to the block update information, and the output thereof is used as the reproduced image signal 568. Output.

Here, the motion parameter detection and motion compensation may be performed not for each block but for global motion parameter detection and global motion compensation for the entire screen. In addition, the block size for motion compensation and the block size for encoding may be different.

As described above, in this embodiment, when the conditional pixel replenishment is performed, the reference frame memory 567 is set to the reproducing circuit 5.
Since it is rewritten by the reproduced image signal obtained from 64, unlike the conventional method of rewriting by the reproduced image signal finally obtained, the reference image signal stored in the reference frame memory 567 is also stored in the still frame. If the coding distortion is large, conditional pixel supplementation is performed. Therefore, when a still scene with a small amount of motion continues, there are advantages that the image quality in the still area is improved and the error resilience is improved when an error occurs in the still area.

(Embodiment 12) FIG. 31 is a block diagram showing another embodiment of the moving picture decoding apparatus according to the eighth invention. In the decoding circuit 572, the block decoding circuit 573
After separating the code string (bit stream) multiplexed in, the side information including the motion parameter and the DCT coefficient information of the signal in the block are decoded and output.

The reproduction circuits 574a and 574b have different characteristics from the DCT coefficient information of the signal in the block which is the coding information supplied from the block decoding circuit 2610.
The decoded signal is reproduced for each one of the components (for example, the low frequency component and the high frequency component, the average value of the pixels in the block and the edge component), and is supplied to the block update circuit 575. The first component is motion-compensated and predicted in the motion compensation prediction circuit 576a, and the second component is motion-compensated and predicted in the motion compensation prediction circuit 576b based on the motion parameter supplied from the block decoding circuit 573. The first component is stored in the reference frame memory 577a and the second component is stored in the reference frame memory 577b.

FIG. 32 shows the configuration of the block update circuit 576. In the block update circuit 576, according to the block update information, the reproduced image signal of the first component and the motion-compensated predicted image signal of the first component are switched by the selector 581a, and the output thereof is supplied to the post-processing circuit 582a.

On the other hand, according to the block update information, the reproduced image signal of the second component and the motion-compensated predicted image signal of the second component are switched by the selector 581b, and the output thereof is supplied to the post-processing circuit 582b. Post-processing circuits 582a and 582b
Then, for example, for low frequency components and block average values, LP
F (low-pass filter) processing is performed, and high-frequency components and edge components are subjected to edge enhancement filter processing to perform post-processing adapted to the properties of each component. Component synthesis circuit 583
Generates a reproduced image signal by combining the components output from the post-processing circuits 582a and 582b, and outputs the reproduced image signal.

Here, the motion parameter detection and motion compensation may be carried out not for each block but for global motion parameter detection and global motion compensation for the entire screen. In addition, the block size for motion compensation and the block size for encoding may be different.

In this embodiment, the block decoding circuit 5
In 73, the encoded information is divided into two components,
It may be divided into two or more.

(Embodiment 13) FIG. 33 is a block diagram showing the structure of a motion parameter detecting device in a moving picture coding device according to the ninth invention. The input moving image signal 601 of the current frame is stored in the current frame memory 602, divided into blocks composed of a plurality of pixels, and supplied to the error evaluation value calculation circuit 603. On the other hand, the motion compensation prediction circuit 604 generates a motion compensation prediction value of the block according to the value of the motion parameter supplied from the motion parameter generation circuit 607 from the image signal of the reference frame stored in the reference frame memory 605, and the error It is supplied to the evaluation value calculation circuit 603. Error evaluation value calculation circuit 604
Calculates an error evaluation value between the input moving image signal 601 of the current frame of the block and the motion compensation prediction signal, and the comparison circuit 6
Supply to 09. As the error evaluation value, the sum of squares of the error signal or the sum of absolute values is used.

The comparison circuit 608 is the error evaluation value calculation circuit 6
The error evaluation value from 03 is compared with an external threshold value, and a signal indicating whether the error evaluation value exceeds the threshold value is supplied to the counter 610. The counter 610 counts the number of blocks whose error evaluation value exceeds a threshold value, and supplies the count value to the motion parameter determination circuit 611. The motion parameter determination circuit 611 outputs the motion parameter 612 that minimizes the number of blocks whose error evaluation value exceeds the threshold value. After completion of this series of motion parameter detection, the input moving image signal 603 of the current frame is supplied to the reference frame memory 605 and becomes the reference image signal of the next frame.

In this embodiment, as shown in FIG. 34, the following steps are repeated a predetermined number of times to detect a plurality of movements within a frame.

[Step 1] First, an inter-frame difference signal between the input moving image signal of the current frame and the reference image signal of the reference frame is obtained, and a block in which the error evaluation value of the inter-frame difference signal is smaller than the threshold value is shown. The block with motion parameter 0 shown in FIG.

[Step 2] Next, by using the block (white block) whose error evaluation value is larger than the threshold value in FIG. 34A, the configuration shown in FIG.
The motion parameter 1 is detected as shown in FIG.

[Step 3] Next, using the block (white block) whose error evaluation value exceeds the threshold value in FIG. 34B, the motion parameter 2 shown in FIG. To detect.

The above steps 1 to 3 are repeated a predetermined number of times or until the error evaluation values of all blocks fall below the threshold value.

As described above, according to this embodiment, step 3
After that, by using the block whose error evaluation value exceeds the threshold value in steps 1 and 2 before that, the motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value is detected. Even when a plurality of global motion parameters are included in the image, when extracting the plurality of motion parameters from the screen, the influence of the region having other motion parameters when extracting the parameters of each region is removed, and Appropriate motion parameters can be extracted.

(Embodiment 14) FIG. 35 is a block diagram showing an embodiment of a moving picture coding apparatus according to the tenth invention. The input moving image signal 701 is stored in the frame memory 702 as an image signal of the current frame. In the motion vector detection circuit 703, the image signal of the current frame stored in the frame memory 702 and the reference frame memory 7
A global motion vector representing the motion of the entire screen is detected with the image signal of the reference frame stored in 02. The motion vector may be obtained for each small area. Further, the motion vector detection circuit 703 determines for each small area (for example, 16 × 16 pixels) whether or not to perform motion compensation on the small area, and information indicating that fact (MC / noM
C) is output.

The block mode selection circuit 705 calculates an error signal between the input moving image signal supplied from the frame memory 702 and the motion-compensated prediction image supplied from the reference frame memory 702, and sets the error evaluation value and the threshold value. Are compared to determine whether or not conditional pixel replenishment is performed for that block (for example, 8 × 8 pixels), and information (CR / notCR) indicating that is output. The block for performing conditional pixel replenishment is a conversion circuit 706 as will be described later.
After being encoded in 1, the reproduction circuit 707 restores the reproduced image signal, and the block of the motion compensation prediction image signal in the reference frame memory 704 is updated with the reproduced image signal.

The coding / multiplexing circuit 708 multiplexes the coding information and side information of the block to form a code string (bit stream) and supplies it to the output buffer 709. The output buffer 709 outputs the bit stream to the transmission path / accumulation medium at a predetermined speed, and also supplies information indicating the buffer amount (retention amount) to the rate control circuit 710. The rate control circuit 710 includes an output buffer 709.
In order to prevent overflow or underflow, the threshold value supplied to the block mode selection circuit 705 and the quantization parameter supplied to the conversion circuit 706 are controlled.

FIG. 36 is a block diagram showing the structure of a moving picture decoding apparatus corresponding to the moving picture coding apparatus shown in FIG. Code string 721 input from transmission line / storage medium
The (bit stream) is stored in the input buffer 722 and then supplied to the separation / decoding circuit 723. Separation
The decoding circuit 723 separates the coded information and the side information of the block from the bitstream, and then decodes and outputs them. The reproduced image signal of the block (CR block) restored by the reproduction circuit 724 and subjected to the conditional pixel replenishment is supplied to the block update circuit 725 together with the motion-compensated predicted image signal created in the reference frame memory 726. The block update circuit 725 reproduces the decoded image signal 727 from the reproduced image signal and the motion-compensated predicted image signal and outputs the decoded image signal 727, and updates the reference frame memory 726 with the reproduced image signal of the block.

FIG. 37 (a) is a block diagram showing the configuration of the conversion circuit 706 in FIG. 35, and FIG. 37 (b) is a block diagram showing the configuration of the reproduction circuit 707 in FIG. 35 and the reproduction circuit 724 in FIG. Is.

In the conversion circuit 706 shown in FIG. 37 (a),
First, the input moving image signal of each block divided into blocks (for example, 8 × 8 pixels) in the block mode selection circuit 705 of FIG. 35 is divided into subblocks (for example, 4 × 4 pixels) in the subblock division circuit 731. The average value calculation circuit 732 calculates the average value of the pixels of each sub-block, and the orthogonal transformation circuit 733 calculates the two-dimensional orthogonal transformation (for example, D
CT) is given. In the average value quantizer 734, the orthogonal transform coefficient supplied from the orthogonal transform circuit 1530 is quantized according to the quantization parameter set by the rate control circuit 710, and then output to the encoding / multiplexing circuit 708 in FIG. .

On the other hand, the edge detection circuit 735 calculates the edge gain for each edge pattern by performing the sum-of-products operation with the edge pattern stored in the edge pattern memory 736 for each sub-block, and calculates the edge gain The pattern with the largest absolute value is selected as the edge of the sub-block. Here, the edge pattern memory 736 stores, for example, four different edge patterns p1 to p4 shown in FIG. 38, and the sum of each element is 0.
I am trying to be. The edges detected by the edge detection circuit 735 are quantized by an edge gain quantizer 737 composed of a quantizer with a dead zone, and then, as shown in FIG.
5 to the encoding / multiplexing circuit 708. here,
By providing a dead zone in the edge gain quantizer 737, visually insignificant small amplitude edges are not encoded. The gate circuit 738 includes an edge gain quantizer 73.
The quantized edge gain from 7 is output as an edge pattern index whose non-zero value is 0 (in the example of FIG. 38, information indicating which of p1 to p4 is 2 bits).

Next, the reproducing circuit 707 shown in FIG. 37 (b).
In (724), first, the average value inverse quantizer 741 and the inverse orthogonal transform circuit 742 reproduce the average value of each sub-block. Further, the edge gain dequantizer 743 dequantizes the edge gain, and in the subblock in which the dequantized edge gain is not 0, the edge gain dequantizer is added to the edge pattern indicated by the edge pattern index. The edge signal from 743 is multiplied by the multiplier 745 to reproduce the edge signal. However, the example of FIG. 38 does not require multiplication. Then, the inverse orthogonal transform circuit 7
After adding the average value of each sub-block from 42 and the edge signal from the multiplier 745 with the adder 746, the limiter 7
The dynamic range is limited to a range of 0 to 255 by 47 and is output.

As described above, in the present embodiment, the edge in each block of the input moving image signal is detected by the template matching in the moving image coding apparatus, and the index of the edge pattern that maximizes the absolute value of the edge gain of the edge. And the edge gain and the average value of the pixel values in the block of the input moving image signal are output as encoded information, the encoded information is input in the moving image decoding apparatus, and the edge corresponding to the index is input. By multiplying each element of the pattern by the edge gain and adding the average value to generate a reproduced image signal, a reproduced image of good quality without edge blurring can be obtained.

(Embodiment 15) An embodiment according to the eleventh invention will be described with reference to FIG. FIG. 39 (a) shows ITU-T.
H. 26 is a diagram showing a syntax for explaining an image data multiplexing structure in a frame layer in FIG.
FIG. 39B is a diagram showing the syntax for explaining the image data multiplexing structure in this embodiment.

The frame layer is composed of a frame start code PSC which is composed of a unique word and can recover the synchronization of decoding lost due to an error, and picture header information such as a frame number TR or type information PTYPE. It Here, since the frame number TR and the type information PTYPE are fixed-length coded, they are unlikely to be affected by the out-of-sync of decoding due to an error. In other words, it is not affected in the case of bit inversion. Therefore, in this embodiment, as shown in FIG.
Since the motion compensation information MCI is arranged in the picture header as shown in 9 (b), the motion compensated prediction image signal can be created without being affected by the loss of synchronization due to an error.

(Embodiment 16) An embodiment according to the twelfth invention will be described with reference to FIGS. 39 and 40. In the present embodiment, as shown in FIG. 40, after the motion compensation is performed on the input moving image signal, the changed region (the right eye portion in the example of the diagram), that is, the region where the motion compensation prediction does not hit is framed. It is rewritten by inner encoding (conditional pixel replenishment). Here, the motion-compensated prediction image is created using the motion-compensation information MCI arranged in the picture header as shown in FIG. 39B, and the conditional pixel supplementation is performed in the GOB layer (this is the GOB layer in H.261). Is not limited to)
Executed on the data in.

Therefore, if an error occurs in the GOB layer,
Since pixels are replenished with random playback images, G
When an error is detected in the OB layer, there is a very high probability that a normal reproduced image cannot be obtained until the next sync code is detected, so pixel replenishment is not performed. The error detection referred to here is not limited to the one based on the error detection code, but may be the one to detect a syntax error or a semantic error.

As described above, in this embodiment, the motion compensation prediction error signal is not coded, but the area where motion compensation prediction does not hit is rewritten by intra-frame coding, so that the influence of the image quality deterioration due to the error is affected. There is an advantage that it is difficult to propagate.

[0190]

As described above, according to the present invention, it is possible to provide a moving picture coding / decoding apparatus having high error resistance. More specifically, according to the present invention, the following effects can be achieved.

(1) According to the first invention, a plurality of screens including a screen for intra-coding and a screen for inter-coding are selectively selected for blocks in which the difference between two adjacent screens is equal to or more than a threshold value. By making the setting possible, even when the error rate of the transmission line / storage medium is higher than in the conventional case, the influence of the error can be suppressed as small as possible, and the influence of the error can be refreshed efficiently.

(2) According to the second aspect of the invention, the encoded data obtained by compression-encoding the moving image signal is variable-length encoded and transmitted / transmitted.
In the case of accumulation, a checksum is added to a variable length coded code string for transmission / accumulation, and an error in the code string is detected using the checksum in the moving picture decoding device, thereby transmitting / accumulating the transmission path / accumulation. The accuracy of error detection on the medium can be improved. Furthermore, by not outputting the decoding result in which an error has been detected, that is, by not displaying it on the screen, it is possible to suppress image quality deterioration due to the error.

(3) According to the third aspect of the invention, the input moving image signal is divided into a plurality of subbands, and only the subband of the lowest frequency is subjected to periodic refreshing by intra coding by several blocks, and the other subbands. With respect to, by performing inter-encoding using a prediction signal obtained by performing leak prediction, it is possible to suppress an increase in code amount due to periodic refresh.

(4) According to the fourth aspect of the invention, the input moving image signal is divided into a plurality of blocks made up of a plurality of pixels, and encoding can be selectively switched between intra encoding and inter encoding for each block. With the circuit provided, each of the blocks of the input moving image signal continuously counts the number of screens that are not intra-coded, and for blocks with a larger count value, for example, the period in which they are not refreshed by intra-coding is long. By forcibly refreshing in order from the block, it is possible to quickly recover the image quality deterioration in the still area.

(5) According to the fifth aspect of the invention, a synchronous code in which the structure of a code string output from the moving picture coding apparatus is added to each frame, and a plurality of rectangular blocks each of which is composed of a plurality of pixels in the frame. For a block group including a plurality of block lines when divided by, the syntax has a sync code added to the block group excluding the block group at the top of the screen, and the sync code added to the block group at the top of the screen By omitting, if it has the same synchronization recovery capability as the conventional code string syntax, the total code amount of the synchronization code,
That is, the redundant code amount for synchronization recovery can be further reduced, and when the syntax is the same as the conventional code string syntax, the redundant code amount for synchronization recovery is further improved in synchronization recovery capability.

(6) According to the sixth aspect of the present invention, when the conditional pixel replenishment is performed, the reference frame memory is rewritten with the reproduced image signal (local decoded signal) to rewrite with the input moving image signal. Unlike the method, when the coding distortion of the reference image signal accumulated in the reference frame memory is large even in the still region, conditional pixel replenishment is performed if the output buffer has a margin, so there is less still motion. When the scenes are continuous, the image quality in the still region can be improved, and the error resilience when an error occurs in the still region can be improved.

Further, a value obtained by adding a non-negative value to the error evaluation value indicating the error of each block of the reproduced image signal with respect to the input moving image signal is used as a threshold value, and the error of each block of the reference image signal with respect to the input moving image signal is set. When the error evaluation value indicating is larger than the threshold value, the contents of the block of the reference image signal stored in the reference frame memory are rewritten and the block is rewritten to perform conditional pixel replenishment. By outputting the rewriting information indicating that the conditional pixel replenishment may not be performed when the conditional pixel replenishment is better, or conversely, the conditional pixel replenishment may be performed when the conditional pixel replenishment is better not performed. Therefore, appropriate conditional pixel replenishment can always be performed.

(7) According to the seventh invention, by combining the motion compensation prediction with the conditional pixel replenishment method according to the sixth invention, the same effect as the sixth invention can be obtained, and the motion compensation prediction is performed. Since it is only necessary to perform conditional pixel replenishment for blocks that could not be predicted by, the coding efficiency is improved especially in a scene including motion, as compared with the conventional conditional pixel supplement method.

(8) According to the eighth invention, in the moving picture decoding apparatus, when the conditional pixel replenishment is performed, the reference frame memory is rewritten with the reproduced picture signal obtained by the local decoding. Unlike the conventional method of rewriting with the finally obtained reproduced image signal, if the coding distortion of the reference image signal accumulated in the reference frame memo is large even in the still region, conditional pixel supplementation is performed. . Therefore, when a still scene with a small amount of motion continues, the image quality in the still area is improved, and the error resilience when an error occurs in the still area is improved.

(9) According to the ninth invention, the image signal of the current screen is divided into a plurality of blocks composed of a plurality of pixels, and the error evaluation value with respect to the reference image is calculated for each block.
And a motion parameter that minimizes the number of blocks whose error evaluation value exceeds the threshold value using at least the block whose error evaluation value exceeds the threshold value.
And a block in which the error evaluation value exceeds the threshold value by the third step and subsequent steps according to the request, the number of blocks whose error evaluation value exceeds the threshold value is the minimum. By detecting a motion parameter that is defined as follows, even when multiple global motion parameters are included in the screen, when multiple motion parameters are extracted from the screen, other motion parameters used when extracting the parameters of each region are detected. It is possible to remove the influence of the region having the and extract appropriate motion parameters.

(10) According to the tenth aspect of the invention, an edge in each block of an input moving image signal is detected by template matching in a moving image coding apparatus, and an edge having the maximum absolute value of the edge gain of the edge is detected. The index of the pattern, the edge gain, and the average value of the pixel values in the block of the input moving image signal are encoded and output as encoded information, and the encoded information is input to the moving image decoding device, and the index is input to the index. By multiplying each element of the corresponding edge pattern by the edge gain and adding the average value to generate a reproduced image signal, a reproduced image of good quality without edge blurring can be obtained.

(11) According to the eleventh invention, the motion parameter information is fixed-length coded and arranged in the picture header, so that the influence of an error on the transmission line / storage medium can be reduced.

(12) According to the twelfth aspect of the invention, the image quality deterioration due to an error is caused by rewriting an area where motion compensation prediction does not hit by intra-frame coding instead of coding the motion compensation prediction error signal. It is possible to make it difficult to propagate the influence of.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a moving picture coding apparatus according to a first embodiment of the first invention.

FIG. 2 is a block diagram showing a configuration of a moving picture decoding apparatus according to the first embodiment.

FIG. 3 is an explanatory diagram of mode information according to the first embodiment.

FIG. 4 is a flowchart for explaining the operation of the mode selection circuit in the first embodiment.

FIG. 5 is a diagram showing an example of configurations of a CR picture and an MC picture and combinations thereof in the first embodiment.

FIG. 6 is a diagram showing the syntax of a code string in the first embodiment.

FIG. 7 is a diagram for explaining a conventional periodic refresh method.

FIG. 8 is a diagram for explaining a problem of the conventional periodic refresh.

FIG. 9 is a block diagram showing a configuration of a moving picture coding / decoding device according to a second embodiment of the second invention.

FIG. 10 is a diagram showing an example of a variable length code in the second embodiment.

FIG. 11 is a diagram showing an example of checksum calculation and inspection according to the second embodiment.

FIG. 12 is a video encoding / embodiment example 3 of the second invention.
Block diagram showing the configuration of the decoding device

FIG. 13 is a block diagram showing the configuration of a moving picture coding apparatus according to embodiment 4 of the third invention.

FIG. 14 is a block diagram showing a configuration of a moving picture decoding apparatus according to a fourth embodiment.

FIG. 15 is a diagram showing measurement results of S / N of a decoded image with respect to an original image by the moving picture coding / decoding device according to the fourth embodiment.

FIG. 16 is a block diagram showing the configuration of a moving picture coding apparatus according to embodiment 5 of the fourth invention.

17 is a block diagram showing the configuration of a mode determination circuit in FIG.

FIG. 18 is a diagram for explaining the operation of the fifth embodiment.

FIG. 19 is a block diagram showing the configuration of a moving picture coding apparatus according to a sixth embodiment of the present invention.

FIG. 20 is a diagram showing the syntax of a code string output from the moving picture coding apparatus for explaining the seventh embodiment according to the fifth invention.

FIG. 21 is a diagram showing the syntax of a code string output from a conventional moving image encoding device.

FIG. 22 is a diagram showing another syntax of a code string output from the conventional moving image coding apparatus.

FIG. 23 is a block diagram showing the configuration of a moving picture coding apparatus according to embodiment 8 of the sixth invention.

FIG. 24 is a block diagram showing a configuration of a block selection circuit in FIG.

FIG. 25 is a block diagram showing a configuration example of a conversion circuit and a reproduction circuit in FIG.

FIG. 26 is a block diagram showing another configuration example of the conversion circuit and the reproduction circuit in FIG.

FIG. 27 is a block diagram showing the configuration of a moving picture coding apparatus according to embodiment 9 of the sixth invention.

FIG. 28 is a block diagram showing the configuration of a motion compensation loop in a moving picture encoding system according to a tenth embodiment of the invention.

FIG. 29 is a block diagram showing the configuration of a moving picture decoding apparatus according to an eleventh embodiment of the present invention.

30 is a block diagram showing the configuration of the block update circuit in FIG.

FIG. 31 is a block diagram showing the configuration of a moving image decoding apparatus according to embodiment 12 of the eighth invention.

32 is a block diagram showing the configuration of the block update circuit in FIG. 31.

FIG. 33 is a block diagram showing the configuration of a motion parameter detecting device in a moving image encoding device according to a thirteenth embodiment of the invention.

FIG. 34 is a view for explaining a motion parameter detection procedure in the embodiment.

FIG. 35 is a block diagram showing the arrangement of a moving picture coding apparatus according to the fourteenth embodiment of the present invention.

FIG. 36 is a block diagram showing the configuration of the moving picture decoding apparatus according to the fourteenth embodiment.

37 is a conversion circuit in FIG. 35 and FIGS. 35 and 3;
6 is a block diagram showing the configuration of a reproducing circuit in FIG.

FIG. 38 is a diagram showing an example of an edge pattern used in Example 14;

FIG. 39 is a fifteenth embodiment according to the eleventh and twelfth inventions.
Of the syntax of the code string output from the moving picture coding device

FIG. 40 is a view for explaining the sixteenth embodiment according to the twelfth invention.

[Explanation of symbols]

101 ... Input moving image signal 102 ... Blocking circuit 103 ... Motion compensation prediction circuit 104 ... Frame memory 105 ... Mode selection circuit 106 ... Encoding control circuit 107 ... Switch 108 ... DCT circuit 109 ... Quantization circuit 110 ... Inverse quantization circuit 111 ... Inverse DCT circuit 112, 113 ...
Variable length coding circuit 114 ... Multiplexer 115 ... Output buffer 116 ... Code string 121 ... Code string 122 ... Input buffer 123 ... Demultiplexer 124, 125 ... Variable length decoding circuit 126 ... Mode decision circuit 127 ... Switch 128 ... Inverse quantization Circuit 129 ... Inverse DCT circuit 130 ... Switch 131 ... Frame memory 132 ... Reproduced image signal

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yoshihiro Kikuchi 1 Komukai Toshiba-cho, Sachi-ku, Kawasaki-shi, Kanagawa Inside the Corporate Research and Development Center, Toshiba Corporation (72) Takashi Ida Komukai-Toshiba, Kawasaki-shi, Kanagawa Town No. 1 Inside the Toshiba Research and Development Center, a stock company (72) Inventor Noboru Yamaguchi No. 1 Komukai Toshiba Town, Kouki-ku, Kawasaki-shi, Kanagawa Inside the Toshiba Research and Development Center, a stock company

Claims (15)

[Claims] 1. A block dividing means for dividing an input moving image signal into a plurality of blocks composed of a plurality of pixels, a mode selecting means for selecting an encoding mode for each of the blocks divided by the block dividing means, and this mode. First encoding means for encoding each block of the input moving image signal in the encoding mode selected by the selecting means, and encoding mode information indicating the encoding mode selected by the mode selecting means 2 encoding means, and the mode selecting means comprises a first screen for performing intra coding and a second screen for performing inter prediction coding for blocks in which the difference between two adjacent screens is greater than or equal to a predetermined threshold value. A moving picture coding apparatus, wherein a plurality of types of screens including a screen are selectively set. 2. The mode selecting means performs inter-prediction coding on a block in which the difference between the two adjacent screens is equal to or more than a predetermined threshold value, and the intra code is applied to at least a part of the block whose difference is less than the threshold value. Third
2. The moving picture coding apparatus according to claim 1, further comprising the screen of FIG. 1 as one of the plurality of types of screens. 3. A first decoding means for decoding a code string obtained by coding a moving picture signal in a coding mode selected for each block composed of a plurality of pixels to reproduce the original moving picture signal, Second decoding means for decoding a code string obtained by encoding mode information indicating an encoding mode, and the second decoding means based on the mode information decoded by the second decoding means. And a mode selecting means for selecting a decoding mode of the first screen, the mode selecting means performing intra coding on a block in which a difference between two adjacent screens is equal to or more than a predetermined threshold, and inter prediction coding. A moving picture decoding apparatus characterized in that the decoding mode is selectively set in correspondence with a plurality of types of screens including a second screen for performing. 4. A variable-length coding means for variable-length-coding coded data obtained by compression-coding an input moving image signal, and a code string output from this coding means is composed of a plurality of variable-length codes. Checksum generating means for generating a checksum with a value of the lower predetermined number of bits of the addition value obtained by adding the code of each division by dividing each unit into a predetermined number of bits, and the check generated by the checksum generating means. And a checksum adding means for adding and outputting a sum to each of the coding units. 5. Decoding means for decoding a code string obtained by performing variable length coding of coded data obtained by compressing and coding a moving picture signal to reproduce an original moving picture signal, and a plurality of variable length codes for the code string. The code of each division is added by dividing into a predetermined number of bits for each encoding unit consisting of
A moving image having an error detecting means for detecting an error in the code string by comparing a value of a predetermined number of lower bits of the added value with a checksum added to each of the coding units. Image decoding device. 6. A first and second sub-band dividing means for dividing an input moving image signal and a motion-compensated prediction signal obtained by performing motion compensation on the input image signal into a plurality of sub-bands having different frequency bands. And, regarding the lowest frequency sub-band of the plurality of sub-bands, the lowest-frequency sub-band from the first sub-band dividing means is at least once within a predetermined period for each area obtained by dividing the screen into a plurality of areas. First coding means for intra-coding the input moving image signal of the sub-band, and sub-bands other than the lowest frequency among the plurality of sub-bands other than the lowest frequency from the second sub-band dividing means. Leakage prediction signal obtained by multiplying the motion-compensated prediction signal of the sub-band by a coefficient smaller than 1 and the motion of the sub-band other than the lowest frequency from the first sub-band dividing means. Video encoding apparatus characterized by a second coding means for coding a prediction error signal between the image signal. 7. A subband dividing means for dividing a motion-compensated prediction signal obtained by performing motion compensation on a reproduced moving image signal into a plurality of subbands having different frequency bands, and a subband dividing means of the lowest frequency of the plurality of subbands. For the sub-band, the motion compensation prediction signal of the lowest frequency and the prediction error signal from the sub-band dividing means are added at least once within a predetermined period for each area obtained by dividing the screen into a plurality of areas, and First decoding means for decoding a reproduced moving image signal, and motion compensation prediction of subbands other than the lowest frequency from the subband dividing means for subbands other than the lowest frequency among the plurality of subbands. It has a second decoding means for generating the reproduced moving image signal by adding the leak prediction signal obtained by multiplying the signal by a coefficient smaller than 1 and the reproduced moving image signal. Moving picture decoding apparatus according to. 8. A block dividing means for dividing an input moving image signal into a plurality of blocks made up of a plurality of pixels, and a code for selectively switching the input moving image signal between intra coding and inter coding for each block. Encoding means for encoding, a counting means for counting the number of screens in which each block of the input moving image signal is not continuously intra-encoded, and the encoding for a block having a larger count value of the counting means. And a control means for forcibly performing intra-coding by the means. 9. A block dividing means for dividing an input moving image signal into a plurality of blocks composed of a plurality of pixels, and an encoding means for encoding the input moving image signal for each block divided by the block dividing means. A local decoding means for decoding the coded information obtained by the coding means to generate a reproduced image signal; a storage means for storing a reference image signal for at least one screen; First calculating means for calculating a first error evaluation value indicating an error of each block of the reproduced image signal, and a second error evaluation value indicating an error of each block of the reference image signal with respect to the input moving image signal, Second calculation means for calculating, and using the value obtained by adding a non-negative value to the first error evaluation value as a threshold value, it is determined whether or not the second error evaluation value is larger than the threshold value. You And a reference image signal accumulated in the accumulating unit for the block having the second error evaluation value larger than the threshold value, based on the determination result of the determining unit, and the reproduction image signal. A moving image coding apparatus, comprising: the coding information of the block and a unit that outputs rewriting information indicating that the block has been rewritten. 10. A block dividing means for dividing an input moving image signal into a plurality of blocks composed of a plurality of pixels, and an encoding means for encoding the input moving image signal for each block divided by the block dividing means. A local decoding means for decoding the coded information obtained by the coding means to generate a reproduced image signal; a storage means for storing a reference image signal for at least one screen; the input moving image signal and the Motion parameter detecting means for detecting and outputting a motion parameter between the reference image signal and the reference image signal, and motion compensating the reference image signal accumulated in the accumulating means on the basis of the motion parameter to generate a motion compensation predicted image signal. Motion compensation prediction means, means for rewriting the reference image signal accumulated in the accumulation means by the motion compensation predicted image signal, the input moving image A first calculating means for calculating a first error evaluation value indicating an error of each block of the reproduced image signal with respect to a signal; and a second calculating means showing an error of each block of the motion compensation prediction image signal for the input moving image signal. The second error evaluation value is larger than the threshold value, the second error evaluation value being greater than the threshold value with a value obtained by adding a non-negative value to the first error evaluation value as a threshold value. Determining means for determining whether or not the reference image signal accumulated in the accumulating means for the block having the second error evaluation value larger than the threshold value is reproduced based on the determination result of the determining means. A moving picture coding apparatus, comprising: rewriting with an image signal, and means for outputting the coding information of the block and rewriting information indicating that the rewriting has been performed for the block. 11. Decoding means for decoding a coded information obtained by coding a moving image signal for each of a plurality of blocks made up of a plurality of pixels to generate a reproduced image signal, and a reference image for at least one screen. Accumulating means for accumulating a signal, and motion compensation for compensating the reference image signal accumulated in the accumulating means on the basis of the motion parameter between the moving image signal and the reference image signal to create a motion compensated prediction image signal For the prediction unit and the block to which the rewriting information is sent among the plurality of blocks, the reproduced image signal of the block is output, and the reference image signal stored in the storage unit is rewritten with the reproduced image signal. A moving image decoding apparatus, comprising: a unit that outputs the motion-compensated prediction image signal for a block to which rewriting information is not sent. 12. The decoding means divides the encoded information into a plurality of different components, performs different post-processing on each of these components, synthesizes the components, and outputs the synthesized image signal. The moving picture decoding apparatus according to claim 11. 13. A block dividing means for dividing an input moving image signal into a plurality of blocks composed of a plurality of pixels, and an encoding means for encoding the input moving image signal for each block divided by the block dividing means. A local decoding means for decoding the coded information obtained by the coding means to generate a reproduced image signal; a storage means for storing a reference image signal for at least one screen; the input moving image signal and the Motion parameter detecting means for detecting and outputting a motion parameter between the reference image signal and the reference image signal, and motion compensating the reference image signal accumulated in the accumulating means based on the motion parameter to generate a motion-compensated predicted image signal. Motion compensation prediction means, means for rewriting the reference image signal stored in the storage means by the motion compensation prediction image signal, the input moving image Determination means for determining whether or not an error evaluation value indicating an error for each block of the motion compensation prediction image signal for a signal is larger than a predetermined threshold value, and the error evaluation value based on the determination result of the determination means. For a block whose value is larger than the threshold value, the reference image signal stored in the storage means is rewritten with the reproduction image signal, and the coding information of the block and the rewriting indicating that the block has been rewritten. And a means for outputting information, wherein the encoding means detects an edge in each block of the input moving image signal by template matching, and an edge pattern index in which an absolute value of an edge gain of the edge is maximum. And the edge gain and the average value of the pixel values in the block of the input moving image signal are encoded to obtain the encoded information. Moving image encoding apparatus and outputs a. 14. Decoding means for decoding a coded information obtained by coding a moving image signal for each of a plurality of blocks each having a plurality of pixels to generate a reproduced image signal, and a reference image for at least one screen. Accumulating means for accumulating a signal, and motion compensation for compensating the reference image signal accumulated in the accumulating means on the basis of the motion parameter between the moving image signal and the reference image signal to create a motion compensated prediction image signal For the prediction unit and the block to which the rewriting information is sent among the plurality of blocks, the reproduced image signal of the block is output, and the reference image signal stored in the storage unit is rewritten with the reproduced image signal. For a block to which rewriting information is not sent, there is provided means for outputting the motion-compensated predicted image signal, and the decoding means is provided as the encoding information. The input is the information of the edge pattern index that maximizes the absolute value of the edge gain of each block in the moving image signal, the edge gain, and the average value of the pixel values in the block of the moving image signal. A moving image decoding apparatus, wherein the reproduced image signal is generated by multiplying each element of the edge pattern corresponding to the index by the edge gain and then adding the average value. 15. Determining means for determining whether or not the encoded information is normally decoded by the decoding means,
The means for prohibiting the rewriting regardless of the presence or absence of the rewriting information with respect to a block in which the coded information is not normally decoded by the judging means, further comprising: Video decoding device.