JPH10327412A - Image decoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image decoder that is capable of high speed processing and has a sophisticated function where a futile processing in a conventional image decoder is eliminated. SOLUTION: A bit stream decomposing section 111 extracts coded and quantized DCT(discrete cosine transform) coefficients for each block and a decoding section 112 decodes the coded and quantized DCT coefficients into run length and valid coefficients. An inverse quantization section 115 generates orthogonal transform coefficients from the run length and the valid coefficients, and an inverse discrete cosine transform section 116 generates a difference image from the orthogonal transform coefficients. In the case of a skipped block, a 1st constant generating section 117 outputs a 'constant 0' and an image decoding section 119 adds a reference unit image and a difference image to generate a new unit image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

TECHNICAL FIELD The present invention relates to a moving picture E
The present invention relates to an image decoding device that decodes a moving image encoded with high efficiency based on an xperts group (MPEG) or the like.

[0002]

2. Description of the Related Art In recent years, techniques for highly efficient encoding and compression of moving pictures such as MPEG have been actively studied and used in fields such as computers, communications, and broadcasting. According to this technique, a moving image is composed of a plurality of frame images, and each frame image is divided into a plurality of blocks. Each divided block usually has 64 pixels (8 pixels × 8 pixels).
Pixel). The moving image is predictively encoded for each block in each frame, and is compressed through processes such as discrete cosine transform (DCT), quantization, and variable-length encoding, which are a kind of orthogonal transform. Generated. The compressed code string (bit stream) undergoes inverse processing of variable length encoding (decomposition processing and decoding processing of bit stream), inverse quantization, inverse DCT, and inverse processing of predictive encoding (image restoration processing), Restored to image.

Next, the principle of predictive coding will be described with reference to "Basic knowledge of image data compression" (CQ Publishing Company, published in December 1991). The predictive coding is a method of coding a difference between a true value and a predicted value (referred to as a prediction error) in order to predict a current pixel value from a transmitted pixel value. The characteristic of the image signal that there is a strong correlation between the luminances of neighboring pixels is used.

[0004] As shown in FIG.
The original signal X that gradually gets brighter_i
1102 series "0 0 0 1 0 3 1 0 1
 1 ... "1112. Here, the pixel number
The signal i1101 is the original signal X _i Sort in the order that 1102 appears
The original signal X_i 1102 is 0-15
The pixel has 16 luminance values. This original signal X_i 11
Looking at the series 1112 of 02, the neighboring pixel values are too large
It turns out that it has not changed sharply. Therefore, this sex
With quality, the original signal X_i Receiving 1102 series 1112
The believer is a certain original signal X_i When you receive 1102,
You can roughly predict what the value of will be. There
Then, from the standpoint of the receiver, the sender sets the first original signal X
₁ Is sent to the receiver, the receiver receives the next original signal X_Two Nitsu
And the original signal X₁ Predict that the same value will come
The predicted value (in this case, Y₁ ) And true value X_Two 
And the difference D_Two = X_Two -Y₁ Will be transmitted. Receiver
Is the signal X sent earlier₁ The next sent message
No. D_Two By adding_Two Know the true value of
Can be.

[0005] The sender sets the original signal X _i-1 appearing immediately before the original signal X _i 1102 as the predicted value Y _i 1103 (Y _i =
X _i-1), calculates the difference D _i 1104 between the predicted value Y _i 1103 and the true value _{X i 1102 (D i = X} i -Y i), the difference D
_i 1104 is the transmission value T _i 1105, and the transmission value T _i 11
05 to the receiver via the transmission path 1106. The receiver receives the transmission value T _i 1105 as the reception value R _i 1107 (R _i = T _i ), adds the previous decoding value Z _i−1 to the reception value R _i 1107, and obtains the current decoding value. Generate Z _i 1108 (Z _i = Z _i-1 + R _i ). Thus, the decoded value sequence “0 0 0 1 0 3 1...” 1118 is restored.

The original signal X _i-1 that appears immediately before the current original signal X _i is called a reference pixel. This method is the simplest method in which the immediately preceding pixel is used as a prediction value.
Such predictive coding using the immediately preceding pixel is called forward predictive coding. Also, predictive coding using the immediately succeeding pixel is called backward predictive coding, and predictive coding using the immediately preceding and succeeding pixels is called bidirectional predictive coding.

Further, a case where the above-described predictive coding is applied to a plurality of pixels adjacent to each other in one frame image is called intra-frame predictive coding. The case where the above-described predictive coding is applied is called inter-frame predictive coding. In the above description of the principle, predictive coding is performed for each pixel. However, predictive coding is generally performed for each block composed of 8 pixels × 8 pixels. In this case, if the previous block to be compared with one block (referred to as a reference block) is exactly the same, instead of transmitting 64 pixel values 0, the same as the previous block is used. Is transmitted, the amount of information to be transmitted can be reduced.

FIG. 24 shows a hierarchical structure of a compressed code string generated by compressing and encoding a moving image using the moving image compression technique. Compressed code string 1 corresponding to one moving image data
211, as shown in the sequence layer 1201 of this figure, a plurality of groups of pictures (Group of pictures)
(Picture, GOP) 1212, and each GOP 1212 has a sequence header code (Sequence Header C) indicating the start of a sequence layer.
mode, SHC) is transmitted. Each GOP 1212, as shown in the GOP layer 1202,
Group start code (Gr) indicating the start code of the OP
up Start Code (GSC) and a plurality of I pictures 1221, B pictures 1222,
It consists of a P picture 1223.

I picture 1221, B picture 122
2. Each of the P pictures 1223 corresponds to one frame image. The I picture is called an intra-frame coded image, and is coded using only that frame image without taking a difference from other frame images. A P picture is referred to as an inter-frame forward prediction coded image. A frame image is predictively coded by forward prediction coding, and a pixel value of a certain pixel is changed to a corresponding pixel in a previous (past) frame image. Are represented using the difference from the pixel value. A B picture is called an inter-frame bidirectional predictive coded image, and a frame image is predictively coded by bidirectional predictive coding, and a pixel value of a certain pixel is used for both a past frame image and a future frame image. This is represented by using a difference from a pixel value of a corresponding pixel in the frame image. Therefore, the decoding of the B picture is performed after the frame images before and after the B picture are decoded. Similarly P
In order to decode a picture, it is necessary that the preceding I picture and P picture have been decoded.

As shown in a picture layer 1203, each picture has a picture start code (Picture Start Code, PSC) indicating the start of the picture layer.
1233, a picture coding type (P which indicates a picture type such as an I picture, a B picture, and a P picture).
image Coding Type (PCT) 12
32, and a plurality of slices 1231.

A slice 1231 corresponds to one horizontal row of images in the frame image. As shown in the slice layer 1204, each slice 1231 has a slice start code (Slice Start Co) indicating the start of the slice layer.
de, SSC) and a plurality of macro blocks (Macro Block, MB) 1241.

Each macro block 1241 includes, as shown in a macro block layer 1205, intra, forward prediction,
Encoded macroblock type (Macro type) indicating the type of macroblock such as backward prediction and bidirectional prediction
Block Type (MTB) 1615, Motion Horizont indicating the horizontal and vertical components of the forward motion vector of macroblock 1241 respectively.
al ForwardCode (MHF) 1252, M
otion Vertical Forward Co
de (MVF) 1253, Motion Horizontal Backwa indicating horizontal and vertical components of the backward motion vector of macroblock 1241 respectively.
rd Code (MHB) 1254, Motion V
optical Backward Code (MV
B) Coded Block Pat indicating the pattern of 6 blocks in 1255, macroblock 1241
It includes information such as turn (CBP) 1256 and a plurality of blocks (Block, B) 1251.

The plurality of blocks 1251 are usually 6
Blocks 1261, 1262, 1263, 126
4, 1265 and 1266. Block 1261,
1262, 1263, and 1264 are composed of components representing luminance, and blocks 1265 and 1266 are composed of components representing color differences. Each block 1271 is usually configured by arranging eight pixels 1272 in the vertical direction and eight pixels in the horizontal direction, and is composed of a total of 64 pixels 1272.

Coded Block Pattern
(CBP) 1256 (hereinafter, Description of the Relat
In ed Art, this is called a block pattern).
With respect to a macroblock in a picture or a P-picture, it indicates, for each pattern, whether any of the six blocks constituting the macroblock has a difference from the corresponding macroblock in the previous frame or the two preceding and succeeding frames. is there.

In the case of a macroblock of an I picture, the pixel value of each block includes six blocks constituting the macroblock. However, in the case of a macroblock of a P picture or a B picture, other pixel values depend on the block pattern. The pixel value is only the pixel value of a block that indicates that a difference has occurred with respect to the image of the corresponding macroblock of the frame. Naturally, when there is no difference between any of the P-picture and B-picture macroblocks and the corresponding macroblocks of other frames in any of the six blocks,
No pixel value is included for that macroblock.

As described above, a block that does not include a pixel value is referred to as a skipped block (a skipped block). In a conventional image decoding apparatus, a decoding unit that decodes an encoded pixel, and a constant that generates a block having a pixel value of 0 as an element when the pixel value is not included in a macroblock of a P picture or a B picture A generator is provided in parallel, and when decoding information including pixels, a code is decoded by the decoding unit to generate a quantized orthogonal transform coefficient, and information that does not include pixels in a P picture or a B picture is decoded. In this case, a block having a pixel value of 0 as an element is generated by the constant generation unit. Subsequently, the block including the generated quantized orthogonal transform coefficient and the pixel value 0 as elements is inversely quantized by an inverse quantization unit to generate an orthogonal transform coefficient, and the generated orthogonal transform coefficient is subjected to an inverse discrete cosine transform. The image is generated by the inverse discrete cosine transform by the unit.

Further, when data is lost or garbled, such as when coded data is transmitted, when such data is decoded, a variable length code that does not exist in the decoding table is obtained. Error, the value of the motion vector may be out of the predetermined range, or the value of the motion vector may point outside the reference image.

Even when such an error occurs, a block having a pixel value of 0 as an element is generated from the occurrence of the constant. Subsequently, the block having the generated pixel value of 0 as an element is inversely quantized by an inverse quantization unit, and the obtained coefficients are inverse discrete cosine transformed by an inverse discrete cosine transform unit to generate an image, and an error occurs. As a substitute for the image. FIG.
2 is a time chart showing the transition of processing over time for each block of the conventional image decoding apparatus. In the vertical direction, bit stream decomposition processing C401, decoding processing C4
02, a constant generation process C403, an inverse quantization process and an inverse discrete cosine transform process C404, and an image restoration process C405 are listed, and time is taken in the horizontal direction. A block B10 including a pixel, a block B11 to be skipped, a block B12 including a pixel, a block B13 having an error, and a block B14 including a pixel flow through the conventional image decoding apparatus in this order, and are C30 to C33, C34 to C37, and C38.
To C41, C42 to C45, and C46 to C49 indicate processes of the respective processing units of the blocks B10, B11, B12, B13, and B14, respectively.

The processing for each block includes C30, C3
1, C32, C33, C34, C35, C36, C3
7, C38, C39, C40, C41, C42, C4
3, C44, C45, C46, C47, C48, C49
Work in order. As described above, in the conventional image decoding apparatus, the case where the coded pixel is decoded, the case where the P picture,
The inverse quantization unit and the inverse discrete cosine transform unit can be used in common when a macroblock of a B picture does not include a pixel value. Also, if an error occurs,
The inverse quantization unit and the inverse discrete cosine transform unit can be used in common.

[0020]

However, in the above-described conventional image decoding apparatus, when a block being decoded is included in a P picture or a B picture and is a block to be skipped, an error occurs in the block being decoded. In this case, although the device can be shared, on the other hand, a block having a pixel value of 0 as a pixel is subjected to an inverse quantization process and an inverse discrete cosine transform process, so that unnecessary processing is performed and the processing speed is reduced. There is a problem that.

When a motion vector error occurs, after the image decoding process at the time of the error is completed, the slice including the block containing the error being decoded is skipped, and the head of the next slice is detected. Must be performed. If no error occurs, this is essentially unnecessary processing, and there is a problem that the processing speed is reduced.

In the above-mentioned prior art, a problem has been pointed out when processing is performed on a block basis. However, the same problem exists when a processing unit image is a macroblock. Here, the unit image is defined as a minimum unit image in which a code string is decomposed by a bit stream decomposition process and a minimum unit image in a case where inverse processing of predictive coding is performed by an image restoration process. Point to two. However, in order to distinguish between the two, the term “unit image” is simply referred to as the former minimum unit image, and the latter unit image is referred to as the “restored unit image”. By the way, in the above prior art,
Both the “unit image” and the “restoration unit image” are blocks.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a high-performance image decoding apparatus capable of performing high-speed processing while avoiding useless processing of a conventional image decoding apparatus. Specifically, a first object of the present invention is to speed up image decoding processing when there is a unit image to be skipped.
It is a second object of the present invention to avoid a reduction in image decoding processing speed when an error occurs in a unit image to be decoded.

Further, a third object of the present invention is to generate a preferable substitute image when an error occurs in a unit image to be decoded.

[0025]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention relates to an image decoding apparatus for decoding a compressed code sequence and generating a moving image. The encoded basic unit image is extracted, and the encoded basic unit image is the same or different information indicating whether the non-coded basic unit image is the same as the reference unit image in the already decoded frame image. When the same or different information indicates that the basic unit image and the reference unit image are different, further includes an encoded differential unit image, and the differential unit image is the basic unit image and the basic unit image. When the reference unit image of another frame image referred to when the image is predictively encoded is different, a difference value between a pixel value forming the original unit image and a pixel value forming the reference unit image is used. The unit image and the reference The rank image is composed of a decomposing unit including a predetermined number of pixels, an image decoding unit that sequentially decodes the extracted coded difference unit image to generate a difference unit image, and a unit image having a pixel value of 0 as an element. First constant generating means for generating a first constant image, and the first constant generating means generating the first constant image when the extracted difference information indicates that the basic unit image and the reference unit image are the same. Selecting the first constant image, and selecting the difference unit image decoded by the image decoding means when the extracted difference information indicates that the original unit image and the reference unit image are different. By adding a selection result by the first image selection unit and the first image selection unit and the reference unit image,
Image restoration means for generating a basic unit image.

In this case, the first constant generation means includes: a first constant generation unit for generating one pixel value 0; It may be configured to include a first constant control unit that controls so as to generate a value 0 and generates a first constant image having pixel values 0 as elements of the number of pixels included in the unit image. Here, the first constant generating means generates one pixel value 0 as a first constant image, and the first image selecting means, when the reference unit image and the basic unit image are the same, One pixel value 0 generated by the first constant generation means
Is selected by the number of pixels included in the unit image, and the image restoring unit selects the unit image selected by the first image selecting unit when the reference unit image and the original unit image are the same. May be configured to generate a basic unit image by adding the pixel values 0 corresponding to the number of pixels included in the frame image and the reference unit image in the already decoded frame image.

Here, the first constant generating means generates a plurality of pixel values 0 as a first constant image, and the first image selecting means determines that the reference unit image and the basic unit image are the same. In some cases, a plurality of first constant images composed of a plurality of pixel values 0 generated by the first constant generating means are selected, and the pixel values 0 are selected in total and the number of pixels included in the unit image is selected. And, when the reference unit image and the original unit image are the same, the image restoration unit includes a pixel value 0 for the number of pixels included in the unit image selected by the first image selection unit, The unit image may be generated by adding the reference unit image in the already decoded frame image to the reference unit image.

Here, the image decoding means sequentially decodes the extracted coded difference unit image to generate an effective coefficient value and a run length of at least one pair; A second constant generating means for generating a second constant consisting of: the at least one run length generated by the code string decoding means; selecting the second constant for the obtained run length; A second coefficient for generating a coefficient matrix by selecting at least one effective coefficient value thus obtained and combining a second constant for the selected run length and the effective coefficient.
Selecting means, inverse quantizing means for performing an inverse quantization process on the coefficient matrix generated by the second selecting means to generate an orthogonal transform coefficient, and inverse transform of orthogonal transform to the generated orthogonal transform coefficient The image decoding apparatus according to claim 1, further comprising: an inverse orthogonal transform unit that performs processing and generates the difference unit image.

Here, the inverse transform of the orthogonal transform may be configured to be an inverse discrete cosine transform. Here, the image restoring means includes: an image storage means for storing one or a plurality of reference frame images and a first frame image to be restored therefrom; and a reference unit image from the reference frame images stored in the image storage means. The read-out reference unit image is added to the read-out reference unit image and the first constant image or the difference unit image selected by the first image selecting means to generate a basic unit image, and the generated basic unit image is added to the reference unit image. A sub-image restoring unit that writes the first frame image in the image storage unit may be included.

Here, the image decoding means may further include, for each of the original unit images, when the difference unit image is generated from the encoded difference unit image, the encoded difference unit image may have a value other than a specified value. The image decoding apparatus further includes a first error detection unit that detects an error indicating that the image decoding unit detects an error when the first error detection unit detects an error. First error control means for controlling the decoding of the encoded coded difference unit image to be stopped, wherein the image restoring means further comprises, when an error is detected, replacing the coded difference unit image in which the error has occurred. Generating an alternative unit image using the reference frame image stored in the image storage unit, and writing the generated alternative unit image as a first frame image in the image storage unit It may be configured to include an error image restoring means.

Here, the image restoring means further includes a second error detecting means for detecting an error that occurs with a motion compensation process when a basic unit image is generated, and the first error controlling means. May further be configured to control the image decoding unit to stop decoding the coded difference unit image in which the error has occurred when the second error detection unit detects an error. .

Here, the image restoring means further reads the reference frame image stored in the image storage means before starting restoration of the frame image, and stores the read reference frame image in the image storage. Means for copying a frame image to be written as a first frame image, wherein the sub-image restoring means generates the basic unit generated when no error is detected by the first error detecting means and the second error detecting means. A unit image writing unit that writes an image to the image storage unit as a first frame image, wherein the image restoration unit further includes a unit that detects an error by the first error detection unit or the second error detection unit. An image writing suppressing unit that suppresses writing of the alternative unit image to the image storage unit with respect to the error image restoring unit; It may be configured to include.

Here, the error image restoring means, when an error is detected by the first error detecting means or the second error detecting means, from the image storage means, encodes the detected error. An image reading means for reading a difference unit image and a unit image at the same position in the reference frame image, an alternative for writing the read unit image as an alternative unit image, and writing the alternative unit image in the image storage means as a first frame image An image writing unit may be included.

Here, the image restoring means further comprises: a second error detecting means for detecting an error occurring along with the motion compensation processing when the basic unit image is generated; Instead of the coded difference unit image in which an error has occurred, an alternative unit image is generated using the reference frame image stored in the image storage unit, and the generated alternative unit image is stored in the image storage unit. Error image restoration means for writing as a first frame image,
The image decoding apparatus further controls the image decoding unit to stop decoding the coded difference unit image in which the error has occurred when the second error detection unit detects an error. You may comprise so that error control means may be included.

Here, the image restoring means further reads the reference frame image stored in the image storage means before starting restoration of the frame image, and stores the read reference frame image in the image storage. Means for writing frame images as first frame images, wherein the sub-image restoration means stores the generated basic unit image in the image storage means when no error is detected by the second error detection means. A unit image writing unit for writing as a first frame image, wherein the image restoring unit further transmits the substitute unit to the error image restoring unit when an error is detected by the second error detecting unit. It may be configured to include an image writing suppression unit that suppresses writing of an image to the image storage unit.

Here, the error image restoration means, when an error is detected by the second error detection means, from the image storage means, the coded difference unit image in which the error is detected and the reference frame. Image reading means for reading a unit image at the same position in an image, and alternative image writing means for writing the read unit image as an alternative unit image and writing the alternative unit image as a first frame image in the image storage means. It may be configured as follows.

Also, the present invention decodes a compressed code string,
An image decoding apparatus for generating a moving image, wherein a predetermined number of basic unit images encoded by decomposing a compressed code string are extracted, and wherein the predetermined number of basic unit images encoded is the predetermined number of basic unit images. In the basic unit image, which uncoded basic unit image is the same as the reference unit image in the already decoded frame image, and indicates which uncoded basic unit image is different from the reference unit image. Including a pattern identifier for identifying the same or different pattern, the encoded predetermined number of basic unit images,
Of the predetermined number of basic unit images, for the basic unit image indicated to be different from the reference unit image by the same or different pattern identified by the pattern identifier, further including an encoded difference unit image, When the basic unit image is different from a reference unit image of another frame image referred to when the basic unit image is predictively encoded, a pixel value forming the basic unit image and the reference unit The original unit image and the reference unit image are composed of a difference value from a pixel value constituting the image, and the original unit image and the reference unit image are sequentially decoded by a decomposing unit including a predetermined number of pixels, and the extracted encoded difference unit image is sequentially decoded. Image decoding means for generating a unit image, first constant generation means for generating a first constant image comprising a unit image having a pixel value of 0 as an element, a different pattern, and a parameter for identifying each different pattern. A pattern table that stores the pattern identifiers, and a different pattern indicated by the extracted pattern identifiers are specified from the pattern table, and for each of the predetermined number of unit images, an uncoded unit image is referred to. Determining means for determining whether or not the same as the unit image, for each of a predetermined number of the basic unit images, when it is determined that the unencoded basic unit image and the reference unit image are the same, The first constant generated by the first constant generating means;
Selecting a constant image, and selecting the difference unit image decoded by the image decoding unit when it is determined that the unencoded original unit image and the reference unit image are different. The image processing apparatus further includes an image selecting unit, and an image restoring unit that generates a basic unit image by adding a selection result by the first image selecting unit and the reference unit image.

Here, the first constant generating means further comprises:
Pixel value 0 for the number of pixels included in the predetermined number of basic unit images
And generating a third constant image including:
Using the extracted pattern identifier, for each of the predetermined number of basic unit images, all of the predetermined number of basic unit images are the same as the reference unit image, or at least one of the predetermined number of basic unit images is The selection unit determines whether the reference unit image is different from the reference unit image. If the determination unit determines that all of the predetermined number of original unit images are the same as the reference unit image, the selection unit determines the first constant. Selecting the third constant image generated by the generating means, the image decoding means;
A code string decoding unit that sequentially decodes the extracted coded difference unit images to generate an effective coefficient value and a run length of at least one pair; a second constant including a pixel value of 0; A second constant generating means for generating a fourth constant including a pixel value of 0 for the number of pixels included, and a predetermined constant when it is determined that at least one of the predetermined number of basic unit images is different from the reference unit image; For each basic unit image of a number of basic unit images, for the basic unit image in which the basic unit image and the reference unit image are the same, select the fourth constant generated by the second constant generating unit. Then, a coefficient matrix is generated, and for each basic unit image of the predetermined number of basic unit images, a basic unit image in which the basic unit image is different from the reference unit image is generated by the code string decoding unit. Said few And one acquired run length, selects the second constant corresponding to the acquired run length, selects the at least one effective coefficient value generated, and selects the second constant corresponding to the selected run length and the effective constant. Second selecting means for combining the coefficient values with the coefficient values to generate a coefficient matrix, and inverse quantizing means for performing an inverse quantization process on the coefficient matrix generated by the second selecting means to generate an orthogonal transform coefficient And an inverse orthogonal transform unit for performing an inverse transform process of orthogonal transform on the generated orthogonal transform coefficient to generate the difference unit image.

Also, the present invention is an image decoding apparatus for decoding the compressed code sequence to generate a moving image, wherein the compressed code sequence is decomposed to extract an encoded basic unit image, and The converted basic unit image includes homogeneity information indicating whether the unencoded basic unit image and the reference unit image in the already decoded frame image are the same, and the homogeneity information is the basic unit image. When indicating that the image and the reference unit image are different, the image further includes an encoded difference unit image, and the difference unit image is referred to when the original unit image and the original unit image are predictively encoded. When the reference unit image of another frame image is different from the reference unit image, the reference unit image is configured by a difference value between a pixel value forming the basic unit image and a pixel value forming the reference unit image. An image consists of a predetermined number of pixels A decoding unit for sequentially decoding the extracted coded difference unit images to generate an effective coefficient value and a run length of at least one pair, and generating a second constant including a pixel value of 0. A second constant generating means for obtaining the at least one run length generated by the code string decoding means, selecting the second constant for the obtained run length, and generating the at least one effective coefficient value A first processing unit including a second selection unit that generates a coefficient matrix by combining a second constant corresponding to the selected run length and the effective coefficient, and stores the generated coefficient matrix. A first storage unit that reads the coefficient matrix from the first storage unit, performs an inverse quantization process on the read coefficient matrix, and generates an orthogonal transform coefficient; Transform coefficients to orthogonal transform Subjected to a conversion process,
A second processing unit including an inverse orthogonal transform unit that generates a difference unit image; and a second processing unit that stores the generated difference unit image.
A storage unit; a first constant generation unit configured to generate a first constant image including a unit image having a pixel value of 0 as an element; and the extracted difference information is the same as the reference unit image and the basic unit image. In the case of indicating, the first constant image generated by the first constant generating means is selected, and the extracted difference information is different between the reference unit image and the original unit image. In the case shown, the difference unit image is read from the second storage unit, the first image selection unit that selects the read difference unit image, and one or a plurality of reference frame images and the first frame image to be restored therefrom An image storage unit for storing, a reference unit image read from a reference frame image stored in the image storage unit, and the read reference unit image selected by the first image selection unit; A third processing unit including: adding a constant image or the difference unit image to generate a basic unit image, and writing the generated basic unit image to the image storage unit as a first frame image; It is characterized by including sequential control means for controlling the first processing means, the second processing means, and the third processing means by pipeline.

Here, when the difference unit image is generated from the coded difference unit image for each of the original unit images, the first processing means may determine that the coded difference unit image is out of the specified range. First error detecting means for detecting an error indicating that a value is included, and decoding of the coded difference unit image in which the error has occurred to the code string decoding means when the first error detecting means detects an error. First error control means for controlling the image storage means to stop the processing, further comprising, when an error is detected, the image storage means instead of the coded difference unit image in which the error has occurred. An error image restoration unit that generates an alternative unit image using the reference frame image stored in the image storage unit and writes the generated alternative unit image as a first frame image in the image storage unit. Configuration may be.

Here, the third processing means further includes a second error detecting means for detecting an error occurring along with the motion compensation processing when the basic unit image is generated, and the first error control means The means may further be configured to, when the second error detecting means detects an error, control the image decoding means to stop decoding the coded difference unit image in which the error has occurred. Good.

Here, the third processing means further reads the reference frame image stored in the image storage means before starting restoration of the frame image, and stores the read reference frame image in the image. A frame image copying unit for writing as a first frame image in a storage unit, wherein the image restoration unit is configured to generate the basic unit when an error is not detected by the first error detection unit and the second error detection unit. A unit image writing unit that writes an image to the image storage unit as a first frame image, wherein the image restoration unit further includes a unit that detects an error by the first error detection unit or the second error detection unit. An image writing suppressing unit that suppresses writing of the alternative unit image to the image storage unit with respect to the error image restoring unit; Unnecessarily may be configured.

Here, the error image restoration means, when an error is detected by the first error detection means or the second error detection means, encodes the detected error from the image storage means. An image reading means for reading a difference unit image and a unit image at the same position in the reference frame image, an alternative for writing the read unit image as an alternative unit image, and writing the alternative unit image as a first frame image in the image storage means An image writing unit may be included.

[0044]

BEST MODE FOR CARRYING OUT THE INVENTION

1.1 Schematic Configuration of Image Decoding Device FIG. 1 is a block diagram illustrating a schematic configuration of an image decoding device as one embodiment of the present invention. This image decoding apparatus includes a decoding control unit 110, a bit stream decomposition unit 11
1, decoding section 112, second constant generation section 113, second selection section 114, inverse quantization section 115, inverse discrete cosine transform section 11
6, a first constant generation unit 117, a first selection unit 118, an image restoration unit 119, and an image storage unit 120.

It should be noted that the unit image in the processing of the image decoding apparatus shown in this figure is a block. Hereinafter, each component of the image decoding apparatus will be described. (1) Bit stream decomposing unit 111 The bit stream decomposing unit 111 analyzes the serially input bit stream 1601,
Picture head, slice head, macroblock MB
And the beginning of a block are detected. Also, the bitstream decomposition unit 111 converts the coded picture type 1611 (which indicates the I picture, P picture, or B picture), the coded macroblock type 1615, the coded block pattern 1614,
The encoded motion vector 1612 and the encoded quantized DCT coefficient 1613 are extracted, and the extracted encoded picture type 1611, encoded macro block type 1615, encoded block pattern 1614, and encoded motion vector 16 are extracted.
12. The encoding and quantization DCT coefficient 1613 is decoded by the decoding unit 112.
Output to

Further, the bit stream decomposition section 111
The decoding control unit 110 receives an encoding error 1602, an out-of-specified range 1603, or a motion vector error 1604. The bit stream decomposing unit 111 includes the decoding control unit 11
If the coding error 1602, out-of-specified range 1603, or motion vector error 1604 is received from 0, the slice currently being decomposed is skipped and the detection of the next slice is started.

When the bit stream decomposing unit 111 detects an end code from the bit stream 1601, the bit stream decomposing unit 111
1 is terminated. (2) Decoding Unit 112 The decoding unit 112 converts the coded picture type 1611, the coded macroblock type 1615, the coded block pattern 1614, the coded motion vector 1612, and the coded quantized DCT coefficient 1613 into the bit stream
Receive from 1.

The decoding unit 112 performs the encoding picture type 1
611, coded macroblock type 1615, coded block pattern 1614, coded motion vector 161
2. It has a decoding table for decoding the coded quantized DCT coefficients 1613. The decoding unit 112 receives the coded picture type 1611, the coded macro block type 1615, the coded block pattern 1614,
The encoded motion vector 1612 is decoded using the decoding table to generate a picture type 1623, a macro block type 1628, a block pattern 1624, and a motion vector 1625, respectively, and the generated picture type 1623, macro block type 1628, and block pattern are generated. 1624, and notifies the decoding control unit 110 of the motion vector 1625.

The decoding unit 112 calculates the run length 1621 and the effective coefficient value 1 from the received coded quantized DCT coefficient 1613.
At least one pair consisting of 631 and 631 is decoded using the decoding table, the decoded run length 1621 is sent to the decoding control unit 110, and the decoded effective coefficient value 1631 is output to the second selecting unit 114. Received coded picture type 1611, coded macroblock type 161
5, while decoding the coded block pattern 1614, the coded motion vector 1612, and the coded quantized DCT coefficient 1613, the decoding unit 112 receives the coded picture type 1611, the coded macro block type 1615, the coded block pattern 1614, coded motion vector 1
612, when detecting that the coded quantized DCT coefficient 1613 is not present in the decoding table,
The decoding unit 112 determines that the error is the encoding error 1602, and notifies the decoding control unit 110 of the encoding error 1602. (3) Second Constant Generation Unit 113 The second constant generation unit 113 generates “constant 0” 1632. (4) Decoding Control Unit 110 The decoding control unit 110 outputs a coding error 1602, a run length 1621, a picture type 1623,
Macro block type 1628, block pattern 16
24, receives the motion vector 1625, and receives the image
19, a motion vector error 1604 is received.

If the motion vector 1625 notified from the decoding unit 112 is a value outside the predetermined range, the decoding control unit 110 determines that the error is outside the specified range. Note that the coding error and out of the specified range are errors indicating that the compressed code string is out of the specified range.

When decoding control section 110 receives encoding error 1602 from decoding section 112, or when decoding control section 1
When 10 recognizes an error 1603 outside the specified range, the decoding control unit 110 notifies the bitstream decomposition unit 111 and the image restoration unit 119 of the coding error 1602 or the out-of-specified range 1603, respectively. Also, the decryption control unit 1
10, the motion vector error 1604 is stored in the image restoration unit 119;
When notified, the motion vector error 1604 is notified to the bit stream decomposition unit 111.

The decoding control unit 110 notifies the image restoration unit 119 of the motion vector 1625 received from the decoding unit 112. When the coded quantized DCT coefficient 1613 is converted to the quantized DCT coefficient 1641, the decoding control unit 11
0 controls the second selector 114 as follows. FIG.
Indicates a pattern example of a block pattern. In the figure,
The shaded portion is a block that has a difference from the corresponding macroblock in the previous frame or the two preceding and succeeding frames. As shown in FIG. 25, the block pattern 327 expresses, for each pattern 326, whether or not each block included in the macroblock has a difference from the previous frame or two frames before and after.
Referring to the pattern 326, it can be determined whether each block includes a difference. The decoding control unit 110 stores the block pattern 324 and the pattern 326 in association with each other as a pattern correspondence information table 327. The decoding control unit 110 specifies one of the patterns 1626 from the pattern correspondence information table 327 using the block pattern 1624. Next, the decryption control unit 110
Determines, using the specified pattern, whether the block being decoded is a block having a difference or a block having no difference, and the decoding control unit 110 provides a flag called a block difference flag, and When there is a block, a value of 1 is set to the block difference flag, and when the block has no difference, a value of 0 is set to the block difference flag.

When the block difference flag is 1, the decoding control section 110 outputs to the second selecting section 114 the “constant 0” 1632 output by the second constant generating section 113 by the number of the run lengths 1621 received. Selection instruction 1 to select
633 is output, and thereafter, control is performed so that the effective coefficient value 1631 output from the decoding unit 112 is selected. When the block difference flag is 0, that is, when the block is a block to be skipped, the decoding control unit 110 instructs the first selection unit 118 to output “constant 0”, which is the output of the first constant generation unit 117.
A selection instruction 1712 is selected so as to select 64 1632s.
Is output.

In this way, the selected effective coefficient value 1
Quantized DCT combining 631 and “constant 0” 1632
A coefficient 1641 is generated. The decryption control unit 110
Specifies the predictive encoding method of the frame image of the macroblock being decoded, using the received macroblock type 1628. Macro block type 1628
Has a predictive coding method, forward predictive coding,
Each type of backward prediction coding, bidirectional prediction coding, or intra-frame coding is included. When the macroblock type 1628 is the forward predictive coding, the backward predictive coding, or the bidirectional predictive coding, the decoding control unit 110 recognizes that at least one block includes a difference, and 114 is controlled. (5) Second Selection Unit 114 The second selection unit 114 selects the first instruction from the decoding control unit 110.
633, the effective coefficient value 16 output from the decoding unit 112
31 or “constant 0” 1632 output by the second constant generator 113 is selected. Thus, the selected effective coefficient value 1631 and the “constant 0” 1632 are combined to generate a quantized DCT coefficient 1641. The generated quantized DCT coefficient 1641 is output to inverse quantization section 115.

When selecting “constant 0” 1632 output from second constant generating section 113, second selecting section 114 outputs “constant 0” 1632 output from decoding control section 110.
In response to the designation of the number and the selection instruction 1633, “Constant 0” 1632 for the designated number is selected. (6) Inverse Quantization Unit 115 The inverse quantization unit 115 inversely quantizes the quantized DCT coefficient 1641 output from the second selection unit 114 to obtain the DCT coefficient 1
651, and outputs the generated DCT coefficient 1651 to the inverse discrete cosine transform unit 116. (7) Inverse discrete cosine transform unit 116 The inverse discrete cosine transform unit 116 performs an inverse discrete cosine transform on the DCT coefficient 1651 output from the inverse quantization unit 115 to generate a restoration unit image 1661, and generates the generated restoration unit. The image 1661 is output to the first selection unit 118. (8) First Constant Generation Unit 117 The first constant generation unit 117 uses “constant 0” 1662 instead of pixel data when there is a block to be skipped.
Are output to the image restoration unit 119 by the selection of the first selection unit 118. (9) First Selection Unit 118 The first selection unit 118 selects the first instruction from the decoding control unit 110.
In response to 712, a restoration unit image 1661 output from the inverse discrete cosine transform unit 116 or “constant 0” 1662 output from the first constant generation unit 117 is selected. Next, the first selection unit 118 outputs the selected “constant 0” 1662 and the selected restoration unit image 1661 to the image restoration unit 119.

Restore unit image 1661 and “constant 0” 166
2 together will be referred to as an integrated unit image 1663. In addition,
When selecting the “constant 0” 1662 output from the first constant generator 117, the first selector 118 selects the decoding controller 110
"Constant 0" 1 together with selection instruction 1712 output from
A value of “64” is designated as the number of 662, and 64 designated “constants 0” 1662 are selected. (10) Image Storage Unit 120 The image storage unit 120 stores the frame image 1692 being decoded and the plurality of reference frame images 1691 already restored. (11) Image Restoring Unit 119 The image restoring unit 119 is notified of the motion vector 1625, the picture type 1623, the macroblock type 1628, the encoding error 1602, and the out-of-specified range 1603 from the decoding control unit 110, and performs the first selection. From part 118,
“Constant 0” 1662 and the restoration unit image 1661 are received.

Using the notified macroblock type 1628, the image restoration unit 119 is a block in which the block being decoded is included in any one of an intra macroblock, a forward prediction macroblock, and a backward prediction macroblock. Judge. If the integrated unit image 1663 composed of the block being decoded is a part of the I picture, the image restoration unit 119 converts the restored unit image 1661 composed of the block being decoded into a frame image being decoded in the image storage unit 120. Write directly to 1692. If the integrated unit image 1663 composed of the block being decoded is a part of a P picture or a B picture, the image storage unit 1
20, a reference unit image 1686, which is a block corresponding to the integrated unit image 1663 composed of the block being decoded, is read out, and the integrated unit image 1663 composed of the block being decoded and the read reference unit read out. The image 1686 is added to generate a combined unit image 1687 composed of a new block, and the generated combined unit image 1687 is written to the frame image 1692 being decoded in the image storage unit 120.

The image restoration section 119 sets the restoration unit image at the position indicated by the motion vector 1625 in the reference frame image 1691 as the reference unit image 1686. A process of determining a unit image in a reference frame image based on a motion vector and adding the unit image to a difference image to generate a restored unit image is called a motion compensation process. When the integrated unit image 1663 composed of the block being decoded is a P picture or a B picture, the notified motion vector 1625 of the integrated unit image 1663 being decoded is stored in the reference frame image 1691 in the image storage unit 120. If the value indicates a value outside the range, the image restoration unit 119 determines that the motion vector error is 1604.

When the image restoration unit 119 determines that the motion vector error 1
If it is determined that the motion vector 604 exists, the motion vector 162
Since 5 is represented by the difference from the motion vector of the previous macroblock, the image restoration unit 119 cannot correctly restore the motion vector of this slice and subsequent blocks in the slice. For this reason, the image restoration unit 119 stops decoding of an image after the block being decoded, among slices including the block being decoded. Instead, the image restoration unit 119 outputs the frame image 1 being decoded.
The image after the second block included in the second slice in the reference frame image 1691 at the same position as the first block included in the first slice in 692 is read out,
The read image after the second block is written into the frame image 1692 being decoded in the image storage unit 120.
This process is called an image compensation process at the time of error. The details of the image compensation processing at the time of error will be described later.

The image restoration unit 119 detects the coding error 1602
And the out-of-specified range 1603, the image of the block in which these errors have occurred cannot be decoded. Instead, as in the case of the motion vector error 1604, the image restoration unit 119 outputs the frame image 1 being decoded.
The image after the second block included in the second slice in the reference frame image at the same position as the first block included in the first slice in 692 is read, and the read image after the second block is read out. The image is written into the frame image 1692 being decoded in the image storage unit 120.

The unit image read from the reference frame image 1691 and written to the frame image 1692 being decoded instead of the block in which the error has occurred is called an alternative unit image. The image restoration unit 119 notifies the decoding control unit 110 of the motion vector error 1604. 1.2 Data Flow of Image Decoding Device Next, a data flow of the image decoding device shown in FIG. 1 will be described with reference to FIGS. Note that FIG.
In FIG. 3 and FIG. 3, a box surrounded by a single line indicates a component of the image decoding apparatus shown in FIG. 1, and a box surrounded by a double line indicates data exchanged between the components. .

The bit stream 1601 is externally output to the bit stream decomposition unit 111. The encoding error 1602, out-of-specified range 1603, and motion vector error 1604 are output from the decoding control unit 110 to the bitstream decomposition unit 111. Encoded picture type 161
1, coded macroblock type 1615, coded motion vector 1612, coded quantized DCT coefficient 1613,
The encoded block pattern 1614 is output from the bitstream decomposition unit 111 to the decoding unit 112.

A run length 1621, a coding error 1602,
Picture type 1623, macro block type 162
8, block pattern 1624, motion vector 1625
Is output from the decoding unit 112 to the decoding control unit 110. Effective coefficient value 1631 is output from decoding section 112 to second selection section 114. “Constant 0” 1632 is the second
It is output from constant generation section 113 to second selection section 114.

The run length 1621 and the selection instruction 1633 are output from the decoding control unit 110 to the second selection unit 114.
The quantized DCT coefficient 1641 is obtained from the second selecting unit 114.
Output to inverse quantization section 115. DCT coefficient 1651
From the inverse quantization unit 115 to the inverse discrete cosine transform unit 11
6 is output.

The restored unit image 1661 is output from the inverse discrete cosine transform unit 116 to the first selecting unit 118. “Constant 0” 1662 is output from first constant generating section 117 to first selecting section 118. The selection instruction 1712 is output from the decoding control unit 110 to the first selection unit 118.

Restore unit image 1661 and “constant 0” 166
The unit image 1663 composed of the first selection unit 11
8 to the image restoration unit 119. The motion vector error 1604 is sent from the image restoration unit 119 to the decoding control unit 1
It is output to 10. The motion vector 1625, the picture type 1623, the macroblock type 1628, the encoding error 1602, and the out-of-specified range 1603
10 to the image restoration unit 119.

The reference unit image 1686 is stored in the image storage 12
0 to the image restoration unit 119. The combined unit image 1687 is transmitted from the image restoration unit 119 to the image storage unit 120.
Output to 1.3 Operation of Image Decoding Device Next, the operation of the image decoding device shown in FIG.
P picture or B picture
The processing operation of a block to be skipped when it is a part of a picture will be briefly described.

The bit stream decomposing unit 111 analyzes the bit stream 1601 and encodes a coded picture type 1611, a coded macro block type 1615, a coded block pattern 1614, and a coded motion vector 16.
12, the coded quantized DCT coefficient 1613 is taken out, and the taken out coded picture type 1611, coded macro block type 1615, coded block pattern 16
14, coded motion vector 1612, coded quantization DC
Output T coefficient 1613 to decoding section 112 (step S
20).

When bit stream decomposition section 111 detects an end code indicating the end of data from bit stream 1601, the image decoding apparatus ends the processing (step S21). The decoding unit 112 includes a coded picture type 1611, a coded macroblock type 1615, a coded block pattern 1614, a coded motion vector 1
612, the coded quantized DCT coefficient 1613 is received from the bit stream decomposition unit 111, and the received coded picture type 1611 and coded macro block type 1 are received.
615, the encoded block pattern 1614, and the encoded motion vector 1612,
Macro block type 1628, block pattern 16
24, a motion vector 1625 is generated, and the generated picture type 1623, macro block type 1628,
The block pattern 1624 and the motion vector 1625 are notified to the decoding control unit 110, and the decoding unit 112 calculates the run length 16 from the received coded quantized DCT coefficient 1613.
21 and the effective coefficient value 1631, and the decoding unit 112 further receives the coded picture type 1611,
While decoding the encoded macroblock type 1615, the encoded block pattern 1614, the encoded motion vector 1612, and the encoded quantized DCT coefficient 1613, the received encoded picture type 1611, encoded macroblock type 1615, encoded block pattern 1614, encoded motion vector 1612, encoded quantized DCT coefficient 16
When the decoding unit 112 detects that the encoding error 160 does not exist in the decoding table, the decoding unit 112
2, and notifies the decoding control unit 110 of the encoding error 1602. (Step S29).

The decoding control unit 110
When the motion vector 1625 is received and the received motion vector 1625 is out of the predetermined range, it is determined that the error is out of the predetermined range 1603 (step S22).
Has received the coding error 1602 from the decoding unit 112 (step S23), the decoding control unit 110 sends the coding error 1602 or the out-of-specified range 1603 to the bitstream decomposition unit 111 and the image restoration unit 119, respectively. Then, the same operation as when a motion vector error 1604 described later occurs is performed.

The decoding control unit 110
The block pattern 1624 is received, and a value of 0 or 1 is set in the block difference flag based on the block pattern 1624. When the block difference flag is 1, that is, when the block is not a block to be skipped (step S2).
4) The decoding control unit 110 selects the “constant 0” 1632 output from the second constant generation unit 113 by the number of the received run lengths 1621 to the second selection unit 114, and thereafter, the decoding unit 112 The second selector 114 selects the effective coefficient value 1631 or “constant 0” 1632 under the control of the decoding control unit 110 so as to select the effective coefficient value 1631 output by Quantization D
The CT coefficient 1641 is generated, and the inverse quantization unit 115 inversely quantizes the quantized DCT coefficient 1641 to generate the DCT coefficient 165.
1, and the inverse discrete cosine transform unit 116 performs an inverse discrete cosine transform of the DCT coefficient 1651 to obtain the restored unit image 1
661, and outputs the generated restoration unit image 1661 to the first selection unit 118 (step S26).

If the integrated unit image 1663 composed of the block being decoded is a part of an intra macro block, the image restoration unit 119 decodes the restored unit image 1661 composed of the block being decoded in the image storage unit 120. The data is directly written into the middle frame image 1692. If the integrated unit image 1663 composed of the block being decoded is a forward prediction macroblock or a part of a backward or bidirectional prediction macroblock, the block being decoded is obtained from the reference frame image 1691 in the image storage unit 120. The reference unit image 1686, which is a block corresponding to the integrated unit image 1663, is read, and the integrated unit image 1663, which is the block being decoded, and the read reference unit image 1686 are added to generate a combined unit, which is a new block. An image 1687 is generated, and the generated combined unit image 1687 is written in the frame image 1692 being decoded in the image storage unit 120 (step S27).

When the block difference flag is 0, that is, when the block is a block to be skipped (step S24), the decoding control unit 110 sends the first selection unit 118
“Constant 0” 1662 which is the output of the constant generation unit 117 is
The first constant generation unit 117 is controlled so as to select four.
Outputs the block consisting of “constant 0” 1662 to the image restoration unit 119 by the selection of the first selection unit 118 (step S28). After that, in step S27, the integrated unit image 1663 being decoded and the reference unit image 1686 in the image storage unit 120 are added, and the combined unit image 1687 is added.
Is generated, and the combined unit image 1687 is written into the frame image 1692 being decoded in the image storage unit 120.

When step S27 ends, step S
Return to 20. Next, a processing operation when an encoding error 1602, an out-of-specified range 1603, or a motion vector error 1604 has occurred will be briefly described with reference to the flowchart of FIG. The flowchart shown in FIG. 5 is based on the image restoration unit 119, the decoding unit 112, and the decoding control unit 1.
10. Note that only the processing operations related to the case where an encoding error 1602, out of the prescribed range 1603, or a motion vector error 1604 has occurred in the bit stream decomposition unit 111 are shown, and not all the processing operations are shown. Cost. In the flowchart of FIG.
The part shown by the broken line indicates that the processing not shown in the figure operates.

The image restoration unit 119 determines that the motion vector error 1
When the image restoration unit 119 detects the motion vector error 1604 (step S201), the image restoration unit 119
10 (step S202), and then performs image compensation processing at the time of error (step S204). On the other hand, when receiving the motion vector error 1604 from the image restoration unit 119, the decoding control unit 110
1 is notified (step S211). Upon receiving the motion vector error 1604 (step S221), the bitstream decomposition unit 111 skips the slice currently being decomposed (step S222) and starts reading the next slice (step S223). In this way,
The image compensation processing at the time of error and the skip processing of the slice currently being decomposed are performed in parallel.

Next, the decoding unit 112 determines that the encoding error 16
02 (step S205), the decoding unit 112
Notifies the decoding error 1602 to the decoding control unit 110 and the image restoration unit 119 (step S206). The decoding control unit 110 receives the encoding error 1602, and notifies the encoding error 1602 to the bitstream decomposing unit 111 (step S212). As in the case where the motion vector error 1604 is detected, the bitstream decomposition unit 111 skips the slice currently being decomposed (step S222) and starts reading the next slice (step S223). Further, the image restoration unit 119 performs an image compensation process at the time of error (step S204).

Further, when the decoding control unit 110
Upon detecting 603, the decoding control unit 110 notifies the bitstream decomposing unit 111 and the image restoring unit 119 of the out-of-specified range 1603 (step S210). As in the case where the motion vector error 1604 is detected, the bitstream decomposition unit 111 skips the slice currently being decomposed (step S222) and starts reading the next slice (step S223). Further, the image restoration unit 119 performs image compensation processing at the time of error (Step S).
204). 1.4 Transition of Processing with Time Elapsed by Image Decoding Apparatus FIG. 6 is a time chart showing transition of processing with time elapsed for each block of the image decoding apparatus shown in FIG.
In the vertical direction, the bit stream decomposing unit 111 and the decoding unit 11
2. Each processing unit of the first constant generation unit 117, the inverse quantization unit 115, the inverse discrete cosine transform unit 116, and the image restoration unit 119 is listed, and time is taken in the horizontal direction. A block B1 including a pixel, a block B2 to be skipped, a block B3 including a pixel, a block B4 including an error, and a block B5 including a pixel flow through the image decoding apparatus illustrated in FIG. 1 in this order, and C1 to C4, C5 to C5. C7, C8-C11, C1
2 to C13 and C15 to C18 are block B, respectively.
The processing of each of the processing units 1, B2, B3, B4, and B5 is shown.

The processing for each block includes C1, C2, C
3, C4, C5, C6, C7, C8, C9, C10, C
It operates in the order of 11, C12. Thereafter, the processes are simultaneously started in C13 and C14, and after completion of C14, C1
5, C16, and C17. In this manner, the inverse quantization and the inverse discrete cosine transform are omitted for the blocks to be skipped, as shown at C5 to C7, and for the blocks containing errors, as shown at C12 to C13. 1.5 Modifications of Embodiment Although the present invention has been described based on the above-described embodiment, it is needless to say that the present invention is not limited to the above-described embodiment. That is, the following cases are also included in the present invention. (1) Embodiment in which the Unit Image of the Image Decoding Device is a Macroblock In the image decoding device shown in FIG. 1, the unit image is a block, but may be a macroblock. The operation of a case where a unit image is a macroblock will be briefly described below.

As described above, there are a case where a block constituting a macroblock is a block which is not skipped, and a case where it is a block which is skipped. If all the blocks included in the macroblock are blocks to be skipped, the first constant generation unit 117 generates a macroblock with a constant 0, and the decoding control unit 110 sends a first selection unit 118 Selecting a macroblock having the pixel value 0 generated by the first constant generation unit 117 as an element,
The output is controlled to be output to the image restoration unit 119.

If at least one block included in the macro block is a block that is not skipped, the second constant generation unit 113 generates a block with a constant 0, and the decoding control unit 110 determines whether the skipped block is , The second selection unit 114, the second constant generation unit 11
The decoding control unit 110 controls so as to select a block including the constant 0 generated by 3 as an element, and for a block that is not skipped, the decoding control unit 110
An effective coefficient value 1631 decoded by the decoding unit 112;
Control is performed such that a constant 0 equal to the number of run lengths 1632 decoded by the decoding unit 112 is selected. After that, the blocks composed of the constants and the constants for the number of the effective coefficient values 1631 and the run lengths 1632 are output from the second selecting unit 114 to the inverse quantization unit 115.

With this configuration, when a macro block is composed of skipped blocks, it is possible to avoid a reduction in processing speed without passing through a process of dequantizing the macro block and performing an inverse discrete cosine transform. For example, in the case of sequentially decoding a macroblock M1 in which all blocks are blocks to be skipped and a macroblock M2 including at least one non-skipped block, the processing time of the conventional image decoding apparatus is as follows.
The processing time by the image decoding device shown in FIG. 1 can be expressed by (Expression 2). That is, the image decoding device shown in FIG. 1 can reduce the processing time by (image conversion time of the macroblock M1) as compared with the conventional image decoding device. Thus, when all the blocks included in the macroblock do not include the pixel value,
A reduction in processing speed can be avoided without passing through image conversion means that is not originally required. (Equation 1) Processing time by conventional image decoding apparatus = (image decoding time of macro block M1) + (image conversion time of macro block M1) + (image restoration time of macro block M1) + (image of macro block M2) (Decoding time) + (Image conversion time of macroblock M2) + (Image restoration time of macroblock M2) (Equation 2) Processing time by image decoding apparatus shown in FIG. 1 = (Image decoding time of macroblock M1) + ( (Image restoration time of macro block M1) + (Image decoding time of macro block M2) + (Image conversion time of macro block M2) + (Image restoration time of macro block M2) (2) Unit image and restoration of image decoding device In the above embodiment, the unit image and the restored unit image are both blocks. In block,
The restoration unit image may be a macro block.

The image restoring unit 119 includes an integrated unit image storage unit that stores six integrated unit images. Are stored in the integrated unit image storage unit. Upon completion of storing the six integrated unit images in the integrated unit image storage unit, the image restoring unit 119 generates six restored unit images sequentially from the six integrated unit images and generates the restored unit images. Image restoration unit 12 stores six restoration unit images.
0 is written to the restored frame image 1692. (3) Embodiment for Generating Constant from First Constant Generating Unit When Error Occurs Even if encoding error 1602 or error 1603 out of the specified range occurs, the same processing as in the case of a block to be skipped is performed. good. That is, the decryption control unit 11
0 indicates an encoding error 1602 or out of the specified range 1603
To the bit stream decomposition unit 111 and the image restoration unit 119
, The decoding control unit 110 sets the first constant generation unit 11
The first selection unit 118 may be controlled so as to select 64 constants 0 which are the outputs of 7.

In this way, instead of the block in which the error has occurred, 64 blocks composed of 64 constants 0 are output from the first selecting section 118 to the image restoring section 119, so that the image restoring section 119 The error image of the block in which the error has occurred can be compensated without performing a complicated error image compensation process. In this case, the immediately preceding reference image is restored instead of the block in which the error has occurred. (4) Embodiment in which the number of pixels included in a block and a macroblock is different In the above-described embodiment, the block is composed of 64 pixels. However, it is needless to say that the number of pixels is not limited to 64. One block may be configured with 256 pixels by arranging 16 pixels vertically and 16 pixels horizontally.

Although the macro block is composed of six blocks, the macro block may be composed of 16 blocks representing luminance and two blocks representing color difference. Thus, there is no limitation on the number of pixels included in a block or macroblock. (5) Another Embodiment of First Constant Generating Unit The first constant generating unit 117 includes, for example, 4 having a pixel value of 0.
May be output, and the first selection unit 118 may select four elements having a pixel value of 0 16 times. As a result, an element having a total of 64 pixel values 0 is selected. That is, the first constant generation unit 117 outputs a plurality of elements including the pixel value 0, and outputs the first selection unit 118
May be selected a plurality of times so that a total of 64 elements are output.

Further, the first constant generation unit 117 supplies a constant generation unit for generating one pixel value 0 to the constant generation unit for the number of pixels included in the block, that is, 64 pixel values 0. Are repeatedly generated, and the generated 64
And a constant control unit that outputs a pixel value 0 for each pixel. Further, the first constant generation unit 117 generates one pixel value 0, and the first selection unit converts one pixel value 0 generated by the first constant generation unit into the number of pixels included in the block, That is, 64 pieces may be selected. At this time, the image restoration unit 119 adds the 64 pixel values 0 and the reference unit image 1686 read from the image storage unit 120. (6) Embodiment in case of intra-frame prediction In the above-described embodiment, a case has been described in which a compressed code string predictively encoded between frame images is decoded.
The above embodiment can be similarly applied to a case where a compressed code sequence obtained by encoding a unit image in a frame image using a difference from another unit image in the same frame image is decoded. (7) Processing Unit at the Time of Error Occurrence In the above embodiment, when an encoding error 1602, a motion vector error 1604, or an out-of-specified range 1603 is detected, the bitstream decomposition unit 111 Is skipped, and the reading of the next slice is started.
The block currently being decomposed may be skipped, and reading of the next block may be started. Alternatively, the bitstream decomposition unit 111 may skip the macroblock currently being decomposed and start reading the next macroblock. Further, the bit stream decomposition unit 111
A plurality of macroblocks including the currently decomposed macroblock may be skipped, and reading of the next plurality of macroblocks may be started. 2. Second Embodiment Next, an image decoding device according to a second embodiment of the present invention will be described. 2.1 Schematic Configuration of Image Decoding Device FIG. 7 is a block diagram illustrating a schematic configuration of an image decoding device as another embodiment of the present invention.

The image decoding apparatus shown in FIG.
0, information storage unit 10a, decoding control unit 11, bit stream decomposition unit 11a, decoding unit 11b, second constant generation unit 11
c, second selection unit 11d, conversion control unit 12, inverse quantization unit 1
2a, the inverse discrete cosine transform unit 12b, the restoration control unit 13,
It comprises a first constant generation unit 13a, a first selection unit 13b, an image restoration unit 13c, a first image storage unit 14, a second image storage unit 15, and a third image storage unit 16.

The decoding control unit 11, bit stream decomposition unit 11a, decoding unit 11b, second constant generation unit 11c, second selection unit 11d, inverse quantization unit 12a, inverse discrete cosine transform unit 12b, first constant generation unit 13a , The first selection unit 13b, the image restoration unit 13c, and the first image storage unit 14 are respectively a decoding control unit 110, a bitstream decomposing unit 111, a decoding unit 112, and a second constant of the first embodiment shown in FIG. It corresponds to the generation unit 113, the second selection unit 114, the inverse quantization unit 115, the inverse discrete cosine transform unit 116, the first constant generation unit 117, the first selection unit 118, the image restoration unit 119, and the image storage unit 120.

Next, the configuration of the image decoding apparatus shown in FIG. 9 will be described. The second constant generation unit 11c and the first image storage unit 14 are the same as those of the first embodiment shown in FIG. The description is omitted because they are the same as the generation unit 113 and the image storage unit 120.
This will be described below. It should be noted that the unit image of the image decoding apparatus in this figure is a block.

The processing realized by the decoding control unit 11, the bit stream decomposition unit 11a, the decoding unit 11b, the second constant generation unit 11c, and the second selection unit 11d is a code decoding process, a conversion control unit 12, and an inverse quantization unit. 12a, a process realized by the inverse discrete cosine transform unit 12b is an image conversion process, a restoration control unit 13, a first constant generation unit 13a, and a first selection unit 13.
b, the process realized by the image restoration unit 13c is referred to as an image restoration process. (1) Transform Control Unit 12 The transform control unit 12 is provided to control the inverse quantization unit 12a and the inverse discrete cosine transform unit 12b in an integrated manner. Upon receiving the instruction to start image conversion 1703 from the sequential control unit 10, the conversion control unit 12
Then, the image conversion is started 170 such that the inverse quantization process is started.
Indicate 3. Further, when the information of the image conversion completion 1704 is received from the inverse discrete cosine transform unit 12b, the information of the image conversion completion 1704 is sequentially notified to the control unit 10. (2) Second Image Storage Unit 15 The second image storage unit 15 temporarily stores the output of the code decoding process and then inputs the output of the image conversion process so that the code decoding process and the image conversion process operate simultaneously. It is provided for.

Specifically, the second image storage unit 15 stores the second
The output from the decoder 11b and the output from the second constant generator 11c selected by the selector 11d are stored. (3) Third Image Storage Unit 16 The third image storage unit 16 stores the output of the image conversion process once, and then inputs the output of the image restoration process, thereby simultaneously operating the image conversion process and the image restoration process. It is provided for.

More specifically, the third image storage section 16 stores the output of the inverse discrete cosine transform section 12b. (4) First Constant Generation Unit 13a When there is a block to be skipped, the first constant generation unit 13a sends a block consisting of a constant 0 instead of a pixel to the image restoration unit 13c via the first selection unit 13b. Output. (5) First Selection Unit 13b The first selection unit 13b selects the data stored in the third image storage unit 16 or the output of the first constant generation unit 13a. (6) Restoration Control Unit 13 The restoration control unit 13 performs integrated control of the first constant generation unit 13a, the first selection unit 13b, and the image restoration unit 13c, and transmits information between the sequential control unit 10 and the image restoration unit 13c. It is provided for.

More specifically, upon receiving an instruction to start image restoration 1705 from the sequential control unit 10, the restoration control unit 13
The selection instruction 1712 is output to the first selection unit 13b to select the data stored in the third image storage unit 16 or the output of the first constant generation unit 13a, and the image restoration unit 13c notifies the image restoration completion 1706. Are received, the control unit 10 is sequentially notified of the information of the image restoration completion 1706.

Further, the restoration control unit 13 includes the image restoration unit 13
When the motion vector error 1604 is received from c, the control unit 10 is sequentially notified of the motion vector error 1604,
When receiving the motion vector 1625, the picture type 1623, and the macroblock type 1628 from the sequential control unit 10, the motion vector 1625 is sent to the image restoration unit 13c.
Picture type 1623 and macroblock type 1
628 is notified.

Further, the restoration control unit 13 sequentially controls the
From 0, when receiving an encoding error 1602 and an error out of the specified range 1603, it notifies the image restoration unit 13c of the encoding error 1602 and the error out of the specified range 1603. Further, the restoration control unit 13 receives the block pattern 1624 from the sequential control unit 10 and, if the block is to be skipped, performs the first selection so as to select 64 constants 0 output from the first constant generation unit 13a. It controls the unit 13b. (7) Sequential Control Unit 10 The sequential control unit 10 activates the code decoding process when the execution condition of the code decoding process is satisfied, and prevents the code decoding process from operating twice if the execution condition is not satisfied. To control.
The sequential control unit 10 performs the same control on the image conversion processing and the image restoration processing. In order to perform this control, the sequential control unit 10 instructs the decoding control unit 11 to start code decoding 1701.
To the conversion control unit 12,
It instructs the restoration control unit 13 to start image restoration 1705. Further, the sequential control unit 10 receives a notification of the completion of code decoding 1702 from the decoding control unit 11, a notification of the completion of image conversion 1704 from the conversion control unit 12, and a notification of the completion of image restoration 1706 from the restoration control unit 13. The details of the control operation will be described later. In addition, the sequential control unit 10 mediates information transmission between the code decoding process, the image conversion process, and the image restoration process. Specifically, upon receiving the notification of the motion vector error 1604 from the restoration control unit 13, the sequential control unit 10 sends the motion vector error 160
4 is notified. When receiving the notification of the coding error 1602, the out-of-specified range 1603, the block pattern 1624, the motion vector 1625, the picture type 1623, or the macroblock type 1628 from the decoding control unit 11, the sequential control unit 10 Out of specified range 16
03, block pattern 1624, motion vector 162
5. The restoration control unit 13 is notified of the picture type 1623 or the macro block type 1628. Communication of information
The synchronization is performed so that each piece of information matches the block corresponding to the information. (8) Information storage unit 10a The information storage unit 10a stores, for each block to be processed, a motion vector 1625, a picture type 1623, a macro block type 1628, a block pattern 1624, an encoding error 1602, a non-defined range 1603, a motion vector The error 1604 is stored. (9) Bitstream Decomposition Unit 11a When the bitstream decomposition unit 11a receives an instruction to start bitstream decomposition 1711 from the decoding control unit 11,
Start the bitstream decomposition process. (10) Decoding Unit 11b When decoding of one block is completed, the decoding unit 11b notifies the decoding control unit 11 of code decoding completion 1702. (11) Decoding control unit 11 The decoding control unit 11 includes a bit stream decomposition unit 11a, a decoding unit 11b, a second constant generation unit 11c, and a second selection unit 11d.
And a sequential control unit 10, a bit stream decomposition unit 11a, a decoding unit 11b, and a second constant generation unit 11c.
And information communication with the second selection unit 11d. Specifically, upon receiving an instruction to start code decoding 1701 from the sequential control unit 10, the decoding control unit 11 instructs the bit stream decomposing unit 11a to start bit stream decomposition 1711, and the decoding unit 11b outputs a code decoding completion 1702. Is received, the information of the code decoding completion 1702 is sequentially notified to the control unit 10. Further, the decoding control unit 11 sequentially notifies the control unit 10 of an encoding error 1602, an error out of the prescribed range 1603, a motion vector 1625, a picture type 1623, a macroblock type 1628, or a block pattern 1624. Further, upon receiving the motion vector error 1604 from the sequential control unit 10, the decoding control unit 11 notifies the bit stream decomposing unit 11a of the error of the motion vector error 1604. (12) Second Selection Unit 11d The second selection unit 11d outputs the selected output from the decoding unit 11b and the output from the second constant generation unit 11c to the second image storage unit 15. (13) Inverse Quantization Unit 12a The inverse quantization unit 12a stores the image in the second image storage unit 15 when the image conversion start 1703 is output from the conversion control unit 12 to start the inverse quantization process. Start inverse quantization of data. (14) Inverse discrete cosine transform unit 12b The inverse discrete cosine transform unit 12b generates a unit image by performing an inverse discrete cosine transform on the output of the inverse quantization unit 12a, and transmits the generated unit image to the third image storage unit 16. Output. Upon completion of the output to the third image storage unit 16, the inverse discrete cosine transform unit 12 b sends the image conversion completion 170 to the conversion control unit 12.
4 is notified. (15) Image Restoring Unit 13c When detecting the motion vector error 1604, the image restoring unit 13c notifies the restoration control unit 13 of the motion vector error 1604. 2.2 Data Flow of Image Decoding Device Next, a data flow of the image decoding device shown in FIG. 7 will be described with reference to FIGS. 8, 9, 10, and 11. FIG.

The motion vector error 1604 and the code decoding start 1701 are sequentially output from the control unit 10 to the decoding control unit 11. Encoding error 1602, out of specified range 160
3, motion vector 1625, picture type 1623,
Macro block type 1628, block pattern 16
24 and the decoding completion 1702 are sequentially output from the decoding control unit 11 to the control unit 10.

The image conversion start 1703 is sequentially performed by the control unit 10.
Is output to the conversion control unit 12. Image conversion completed 170
4 are sequentially output from the conversion control unit 12 to the control unit 10. Image restoration start 1705, motion vector 1625, picture type 1623, macro block type 162
8, the coding error 1602, the out-of-specified range 1603, and the block pattern 1624 are sequentially output from the control unit 10 to the restoration control unit 13.

The image restoration completion 1706 and the motion vector error 1604 are sequentially output from the restoration control unit 13 to the control unit 10. Encoding error 1602, out of specified range 160
3, the motion vector error 1604 and the start of bitstream decomposition 1711 are output from the decoding control unit 11 to the bitstream decomposition unit 11a.

Code decoding completed 1702, run length 1621,
Encoding error 1602, picture type 1623, macroblock type 1628, block pattern 1624
And the motion vector 1625 are output from the decoding unit 11b to the decoding control unit 11. Encoded picture type 161
1, the encoded macroblock type 1615, the encoded motion vector 1612, the encoded quantized DCT coefficient 1613, and the encoded block pattern 1614 are output from the bitstream decomposition unit 11a to the decoding unit 11b.

The effective coefficient value 1631 is output from the decoder 11b to the second selector 11d. The run length 1621 and the effective coefficient value 1631 are output from the decoding control unit 11 to the second selection unit 11d. The quantized DCT coefficient 1641 is output from the second selector 11d to the second image storage 15.

The quantized DCT coefficient 1641 is output from the second image storage unit 15 to the inverse quantization unit 12a. The image conversion start 1703 is performed by the conversion control unit 12 and the inverse quantization unit 1.
2a. Image conversion completion 1704 is output from the inverse discrete cosine transform unit 12b to the conversion control unit 12.

The restoration unit image 1661 is output from the inverse discrete cosine transform unit 12b to the third image storage unit 16.
The restoration unit image 1661 is output from the third image storage unit 16 to the first selection unit 13b. “Constant 0” 1662 is
The first constant generator 13a outputs the data to the first selector 13b.

The selection instruction 1712 is output from the restoration control unit 13 to the first selection unit 13b. "Constant 0" 166
2 and an integrated unit image 16 composed of a restored unit image 1661
63 is output from the first selection unit 13b to the image restoration unit 13c. The image restoration completion 1706 and the motion vector error 1604 are sent from the image restoration unit 13c to the restoration control unit 13
Output to

Motion vector 1625, picture type 1
623, the macroblock type 1628, the coding error 1602, and the out-of-specified range 1603 are output from the restoration control unit 13 to the image restoration unit 13c. Reference unit image 16
86 is output from the first image storage unit 14 to the image restoration unit 13c.

[0104] The combined unit image 1687 is
c to the first image storage unit 14. 2.3 Transition of State of Code Decoding Processing, Image Conversion Processing, and Image Restoration Processing Next, FIG. 12 shows the state transition of code decoding processing, image conversion processing, and image restoration processing of the image decoding apparatus shown in FIG. It will be described using FIG.

The code decoding process is not executed 42 or being executed 4
It has three states: Bit stream 4 in which the state of the code decoding process is not executed 42 and the code decoding process is waiting
If 1 exists, the code decoding process can be executed.
When the state of the code decoding process is being executed 43 and the code decoding process is completed thereafter, the code decoding process is in a state of not being executed 42.

The image conversion process has two states, that is, not executed 44 or being executed 45. If the state of the image conversion processing is not executed 44 and there is data 48 waiting for the image conversion processing, the image conversion processing can be executed. When the state of the image conversion processing is being executed 45 and the image conversion processing is completed thereafter, the image conversion processing is in a state of not being executed 44.

Further, the image restoration process has two states, that is, “not executed” 46 and “executed 47”. Data 49 in which the state of the image restoration process is not executed 46 and the image restoration process is waiting
Exists, the image restoration processing can be executed. When the state of the image restoration processing is being executed 47 and the image restoration processing is thereafter completed, the image restoration processing is in a state of not being executed 46. 2.4 Operation of Sequential Control Unit Next, the operation of the sequential control unit 10 of the image decoding device shown in FIG. 7 will be described with reference to the flowcharts of FIGS. 13, 14, and 15.

In FIG. 13, first, the sequential control unit 10 initializes a flag used in the sequential control unit 10. The flag has ON and OFF states. The flags used by the sequential control unit 10 include a code decoding process flag, a code decoding process completion flag, an image conversion process flag, an image conversion process completion flag, a restoration process flag, and a restoration process completion flag. The sequential control unit 10 determines whether the code decoding process is in progress, the code decoding process completion flag, the image conversion process flag,
Initial values are set for the image conversion processing completion flag, the restoration processing flag, and the restoration processing completion flag (step S).
501).

When the bit stream decomposing unit 11a detects the end code of the bit stream and the sequential control unit 10 receives the last image restoration completion 1706 from the restoration control unit 13, the sequential control unit 10 sets the decoding control unit 11 and the conversion control Part 1
2. The control for the restoration control unit 13 is terminated (step S502). The sequential control unit 10 determines whether the code decoding process flag is ON or OFF. If the code decoding process flag is OFF, that is, the code decoding process is “unexecuted” (step S503), the bit for performing the code decoding process is determined. When a stream exists (step S504), the sequential control unit 10
Instructs the decoding control unit 11 to start the code decoding process (step S505), turns on the code decoding process flag (step S506), and continues the process. The decoding control unit 11 starts the code decoding process (step S507). When the code decoding process is completed and the code decoding process is notified from the decoding control unit 11, the decoding control unit 11 sequentially executes the code decoding process.
Turns on the code decoding process completion flag (step S
508).

If the code decoding process flag is OFF, ie, the code decoding process is “unexecuted” (step S 503), if there is no bit stream to be subjected to the code decoding process (step S 504), continue processing. When the code decoding process flag is ON, that is, the code decoding process is “executing” (step S50).
3), the sequential control unit 10 sets the code decoding process completion flag to O
Judge whether N or OFF, the code decoding process completion flag is ON
That is, when the code decoding process is completed (step S50).
9), the sequential control unit 10 sets the code decoding process completion flag to O
FF (step S510), the flag during the code decoding process is turned OFF (step S511), and the control unit 10
Continues the processing.

When the code decoding process flag is ON, that is, the code decoding process is “executing” (step S503), the sequential control unit 10 sets the code decoding process completion flag to OFF, that is, when the code decoding process is not completed. Is (Step S50
9), the control unit 10 continues the processing. Next, FIG.
In step 4, the sequential control unit 10 sets the image conversion process flag to ON.
It is determined whether or not the image conversion processing flag is OFF, that is, if the image conversion processing is “not executed” (step S51).
5) If there is data for image conversion processing in the second image storage unit 15 (step S516), the control unit 10
Instructs the conversion control unit 12 to start the image conversion processing (step S517), turns on the image conversion processing flag (step S518), and the control unit 10 continues the processing. When the conversion control unit 12 starts the image conversion process (step S519), the image conversion process is completed, and the conversion control unit 12 notifies the completion of the image conversion process, the sequential control unit 10 sets the image conversion process completion flag. Turn on (step S520).

If the image conversion processing flag is OFF, that is, if the image conversion processing is “unexecuted” (step S 515), the sequential control unit 10 determines that there is no data to be subjected to the image conversion processing in the second image storage unit 15. If it is (Step S516), the control unit 10 continues the process. The sequential control unit 10
If the image conversion processing flag is ON, that is, if the image conversion processing is “executing” (step S515), the sequential control unit 10
Determines whether the image conversion processing completion flag is ON or OFF. If the image conversion processing completion flag is ON, that is, if the image conversion processing is completed (step S521), the sequential control unit 10
Sets the image conversion processing completion flag to OFF (step S
522), the image conversion processing in-progress flag is turned off (step S523), and the control unit 10 continues the processing.

When the image conversion processing flag is ON, that is, when the image conversion processing is "executing" (step S515), the sequential control unit 10 turns off the image conversion processing completion flag, that is, when the image conversion processing is not completed. (Step S52
1) The control unit 10 continues the processing. Subsequently, in FIG. 15, the sequential control unit 10 determines whether the restoration processing flag is ON or OFF, and if the restoration processing flag is OFF, that is, the image restoration processing is “unexecuted” (step S52).
5) If there is data for image restoration processing in the third image storage unit 16 (step S526), the control unit 10
Instructs the restoration control unit 13 to start the image restoration processing (step S527), and turns on the restoration processing flag.
(Step S528), and the control unit 10 continues the processing. The restoration control unit 13 starts an image restoration process,
When the image restoration processing is completed, the sequential control unit 10 receives a notification of the completion of the image restoration processing from the restoration control unit 13, turns on the restoration processing completion flag (step S530), and the sequential control unit 10 continues the processing.

The sequential control unit 10 sets the restoration processing flag to O
If the FF, that is, the image restoration process is “unexecuted” (step S525), if there is no data for performing the image restoration process in the third image storage unit 16 (step S526), the control unit 10 sequentially executes the process. continue. When the restoration processing flag is ON, that is, when the image restoration processing is “executing” (step S525), the sequential control unit 10 determines that the restoration processing completion flag is ON, that is, when the image restoration processing is completed (step S525).
531), the sequential control unit 10 sets the copy processing completion flag to O
FF (step S532), and sets the copy processing flag to O.
FF is set (step S533), and the control unit 10 continues the process.

The sequential control unit 10 determines that the restoration processing flag is
N, that is, when the image restoration process is “executing” (step S525), when the restoration process completion flag is ON, that is, when the image restoration process is not completed (step S531), the control unit 10 continues the process. . Next, step S5
02, the control unit 10 continues the processing.

In this way, the sequential control unit 10 starts the code decoding process when the execution condition of the code decoding process is satisfied, and does not operate the code decoding process twice when the execution condition is not satisfied. Thus, the execution of the next block can be avoided until the processing of one block is completed. The same control is performed for the image conversion processing and the image restoration processing. 2.5 Transition of Processing with Time Elapsed by Image Decoding Apparatus FIG. 16 is a time chart showing transition of processing with time elapsed for each block of the image decoding apparatus shown in FIG. Each processing unit of the bit stream decomposition unit 11a, the decoding unit 11b, the first constant generation unit 13a, the inverse quantization unit 12a, the inverse discrete cosine transform unit 12b, and the image restoration unit 13c is listed in the vertical direction. Take the time. It is assumed that blocks B21, B22, B23, B24 including pixels and blocks B25, B26, B27, B28 to be skipped are processed in this order by the image decoding apparatus shown in FIG.
C61-C64, C65-C68, C69-C72, C
73-C76, C77-C79, C80-C82, C8
3 to C85 and C86 to C88 are blocks B2, respectively.
1, B22, B23, B24, B25, B26, B2
7 and B28 for each processing unit.

C61, C62, C63 and C64 operate in this order, and C65, C66, C67 and C68 operate in this order. In C65, the processing is started after the end of C62. That is, the processes of C63 and C65 are started at the same time. As described above, since different blocks can be processed at the same time, it can be seen that the processing time is shortened as compared with FIG. 3. Third Embodiment Here, as a third embodiment according to the present invention, details of the image compensation processing performed by the image restoration unit 119 at the time of an error in the first embodiment will be described.

The following embodiment can be similarly implemented in the image restoration unit 13c according to the second embodiment. 3.1 Configuration of Image Decoding Device Here, the configuration of the image decoding device according to the third embodiment of the present invention will be described.

Since the image decoding apparatus according to the third embodiment has the same configuration as the image decoding apparatus according to the first embodiment, only different parts will be described below. As shown in FIG. 17, the image storage unit 120 stores a first frame decoded into an image in a first storage area X1.
01 and the second that stores one frame decoded into an image
And a storage area X121. The first storage area X101 stores the frame immediately before the currently decoded frame, and the second storage area X121 stores the currently decoded frame. Alternatively, the first storage area X101 stores the frame currently being decoded, and the second storage area X121 stores the frame immediately before the frame currently being restored.

That is, the first storage area X101 and the second storage area X121 alternately store the frame image 1692 being decoded and one of the already restored reference frame images 1691, respectively. FIG. 17 shows a state where the first storage area X101 stores the frame immediately before the frame currently being decoded, and the second storage area X121 stores the frame currently being decoded.

The second storage area X121 has a plurality of slices X.
151. The second storage area X121 is
It is composed of one or a plurality of slices X131 that have already been decoded, a slice X132 in which an error has occurred during decoding, and one or a plurality of slices X133 that have not been decoded yet. The first storage area X101 includes a plurality of slices X141. Slice X of first storage area X101
111 corresponds to slice X132. Slice X1
32 is the n-th slice in the second storage area X121, the slice X111 in the first storage area X101
Is the n-th slice in the first storage area X101. Here, n is an integer of 1 or more.

The decoding control section 110 has a storage area flag X161 as shown in FIG. The storage area flag X161 indicates that “the first storage area X101 stores the frame immediately before the currently decoded frame, and the second storage area X121 stores the currently decoded frame”.
Alternatively, to indicate any state of “the first storage area X101 stores the frame currently being decoded, and the second storage area X121 stores the frame immediately before the frame currently being decoded”. , Take a value of “1” or “0” in each state.

The decoding control unit 110 sets “1” to the storage area flag X161 as initialization processing immediately after the image decoding apparatus is started and before the bit stream decomposition processing is started. Further, when starting decoding the first image of the frame, the decoding control unit 110 uses the storage area flag X161 to indicate that the currently decoded frame is to be stored in the first storage area X101 or the second storage area. The image restoration unit 119 is instructed to write a blank in X121.

Further, after the decoding of the last image of the frame is completed, the decoding control unit 110 sets the storage area flag X
161 is inverted. That is, the storage area flag X1
If the value of 61 is “1”, the value of “0” is set to the storage area flag X161, and if the value of the storage area flag X161 is “0”, the value of “1” is set to the storage area flag X161. set.

When an error occurs during decoding, the image restoring unit 119 reads an image for one slice from the slice X111 in the first storage area X101 corresponding to the slice X132 in which the error has occurred, and reads one slice. The read image is stored in the slice X1 in the second storage area X121.
Write to 32. In addition, the image restoration unit 119 instructs the decoding control unit 110 to write a blank in the first storage area X101 or the second storage area X121 indicating that the frame currently being decoded is to be stored by the storage area flag X161. Then, the storage area flag X161 writes a blank in the first storage area X101 or the second storage area X121 indicating that the frame currently being decoded is stored. 3.2 Operation of Image Decoding Apparatus Here, the operation of the image decoding apparatus according to the third embodiment will be described with reference to the flowchart shown in FIG.
The following description focuses on the differences from the flowchart shown in FIG.

In the flowchart shown in FIG.
The flowchart shown in FIG. 19 is obtained by adding 31, S32, S33, and S34. In the flowchart shown in FIG. 19, in step S31, the decoding control unit 110
Immediately after the start of the image decoding apparatus and before starting the bitstream decomposition processing, “1” is set to the storage area flag X161 as initialization processing.

In step S32, as the storage area initialization processing, the decoding control section 110 sets the storage area flag X16
1 instructs the image restoration unit 119 to write a blank in the first storage area X101 or the second storage area X121 indicated to store the frame currently being decoded,
Upon receiving this instruction, the image restoration unit 119 uses the storage area flag X161 to indicate that the currently decoded frame is to be stored in the first storage area X101 or the second storage area X101.
Write a blank in 121.

At step S33, decoding control section 110
Determines whether or not decoding of the last image of the frame has been completed. If it is determined that decoding has been completed, in step S34, the value of the storage area flag X161 is inverted. That is, if the value of the storage area flag X161 is “1”, a value of “0” is set to the storage area flag X161, and if the value of the storage area flag X161 is “0”, “0” is set to the storage area flag X161. Set the value of "1". Next, control is returned to step S32.

FIG. 20 is a flowchart showing the detailed operation of the image compensation processing at the time of error. Image restoration unit 119
When an error occurs during decoding, the first storage area X101 corresponding to the slice X132 in which the error has occurred
The image for one slice is read from the slice X111 in the storage area (step SX401), and the read image for one slice is written in the slice X132 in the second storage area X121 (step SX402). 4. Fourth Embodiment Here, as a fourth embodiment according to the present invention, another detail of the image compensation processing at the time of error of the image restoration unit 119 of the first embodiment will be described.

The following embodiment can be similarly implemented in the image restoring unit 13c according to the second embodiment. 4.1 Configuration of Image Decoding Apparatus Here, the configuration of the image decoding apparatus according to the fourth embodiment of the present invention will be described.

The image decoding apparatus according to the fourth embodiment has the same configuration as that of the image decoding apparatus according to the first embodiment, and therefore only different parts will be described below. As shown in FIG. 21, the image storage unit 120 stores a frame image 1692 being decoded in a third storage area X50.
One. The third storage area X501 includes a plurality of slices X521.

Next, the third storage area X of the image restoration unit 119
The details of writing a frame image to 501 will be described. It is assumed that the decoding of one frame image is completed and the time immediately before the decoding of the subsequent frame image is started.
The image restoration unit 119 reads the decoded frame image from the image storage unit 120, and reads the frame image in the third storage area X5.
Write to 01.

Next, when the decoding of the subsequent frame image is started, the image restoring unit 119 outputs
The second slice is written to the first slice of the third storage area X501. At this time, the image restoration unit 119
No writing is performed on the second and subsequent slices stored in the storage area X501. Further, when the decoding of the second slice of the subsequent frame image is started, the image restoration unit 119 sets the second slice to the third slice.
Write to the second slice of the storage area X501. At this time, the image restoration unit 119 does not perform any writing on the third and subsequent slices stored in the third storage area X501.

Hereinafter, the image restoration unit 119 similarly performs
The third and subsequent slices are written, and the subsequent frame image is decoded in the third storage area X501. While decoding the frame image as described above, n
Suppose an error occurs while decoding the th slice. In FIG. 21, a slice X 511 is a frame image currently being decoded, indicates a slice that has already been decoded, and a slice X 511.
Reference numeral 512 denotes a frame image that is currently being decoded and indicates a slice in which an error has occurred during decoding, and slice X513 indicates a decoded slice of the frame decoded immediately before. In the third storage area X501, the slice X511 includes:
Images up to the (n-1) th slice of the currently decoded frame image have been written, and 1
The image of the n-th slice of the immediately preceding frame image is written, and the image of the (n + 1) -th and subsequent slices of the immediately preceding frame image is written in slice X513.

[0135] In this case, the image restoring unit 119 sets the slice X512 of the image of the n-th slice in which the error has occurred.
Is not written. Next, the image restoration unit 119
When the (n + 1) th slice is decoded, the image of the (n + 1) th slice is written to the (n + 1) th slice of the third storage area X501.

Thereafter, the image restoring unit 119 similarly performs
When the subsequent slice is decoded, the decoded slice is written to the subsequent slice of the third storage area X501. In this way, when an error occurs in the n-th slice of the frame image, the image of the n-th slice of the immediately preceding frame image is used instead of the image of the slice in which the error has occurred. Is done. 4.2 Operation of Image Decoding Device Here, the operation of the image decoding device according to the fourth embodiment will be described. The general operation of the image decoding apparatus is the same as the operation shown in the flowchart of FIG.

The processing operation of the image copying section will be described with reference to the flowchart shown in FIG. 22, focusing on differences from the flowchart shown in FIG. Step S204 was deleted from the flowchart shown in FIG. 5, and a flowchart shown in FIG. 22 was obtained. That is, the image restoration unit 119
While restoring the image of slice X512, slice X5
In the case where an error has occurred in No. 12, writing to the slice X512 of the third storage area X501 is not performed at the time of occurrence of the error.

In this way, the image of the slice corresponding to the immediately preceding frame is adopted as the image of the slice in which an error has occurred.

[0139]

As described above, the present invention relates to an image decoding apparatus which decodes a compressed code sequence to generate a moving image, and comprises a unit image decoded by decomposing the compressed code sequence. The encoded basic unit image includes the same / different information indicating whether or not the non-coded basic unit image is the same as the reference unit image in the already decoded frame image. When the different information indicates that the basic unit image is different from the reference unit image, the basic unit image further includes an encoded differential unit image, and the differential unit image includes the basic unit image and the basic unit image obtained by predictive coding. When the reference unit image of another frame image to be referred to when the reference unit image is different from the reference unit image, the reference unit image is constituted by a difference value between a pixel value forming the base unit image and a pixel value forming the reference unit image. The unit image and the reference unit image are a predetermined number. A decomposing means consisting of pixels, an image decoding means for sequentially decoding the extracted coded difference unit image to generate a difference unit image, and a first constant image consisting of a unit image having a pixel value of 0 as an element are generated. A first constant generation unit that, when the extracted difference information indicates that the basic unit image and the reference unit image are the same, divides the first constant image generated by the first constant generation unit; Selected,
A first image selection unit that selects the difference unit image decoded by the image decoding unit when the extracted difference information indicates that the original unit image is different from the reference unit image; Image restoration means for generating a basic unit image by adding the selection result by the image selection means and the reference unit image.

According to this configuration, the unit image being decoded is P
If the unit image is a part of a picture or a B picture and is a unit image to be skipped, a unit image having a pixel value of 0 as a pixel generated by the first constant generating unit is selected, and like a unit image of an I picture, If the unit image is not skipped, the unit image decoded by the image decoding unit is selected, and the selected unit image and the already decoded reference unit image are added to restore the unit image. In this manner, the process of image decoding a unit image having a pixel value of 0 which is originally unnecessary is omitted, and the unit image is restored. Thus, there is an effect that the restoration time of the unit image can be reduced. Here, the first constant generation unit transmits a pixel value 0 for the number of pixels included in the unit image to a first constant generation unit that generates one pixel value 0 and the first constant generation unit. And a first constant control unit that generates a first constant image having a pixel value of 0 corresponding to the number of pixels included in the unit image as an element.

According to this configuration, the first constant generation unit generates one pixel value 0, and the first constant control unit causes the first constant generation unit to generate the required number of pixel values. Since control is performed, there is an effect that a required number of pixel values can be generated. Here, the first constant generating means generates one pixel value 0 as a first constant image, and the first image selecting means, when the reference unit image and the basic unit image are the same, One pixel value 0 generated by the first constant generation means is selected for the number of pixels included in the unit image, and the image restoration means selects the reference unit image and the original unit image when they are the same. Then, the basic unit image is generated by adding the pixel values 0 of the number of pixels included in the unit image selected by the first image selecting unit and the reference unit image in the already decoded frame image. May be configured.

According to this configuration, the first constant generating means is 1
Since one pixel value 0 is generated and the required number of pixel values are selected by the first image selection means, the configuration of the first constant generation means is simplified. Here, the first
The constant generating means generates a plurality of pixel values 0 as a first constant image, and the first image selecting means generates the first constant value when the reference unit image and the basic unit image are the same. Selecting a plurality of first constant images consisting of a plurality of pixel values 0 generated by the means, and selecting pixel values 0 for the number of pixels included in the unit image in total; When the reference unit image and the original unit image are the same, a pixel value 0 corresponding to the number of pixels included in the unit image selected by the first image selecting unit and a reference value in the already decoded frame image The unit image may be added to generate a basic unit image.

According to this configuration, the first constant generation means generates a plurality of pixel values 0, and the first image selection means selects the required number of pixel values. There is an effect that can be shortened. Here, the image decoding unit sequentially decodes the extracted coded difference unit image to generate a code sequence decoding unit that generates an effective coefficient value and a run length of at least one pair, and a pixel sequence including a pixel value of 0. A second constant generating means for generating two constants, the at least one run length generated by the code string decoding means, the second constant for the obtained run length is selected, and the generated at least A second selecting means for selecting one effective coefficient value and combining a second constant for the selected run length and the effective coefficient to generate a coefficient matrix; and Inverse quantization means for performing an inverse quantization process on a coefficient matrix to generate an orthogonal transform coefficient, and an inverse orthogonal transform process for performing an inverse transform process of an orthogonal transform on the generated orthogonal transform coefficient to generate the difference unit image Means to include It may form.

According to this configuration, the image decoding means includes a code string decoding means, a second constant generation means, a second selection means,
Since it has an inverse quantization means and an inverse orthogonal transform means, MP
There is an effect that an image encoded by a general encoding method such as EG can be decoded. Here, the inverse transform process of the orthogonal transform may be configured to be an inverse discrete cosine transform.

According to this configuration, since the inverse discrete cosine transform is performed, there is an effect that an image encoded by a general encoding method such as MPEG can be decoded. here,
The image restoring unit includes: an image storage unit that stores one or a plurality of reference frame images and a first frame image to be restored from the reference frame image; and reads and reads a reference unit image from the reference frame images stored in the image storage unit. The reference unit image and the first constant image or the difference unit image selected by the first image selection unit are added to generate a basic unit image, and the generated basic unit image is stored in the image storage unit. And a sub-image restoring means for writing as a first frame image.

According to this configuration, since the image restoring unit generates the basic unit image by adding the reference unit image and the difference unit image, there is an effect that the image that has been predictively encoded can be restored. Here, the image decoding unit may further include, for each of the original unit images, when the difference unit image is generated from the encoded difference unit image, the encoded difference unit image includes a non-specified value. First to detect an error indicating
Including an error detection means, the image decoding apparatus further comprises:
A first error control unit configured to control the image decoding unit to stop decoding the coded difference unit image in which the error has occurred when the first error detection unit detects an error; The means, further, when an error is detected, instead of the coded difference unit image in which the error has occurred, using the reference frame image stored in the image storage means, to generate an alternative unit image, An error image restoration unit that writes the generated alternative unit image into the image storage unit as a first frame image may be included, and the image restoration unit may further generate an original unit image. In this case, the apparatus further includes a second error detection unit that detects an error that occurs with the motion compensation processing, and the first error control unit further includes a second error detection unit that detects the error. If the may be configured to control so as to stop the decoding of the generated coded difference unit image error on the image decoding means.

According to this configuration, when an error occurs during decoding of the unit image, or by adding the selected unit image and the reference unit image read from the image storage unit, the unit image is restored. If an error occurs,
The decoding of the unit image in which the error occurred is stopped, the unit image is restored using the unit image stored in the image storage unit instead of the unit image in which the error occurred, and the next unit image is , The extraction of the different information corresponding to (1) is started, so that the restoration time of the unit image can be reduced.

For example, when a motion vector error indicating that the motion vector points out of the area of the reference frame image occurs, a compensation process for restoring the restored unit image based on the reference frame image stored in the image storage means is executed. At the same time, the slice including the unit image being decoded is skipped, and the detection of the next slice is started. This parallel processing can reduce the processing time.

That is, when an error occurs in a unit image and error compensation processing is performed, the time from “start of error compensation” to “completion of detection processing of the next slice” is conventionally set to “error compensation”. The processing time '' is the time obtained by adding the `` next slice detection time ''. However, in the image decoding apparatus according to the present invention, the `` error compensation processing time '' or the `` next slice detection time '' is the later time. Become. As described above, the processing time when an error occurs can be reduced. here,
The image restoring unit further reads the reference frame image stored in the image storage unit before starting restoration of the frame image, and stores the read reference frame image in the image storage unit as a first frame image. The sub-image restoring unit writes the generated basic unit image when the first error detecting unit and the second error detecting unit detect no error. And a unit image writing unit for writing as a first frame image, wherein the image restoration unit further includes an error image restoration unit for detecting the error image when an error is detected by the first error detection unit or the second error detection unit. Means for controlling the writing of the alternative unit image into the image storage means. Good.

According to this configuration, before the restoration of the frame image is started, the reference frame image is read, and the read reference frame image is written in the image storage unit as the frame image to be restored from now on. When an error occurs, the decoded unit image is written, and when an error occurs, the writing of an alternative unit image in place of the unit image in which the error occurred is suppressed, so that the error occurs without actively performing the error compensation processing. The unit image can be restored using the reference frame image instead of the unit image. Here, the error image restoring means, when an error is detected by the first error detecting means or the second error detecting means, from the image storage means, the encoded difference unit image in which the error is detected. Image reading means for reading a unit image at the same position in the reference frame image, and an alternative image writing unit for writing the read unit image as an alternative unit image and writing the alternative unit image as a first frame image in the image storage means. Means may be included.

According to this configuration, when an error occurs, a unit image can be restored using the frame image restored immediately before. Here, the image restoring unit further includes a second error detecting unit that detects an error that occurs along with the motion compensation processing when the basic unit image is generated, and an error detecting unit that detects an error when the error is detected. Instead of the generated coded difference unit image, an alternative unit image is generated using the reference frame image stored in the image storage unit, and the generated alternative unit image is stored in the image storage unit as a first frame. Error decoding means for writing as an image, the image decoding apparatus further comprising: an encoded difference unit image in which an error has occurred in the image decoding means when the second error detection means has detected an error. May be configured to include first error control means for controlling so as to stop decoding of.

According to this configuration, when an error occurs when restoring the unit image by adding the selected unit image and the reference unit image read from the image storage means,
The decoding of the unit image in which the error occurred is stopped, the unit image is restored using the unit image stored in the image storage unit instead of the unit image in which the error occurred, and the next unit image is , The extraction of the different information corresponding to (1) is started, so that the restoration time of the unit image can be reduced. Here, the image restoring unit further reads the reference frame image stored in the image storage unit before starting restoration of the frame image, and stores the read reference frame image in the image storage unit. Including frame image copying means for writing as one frame image,
The sub-image restoration unit includes a unit image writing unit that writes the generated unit image as a first frame image to the image storage unit when no error is detected by the second error detection unit, The image restoration means
Further, the image processing apparatus may further include an image writing suppression unit that suppresses writing of the alternative unit image to the image storage unit with respect to the error image restoration unit when an error is detected by the second error detection unit. May be configured.

According to this configuration, before the restoration of the frame image is started, the reference frame image is read, and the read reference frame image is written into the image storage unit as the frame image to be restored from now on. When an error occurs, the decoded unit image is written, and when an error occurs, the writing of an alternative unit image in place of the unit image in which the error occurred is suppressed, so that the error occurs without actively performing the error compensation processing. The unit image can be restored using the reference frame image instead of the unit image. Here, the error image restoration means, when an error is detected by the second error detection means, from the image storage means, the coded difference unit image in which the error is detected and the reference frame image. Image reading means for reading a unit image at the same position, and alternative image writing means for writing the read unit image as an alternative unit image and writing the alternative unit image as a first frame image in the image storage means. May be.

According to this configuration, when an error occurs, a unit image can be restored using the frame image restored just before. Also, the present invention is an image decoding apparatus for decoding a compressed code sequence to generate a moving image, and extracts a predetermined number of basic unit images encoded by decomposing the compressed code sequence, and The predetermined number of the basic unit images is the same as the reference unit image in the frame image in which any uncoded basic unit image is already decoded among the predetermined number of the basic unit images. The pattern includes a pattern identifier for identifying a different pattern indicating whether the basic unit image that is not different from the reference unit image, and the encoded predetermined number of basic unit images includes the pattern unit of the predetermined number of basic unit images. For the basic unit image indicated to be different from the reference unit image by the same or different pattern identified by the identifier, the basic unit image further includes an encoded differential unit image, and the differential unit image includes the basic unit image and the basic unit image. But When the reference unit image of another frame image referred to when being encoded is different, it is configured by a difference value between a pixel value forming the original unit image and a pixel value forming the reference unit image, The original unit image and the reference unit image are decomposed by a predetermined number of pixels, an image decoding unit that sequentially decodes the extracted coded difference unit image to generate a difference unit image, A first constant generating means for generating a first constant image composed of unit images as elements, a pattern table storing same and different patterns, and a pattern identifier for identifying each different pattern; Determining a similar pattern to be indicated from the pattern table, and for each of a predetermined number of basic unit images, determining means for determining whether an uncoded basic unit image is the same as the reference unit image. The first constant image generated by the first constant generating means when it is determined that the unencoded basic unit image and the reference unit image are the same for each of the predetermined number of basic unit images. And selecting means for selecting the difference unit image decoded by the image decoding means when it is determined that the uncoded original unit image and the reference unit image are different from each other. Means and an image restoring means for generating a basic unit image by adding a selection result by the first image selecting means and the reference unit image.

According to this configuration, it is determined whether or not the basic unit image and the reference unit image are the same using the pattern identifier included in the same / different information. The processing of image decoding a unit image having a pixel value of 0 as an element is omitted, and the unit image is restored.
Thus, there is an effect that the restoration time of the unit image can be reduced. Here, the first constant generation unit further generates a third constant image including pixel values 0 corresponding to the number of pixels included in the predetermined number of basic unit images, and the determination unit further includes the extracted constant image. For each of the predetermined number of basic unit images, all of the predetermined number of basic unit images are the same as the reference unit image, or at least one of the predetermined number of basic unit images is the reference unit image. It is determined whether the reference unit image is different from the image. If the determination unit determines that all of the predetermined number of basic unit images are the same as the reference unit image, the first constant generation unit determines The generated third constant image is selected, and the image decoding unit sequentially decodes the extracted coded difference unit image to generate a code sequence decoding that generates an effective coefficient value and a run length of at least one pair. Means and pixel value 0 A second constant generating unit configured to generate a second constant and a fourth constant including pixel values 0 corresponding to the number of pixels included in the basic unit image, wherein at least one of the predetermined number of basic unit images is the reference unit image; When it is determined that the basic unit images are different from each other, for each basic unit image of the predetermined number of basic unit images,
The fourth constant generated by the constant generating means is selected to generate a coefficient matrix, and the basic unit image and the reference unit image are different for each basic unit image of the predetermined number of basic unit images. For the basic unit image, the at least one run length generated by the code string decoding unit is obtained, the second constant corresponding to the obtained run length is selected, and the generated at least one effective coefficient value is obtained. A second selecting means for generating a coefficient matrix by combining and selecting a second constant for the selected run length and the effective coefficient value, and an inverse of the coefficient matrix generated by the second selecting means. Inverse quantization means for performing a quantization process to generate an orthogonal transform coefficient, and inverse orthogonal transform means for performing an inverse transform process of the orthogonal transform on the generated orthogonal transform coefficient to generate the difference unit image. May be configured .

According to this configuration, the unit image being decoded is P
When at least one block included in the macroblock includes a pixel in a case of a unit image to be skipped when the block is a part of a picture or a B picture, and for a block that does not include a pixel, the second selection unit performs When the fourth constant generated by the two constant generating means is selected, and when all the blocks included in the macro block are the blocks to be skipped, the first image selecting means sets the third constant generated by the first constant generating means to the third constant. Since the selection is made, in the case of decoding a macroblock in which all blocks are blocks to be skipped and a macroblock including at least one non-skipped block, the restoration time of a unit image is reduced as compared with the related art. Can be. Further, the present invention is an image decoding apparatus which decodes the compressed code sequence to generate a moving image, and extracts a coded basic unit image by decomposing the compressed code sequence, and The basic unit image is
The same unit information includes whether the original unit image that has not been encoded and the reference unit image in the already decoded frame image are the same, and the same information indicates that the original unit image and the reference unit image are different from each other. When the difference unit image is different, the difference unit image further includes an encoded difference unit image, and the difference unit image is a reference unit of another frame image referred to when the original unit image is predictively encoded. If the image is different,
The basic unit image and the reference unit image are configured by a difference value between a pixel value configuring the basic unit image and a pixel value configuring the reference unit image.The basic unit image and the reference unit image are extracted by a decomposing unit including a predetermined number of pixels. Code sequence decoding means for sequentially decoding the coded difference unit images and generating at least one pair of effective coefficient values and run lengths; and second constant generation means for generating a second constant consisting of pixel values of 0. Acquiring the at least one run length generated by the code string decoding means, selecting the second constant for the obtained run length, selecting and selecting the generated at least one effective coefficient value. A first processing unit including a second selection unit configured to generate a coefficient matrix by combining a second constant corresponding to the run length and the effective coefficient;
First storage means for storing the generated coefficient matrix, and inverse quantization for reading the coefficient matrix from the first storage means, performing an inverse quantization process on the read coefficient matrix, and generating an orthogonal transform coefficient. Means for performing inverse transform processing of orthogonal transform on the generated orthogonal transform coefficients to generate an inverse orthogonal transform means, and a second processing means for storing the generated differential unit image. 2 storage means and pixel value 0
A first constant generating means for generating a first constant image composed of unit images each having an element, wherein the extracted similarity information indicates that the reference unit image and the basic unit image are the same. Selecting the first constant image generated by the first constant generating means, and selecting the second constant image if the extracted similarity information indicates that the reference unit image is different from the basic unit image. A first image selection unit that reads the difference unit image from the storage unit and selects the read difference unit image, an image storage unit that stores one or a plurality of reference frame images and a first frame image to be restored therefrom, A reference unit image is read from the reference frame image stored in the image storage unit, and the read reference unit image and the first unit image selected by the first image selection unit are read out.
A third processing unit including: adding a constant image or the difference unit image to generate a basic unit image, and writing the generated basic unit image to the image storage unit as a first frame image; The first processing means, the second processing means, and the third processing means may be configured to include sequential control means for performing pipeline control.

According to this configuration, the decomposing means, the code string decoding means, the second constant generating means and the second selecting means are collectively referred to as the first selecting means.
Processing means, inverse quantization means and inverse discrete cosine transform means as second processing means, first constant generation means, first image selection means, image storage means, and image restoration means as third processing means; A first storage means is provided between the first processing means and the second processing means, and a second storage means is provided between the second processing means and the third processing means, and the first processing means, the second processing means, and the third processing means are provided. Means and the first processing means, the second processing means, and the third processing means can be operated in parallel for different unit images, and the processing time can be reduced. Becomes Here, the first processing unit may further include, for each of the original unit images, when the difference unit image is generated from the encoded difference unit image, the encoded difference unit image includes a value that is not specified. First error detecting means for detecting an error indicating that the error is detected, and when the first error detecting means detects an error, the decoding of the coded difference unit image in which the error has occurred is stopped for the code string decoding means. First error control means for controlling the image processing means, the third processing means further comprising: when an error is detected, stored in the image storage means instead of the coded difference unit image in which the error has occurred. Using the reference frame image
It may be configured to include an error image restoration unit that generates an alternative unit image, and writes the generated alternative unit image as a first frame image in the image storage unit.
The third processing unit further includes a second error detection unit that detects an error that occurs along with the motion compensation process when the basic unit image is generated, and the first error control unit further includes: When the second error detection unit detects an error, the image decoding unit may be configured to control the image decoding unit to stop decoding the coded difference unit image in which the error has occurred.

According to this configuration, when an error occurs during decoding of the unit image, or by adding the selected unit image and the reference unit image read from the image storage unit, the unit image is restored. If an error occurs,
The decoding of the unit image in which the error occurred is stopped, the unit image is restored using the unit image stored in the image storage unit instead of the unit image in which the error occurred, and the next unit image is , The extraction of the different information corresponding to (1) is started, so that the restoration time of the unit image can be reduced.

For example, if a motion vector error occurs in which the motion vector points out of the area of the reference frame image, a compensation process for restoring a restoration unit image based on the reference frame image stored in the image storage means is executed. At the same time, the slice including the unit image being decoded is skipped, and the detection of the next slice is started. This parallel processing can reduce the processing time.

That is, when an error occurs in a unit image and error compensation processing is performed, the time from “start of error compensation” to “completion of detection processing of the next slice” is conventionally set to “error compensation”. The processing time '' is the time obtained by adding the `` next slice detection time ''. However, in the image decoding apparatus according to the present invention, the `` error compensation processing time '' or the `` next slice detection time '' is the later time. Become. As described above, the processing time when an error occurs can be reduced. here,
The third processing unit further reads the reference frame image stored in the image storage unit before starting restoration of the frame image, and stores the read reference frame image in the image storage unit in the first frame. A frame image copying unit for writing as an image, wherein the image restoration unit stores the generated basic unit image when the first error detection unit and the second error detection unit detect no error; And a unit image writing unit for writing as a first frame image, wherein the image restoration unit further includes an error image restoration unit for detecting the error image when an error is detected by the first error detection unit or the second error detection unit. The means may include an image writing suppressing means for suppressing writing of the alternative unit image to the image storage means. There.

According to this configuration, before the restoration of the frame image is started, the reference frame image is read, and the read reference frame image is written in the image storage unit as the frame image to be restored from now on. When an error occurs, the decoded unit image is written, and when an error occurs, the writing of an alternative unit image in place of the unit image in which the error occurred is suppressed, so that the error occurs without actively performing the error compensation processing. The unit image can be restored using the reference frame image instead of the unit image. Here, the error image restoring means, when an error is detected by the first error detecting means or the second error detecting means, from the image storage means, the encoded difference unit image in which the error is detected. Image reading means for reading a unit image at the same position in the reference frame image, and an alternative image writing unit for writing the read unit image as an alternative unit image and writing the alternative unit image as a first frame image in the image storage means. Means may be included.

According to this configuration, when an error occurs, a unit image can be restored using the frame image restored immediately before. As described above, in the image decoding apparatus according to the present invention, when a block to be skipped or a block including an error occurs, the processing can be speeded up, and the practical effect is large.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of an image decoding device according to a first embodiment of the present invention.

FIG. 2 shows a data flow of the image decoding apparatus shown in FIG.

FIG. 3 shows a data flow of the image decoding apparatus shown in FIG. FIG. 3 is a continuation drawing of FIG. 2.

FIG. 4 is a flowchart illustrating an operation of the image decoding device illustrated in FIG. 1;

FIG. 5 is a flowchart showing an operation of the image decoding apparatus shown in FIG. 1 when an error occurs.

FIG. 6 is a time chart illustrating a transition of a process of the image decoding apparatus illustrated in FIG. 1 over time.

FIG. 7 is a block diagram illustrating a schematic configuration of an image decoding device according to a second embodiment of the present invention.

FIG. 8 shows a data flow of the image decoding apparatus shown in FIG. 7;

FIG. 9 shows a data flow of the image decoding apparatus shown in FIG. 7; FIG. 9 is a continuation drawing of FIG. 8.

FIG. 10 shows a data flow of the image decoding apparatus shown in FIG. 7; 10 is a continuation of the drawing from FIG. 9.

FIG. 11 shows a data flow of the image decoding apparatus shown in FIG. 7; 11 is a continuation of the drawing from FIG. 10.

12 is a diagram illustrating a state transition of a code decoding process, an image conversion process, and an image restoration process of the image decoding device illustrated in FIG. 7;

FIG. 13 is a flowchart illustrating an operation of a sequential control unit of the image decoding device illustrated in FIG. 7;

FIG. 14 is a flowchart illustrating an operation of a sequential control unit of the image decoding device illustrated in FIG. 7; FIG. 14 shows a continuation from the flowchart in FIG. 13.

FIG. 15 is a flowchart illustrating an operation of a sequential control unit of the image decoding device illustrated in FIG. 7; FIG. 15 shows a continuation from the flowchart in FIG. 14.

FIG. 16 is a time chart illustrating transition of a process with time of the image decoding device illustrated in FIG. 7;

FIG. 17 is a diagram illustrating a configuration of an image storage unit of an image decoding device according to a third embodiment of the present invention.

FIG. 18 illustrates a storage area flag included in a decoding control unit of the image decoding device according to the third embodiment of the present invention.

FIG. 19 is a flowchart illustrating an operation of the image decoding device according to the third embodiment of the present invention.

FIG. 20 is a flowchart illustrating an image compensation processing operation when an error occurs in the image decoding device according to the third embodiment of the present invention.

FIG. 21 is a diagram illustrating a configuration of an image storage unit of an image decoding device according to a fourth embodiment of the present invention.

FIG. 22 is a flowchart illustrating a processing operation performed when an error occurs in the image decoding device according to the fourth embodiment of the present invention.

FIG. 23 is a diagram for explaining the principle of predictive coding.

FIG. 24 is a diagram illustrating a hierarchical structure of a compressed code string obtained by compressing a moving image.

FIG. 25 is a diagram illustrating a pattern example of a block pattern;

FIG. 26 is a time chart showing a transition of processing with time of a conventional image decoding apparatus.

[Explanation of symbols]

 Reference Signs List 10 control unit 10a information storage unit 11 decoding control unit 11a bit stream decomposition unit 11b decoding unit 11c second constant generation unit 11d second selection unit 12 conversion control unit 12a inverse quantization unit 12b inverse discrete cosine transform unit 13 restoration control unit 13a First constant generation unit 13b First selection unit 13c Image restoration unit 14 First image storage unit 15 Second image storage unit 16 Third image storage unit 110 Decoding control unit 111 Bit stream decomposition unit 112 Decoding unit 113 Second constant generation unit 114 second selection section 115 inverse quantization section 116 inverse discrete cosine transform section 117 first constant generation section 118 first selection section 119 image restoration section 120 image storage section

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Shuzo Kiyohara 1006 Kadoma Kadoma, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (21)

[Claims] An image decoding apparatus that decodes a compressed code sequence to generate a moving image, extracts a coded unit image by decomposing the compressed code sequence, and extracts the coded unit image. The image includes homogeneity information indicating whether the unencoded basic unit image and the reference unit image in the already decoded frame image are the same, and the homogeneity information includes the basic unit image and the reference unit. When the image indicates that the image is different, the image further includes an encoded difference unit image, wherein the difference unit image includes the basic unit image and another frame image referred to when the basic unit image is predictively encoded. When the reference unit image is different from the reference unit image, the reference unit image is constituted by a difference value between a pixel value forming the base unit image and a pixel value forming the reference unit image. A decomposition means consisting of pixels of Image decoding means for sequentially decoding the coded difference unit images to generate a difference unit image; first constant generation means for generating a first constant image composed of unit images having a pixel value of 0 as an element; If the obtained difference information indicates that the basic unit image and the reference unit image are the same, the first constant image generated by the first constant generation unit is selected, and the extracted difference image is selected. When the information indicates that the original unit image and the reference unit image are different, a first image selecting unit that selects the difference unit image decoded by the image decoding unit, and a selection result by the first image selecting unit And an image restoring unit that generates a basic unit image by adding the reference unit image and the reference unit image. 2. The first constant generation unit includes: a first constant generation unit that generates one pixel value 0; and a pixel value for the number of pixels included in the unit image for the first constant generation unit. 2. A first constant control unit that controls to generate 0 and generates a first constant image having pixel values 0 as elements of the number of pixels included in the unit image. Image decoding device. 3. The first constant generation unit generates one pixel value 0 as a first constant image, and the first image selection unit determines that the reference unit image and the basic unit image are the same. In addition, one pixel value 0 generated by the first constant generation unit is selected for the number of pixels included in the unit image, and the image restoration unit determines that the reference unit image and the original unit image are the same. In some cases, the pixel value of the number of pixels included in the unit image selected by the first image selecting means is 0.
2. The image decoding apparatus according to claim 1, wherein a basic unit image is generated by adding the reference unit image in the already decoded frame image to the reference unit image. 4. The first constant generation unit generates a plurality of pixel values 0 as a first constant image, and the first image selection unit sets the reference unit image and the basic unit image to be the same. In this case, a plurality of first constant images composed of a plurality of pixel values 0 generated by the first constant generating means are selected, and pixel values 0 for the number of pixels included in the unit image are selected in total. The image restoration unit, when the reference unit image and the original unit image are the same, has a pixel value 0 for the number of pixels included in the unit image selected by the first image selection unit.
2. The image decoding apparatus according to claim 1, wherein a basic unit image is generated by adding the reference unit image in the already decoded frame image to the reference unit image. 5. An image decoding unit, comprising: a decoding unit configured to sequentially decode the extracted coded difference unit images to generate an effective coefficient value and a run length of at least one pair; A second constant generation means for generating a second constant, and at least one of the at least one generated by the code string decoding means
Two run lengths, selecting the second constant for the obtained run length, selecting the at least one effective coefficient value generated, and selecting the second constant and the effective coefficient for the selected run length. A second selecting means for generating a coefficient matrix by combining the above, an inverse quantization means for performing an inverse quantization process on the coefficient matrix generated by the second selecting means to generate an orthogonal transform coefficient, 2. The image decoding apparatus according to claim 1, further comprising: an inverse orthogonal transform unit that performs an inverse transform process of the orthogonal transform on the generated orthogonal transform coefficient to generate the difference unit image. 6. The image decoding apparatus according to claim 5, wherein the inverse transform of the orthogonal transform is an inverse discrete cosine transform. 7. The image restoring means includes: one or more reference frame images and a first
An image storage unit that stores a frame image; a reference unit image read from a reference frame image stored in the image storage unit; and the read reference unit image and the first unit selected by the first image selection unit. A sub-image restoring unit that generates a basic unit image by adding a constant image or the difference unit image, and writes the generated basic unit image as the first frame image in the image storage unit. The image decoding device according to claim 6. 8. The image decoding unit further comprises: for each of the original unit images, when the difference unit image is generated from the encoded difference unit image, the encoded difference unit image outputs a non-specified value. The image decoding apparatus further includes a first error detection unit configured to detect an error indicating that the image decoding unit detects an error. A first error control unit for controlling to stop decoding of the coded difference unit image, wherein the image restoring unit further includes, when an error is detected, instead of the coded difference unit image in which the error has occurred; Generating an alternative unit image using the reference frame image stored in the image storage means, and writing the generated alternative unit image as a first frame image in the image storage means. The image decoding apparatus according to claim 7, characterized in that it comprises an image restoring means. 9. The image restoration unit further includes a second error detection unit that detects an error that occurs with a motion compensation process when a basic unit image is generated, wherein the first error control unit includes: 10. The image processing apparatus according to claim 8, further comprising: when the second error detecting unit detects an error, controlling the image decoding unit to stop decoding the coded difference unit image in which the error has occurred. An image decoding device according to claim 1. 10. The image restoring unit further reads the reference frame image stored in the image storage unit before starting restoration of a frame image, and stores the read reference frame image in the image storage unit. A frame image copying unit that writes the first unit image as a first frame image, wherein the sub-image restoration unit is configured to generate the basic unit image when an error is not detected by the first error detection unit and the second error detection unit. As a first frame image in the image storage means, wherein the image restoration means further comprises: when an error is detected by the first error detection means or the second error detection means, An image writing suppressing unit that suppresses writing of the alternative unit image to the image storage unit with respect to the error image restoring unit. The image decoding apparatus according to claim 9, wherein Mukoto. 11. The error image restoring means, when an error is detected by the first error detecting means or the second error detecting means, an encoded difference in which the error is detected from the image storage means. An image reading unit that reads a unit image at the same position in the unit image and the reference frame image, an alternative image that writes the read unit image as an alternative unit image, and writes the alternative unit image into the image storage unit as a first frame image 10. The image decoding apparatus according to claim 9, further comprising a writing unit. 12. The image restoration means further comprises: a second error detection means for detecting an error occurring along with a motion compensation process when a basic unit image is generated; Instead of the coded difference unit image in which the error has occurred, an alternative unit image is generated using the reference frame image stored in the image storage unit, and the generated alternative unit image is stored in the image storage unit. An error image restoration unit that writes as one frame image, wherein the image decoding device further comprises: an encoding difference in which an error has occurred in the image decoding unit when the second error detection unit detects an error. The image decoding apparatus according to claim 7, further comprising a first error control unit that controls to stop decoding the unit image. 13. The image restoring unit further reads the reference frame image stored in the image storage unit before starting restoration of a frame image, and stores the read reference frame image in the image storage unit. A sub-image restoring unit that writes the generated basic unit image to the image storage unit when no error is detected by the second error detection unit. A unit image writing unit for writing as a one-frame image, wherein the image restoring unit further supplies the substitute unit image to the error image restoring unit when an error is detected by the second error detecting unit. 13. The image decoding apparatus according to claim 12, further comprising an image writing suppressing unit that suppresses writing to said image storage unit. 14. The error image restoring means, when an error is detected by the second error detecting means, the coded difference unit image in which the error is detected and the reference frame image are read from the image storage means. Image reading means for reading a unit image at the same position in the image, and alternative image writing means for writing the read unit image as an alternative unit image and writing the alternative unit image as a first frame image in the image storage means. 13. The image decoding apparatus according to claim 12, wherein: 15. An image decoding apparatus for decoding a compressed code sequence to generate a moving image, wherein the image decoding device decomposes the compressed code sequence to extract a predetermined number of coded basic unit images, and The predetermined number of basic unit images are the same as the reference unit images in the already decoded frame images, and among the predetermined number of basic unit images, which non-coded basic unit images are the same, The reference unit image includes a pattern identifier for identifying a different pattern indicating whether or not the reference unit image is different from the reference unit image, and the encoded predetermined number of unit image is the pattern identifier of the predetermined number of unit images. For the original unit image that is indicated to be different from the reference unit image by the same or different pattern identified by, further includes an encoded difference unit image, the difference unit image is the original unit image, and the original unit image is Forecast When the reference unit image of another frame image referred to when being encoded is different, it is configured by a difference value between a pixel value forming the original unit image and a pixel value forming the reference unit image, The original unit image and the reference unit image are decomposed by a predetermined number of pixels; an image decoding unit that sequentially decodes the extracted encoded difference unit image to generate a difference unit image; A first constant generating means for generating a first constant image composed of unit images as elements, a pattern table for storing the same and different patterns, and a pattern identifier for identifying each of the different patterns; Determining a similar pattern to be indicated from the pattern table, and for each of a predetermined number of basic unit images, determining means for determining whether an uncoded basic unit image is the same as the reference unit image. The first constant generated by the first constant generator when it is determined that the uncoded basic unit image and the reference unit image are the same for each of the predetermined number of basic unit images. Selecting an image and, when it is determined that the unencoded original unit image and the reference unit image are different, selecting means for selecting the difference unit image decoded by the image decoding means. An image decoding apparatus comprising: a selection unit; and an image restoration unit that generates a basic unit image by adding a selection result by the first image selection unit and the reference unit image. 16. The first constant generation unit further generates a third constant image including pixel values 0 for the number of pixels included in the predetermined number of basic unit images, and the determination unit further includes: Using the extracted pattern identifier, for each of the predetermined number of basic unit images, all of the predetermined number of basic unit images are the same as the reference unit image, or at least one of the predetermined number of basic unit images is the The selection unit determines whether the reference unit image is different from the reference unit image. If the determination unit determines that all of the predetermined number of original unit images are the same as the reference unit image, the selection unit generates the first constant. Means for selecting the third constant image generated by the means, the image decoding means for sequentially decoding the extracted coded difference unit images, and generating at least one pair of effective coefficient values and run lengths Column decoding means and pixels A second constant generating unit configured to generate a second constant including zero and a fourth constant including pixel values 0 corresponding to the number of pixels included in the basic unit image; and at least one of the predetermined number of basic unit images refers to the reference. When it is determined that the unit image is different from the unit image, for each unit image of the predetermined number of unit images,
For the basic unit image in which the basic unit image and the reference unit image are the same, the fourth constant generated by the second constant generating unit is selected to generate a coefficient matrix, and the predetermined number of basic units is calculated. For each basic unit image of the unit image,
For the basic unit image in which the basic unit image and the reference unit image are different, the at least one run length generated by the code string decoding unit is obtained, and the second constant corresponding to the obtained run length is selected. The at least one generated
Select two effective coefficient values and select the second for the selected run length
A second selecting unit that generates a coefficient matrix by combining a constant and the effective coefficient value; and performs an inverse quantization process on the coefficient matrix generated by the second selecting unit to generate an orthogonal transform coefficient. The image decoding method according to claim 15, further comprising: an inverse quantization unit; and an inverse orthogonal transformation unit that performs an inverse transformation process of an orthogonal transformation on the generated orthogonal transformation coefficient to generate the difference unit image. apparatus. 17. An image decoding apparatus for decoding a compressed code sequence to generate a moving image, comprising: decomposing a compressed code sequence to extract an encoded basic unit image; The unit image includes homogeneity information indicating whether or not the unencoded basic unit image and the reference unit image in the already decoded frame image are the same, and the homogeneity information includes the basic unit image and the reference. When the unit image is different, the image further includes an encoded difference unit image, and the difference unit image includes the basic unit image and another frame referred to when the basic unit image is predictively encoded. When the reference unit image of the image is different from the reference unit image, the reference unit image is configured by a difference value between a pixel value forming the base unit image and a pixel value forming the reference unit image. Decomposition means consisting of a number of pixels; Code string decoding means for sequentially decoding the output coded difference unit images and generating an effective coefficient value and a run length of at least one pair, and generating a second constant value for generating a second constant value having a pixel value of 0 Means for obtaining the at least one run length generated by the code string decoding means, selecting the second constant for the obtained run length, selecting the generated at least one effective coefficient value, First processing means including second selecting means for generating a coefficient matrix by combining a second constant for the selected run length and the effective coefficient; first storage for storing the generated coefficient matrix Means for reading out the coefficient matrix from the first storage means, performing an inverse quantization process on the read out coefficient matrix to generate an orthogonal transform coefficient, and orthogonally transforming the generated orthogonal transform coefficient. Reverse conversion process A second processing unit including an inverse orthogonal transform unit for generating a difference unit image; a second storage unit for storing the generated difference unit image; and a first unit including a unit image having a pixel value of 0 as an element. First constant generation means for generating a constant image; and the first constant generation means when the extracted difference information indicates that the reference unit image and the basic unit image are the same. Selecting the first constant image, and when the extracted difference information indicates that the reference unit image and the original unit image are different, reads the difference unit image from the second storage unit; A first image selecting unit for selecting the read difference unit image, an image storing unit for storing one or a plurality of reference frame images and a first frame image to be restored therefrom, and a reference stored in the image storing unit H Reading a reference unit image from the frame image, adding the read reference unit image and the first constant image or the difference unit image selected by the first image selecting unit, and generating a basic unit image; A third processing unit including an image restoring unit that writes the obtained basic unit image as a first frame image into the image storage unit; and a pipeline that connects the first processing unit, the second processing unit, and the third processing unit. An image decoding device, comprising: a sequential control unit for controlling the image decoding. 18. The method according to claim 18, wherein, when the difference unit image is generated from the coded difference unit image for each of the original unit images, the coded difference unit image has an undefined value. A first error detection unit that detects an error indicating that the first error detection unit detects an error, and decodes the coded difference unit image in which the error has occurred to the code string decoding unit when the first error detection unit detects an error. First error control means for controlling to stop the processing, wherein the third processing means further comprises, when an error is detected, replacing the coded difference unit image in which the error has occurred with the image storage means. An error image restoration unit that generates an alternative unit image using the stored reference frame image, and writes the generated alternative unit image to the image storage unit as a first frame image. The image decoding apparatus according to claim 17, symptoms. 19. The third processing unit further includes a second error detection unit that detects an error that occurs with a motion compensation process when a basic unit image is generated, and the first error control unit. And controlling the image decoding unit to stop decoding the coded difference unit image in which the error has occurred, when the second error detection unit detects an error. 19. The image decoding device according to 18. 20. The third processing means further reads the reference frame image stored in the image storage means before starting restoration of the frame image, and stores the read reference frame image in the image storage. Means for writing a frame image as a first frame image in the means, wherein the image restoring means generates the basic unit image generated when no error is detected by the first error detecting means and the second error detecting means. As a first frame image in the image storage means, wherein the image restoration means further comprises: when an error is detected by the first error detection means or the second error detection means, And an image writing suppressing unit for suppressing the writing of the alternative unit image into the image storage unit for the error image restoring unit. The image decoding apparatus according to claim 19, wherein a. 21. The image processing apparatus according to claim 19, wherein, when an error is detected by the first error detecting means or the second error detecting means, the coded difference in which the error is detected is read from the image storing means. An image reading unit that reads a unit image at the same position in the unit image and the reference frame image, an alternative image that writes the read unit image as an alternative unit image, and writes the alternative unit image into the image storage unit as a first frame image 20. The image decoding apparatus according to claim 19, further comprising a writing unit.