JP4774315B2 - Image decoding apparatus and image decoding method - Google Patents

Description

  The present invention relates to an image decoding apparatus and an image decoding program for decoding encoded moving image data, and in particular, a moving image compressed and encoded based on the H.264 / AVC (Advanced Video Coding) standard. The present invention relates to an image decoding apparatus and an image decoding method for decoding data.

  MPEG (Moving Picture Experts Group) and other encoding methods have been established as a method for efficiently recording and transmitting moving image data such as TV signals, and the MPEG-1 standard, MPEG-2 standard, MPEG- The international standard encoding system is defined as 4 standards. In addition, the H.264 / AVC standard is defined as a method for further improving the compression rate.

  In general, in moving picture coding, interframe difference coding is performed using the correlation between an intra-frame predictive coding mode in which a frame to be coded is coded only from information in the frame and the preceding and following frames. Two encoding modes, the inter-picture prediction encoding mode, are used adaptively.

  The H.264 / AVC standard has a data structure with the same hierarchy as the MPEG-2 standard, etc., but is a unit for switching between the two encoding modes of the intra prediction encoding mode and the inter prediction encoding mode. Is used as a slice. In the MPEG-2 standard and the like, the unit for switching the encoding mode is a picture (frame), whereas in the H.264 / AVC standard, it is a slice in the next lower layer of this screen. In the following, in order to distinguish a one-frame screen from a “picture” in MPEG-2 or the like, it is referred to as a “screen”.

  As described above, the slice is a unit for switching the predictive coding mode. A slice in which the intra-screen predictive coding mode is set is called an I (Intra coding) slice, and prediction from a past screen on the time axis is performed. A slice in which an inter-frame prediction encoding mode to be encoded using (forward prediction) is set is called a P (Predictive encoding) slice, and prediction from a past screen (forward prediction) or prediction from a future screen ( A slice in which an inter-frame predictive coding mode to be coded using backward prediction is set is called a B (Bidirectionally Predictive Coding) slice. The slices called I, P, and B slices are hereinafter referred to as slice types.

  FIG. 6 is a diagram schematically showing the data structure of a slice of one screen. In the H.264 / AVC standard, one screen is divided into a plurality of slices as shown in the figure. , B slices.

  All of these I, P, and B slices are encoded in units of MB (macroblocks) that are blocks in the lower layer of the slice (that is, for each MB), but selectable predictive encoding in MB. The mode is different for each predictive coding mode set in a slice, that is, for each I, P, and B slice. Specifically, in the I slice MB, only the intra prediction encoding mode can be selected (that is, encoding is performed only in the intra prediction encoding mode). In the MB of the P slice, intra prediction encoding and inter prediction encoding (only forward prediction) can be selected (that is, intra prediction encoding mode and inter prediction encoding mode based on forward prediction depending on the MB). Or you can choose). In MB of B slice, intra prediction encoding and inter prediction encoding (forward prediction and backward prediction) can be selected (that is, inter prediction encoding mode and inter prediction prediction mode according to MB. One of the prediction encoding modes can be selected).

  That is, MBs of P and B slices are encoded by the inter prediction encoding mode by forward prediction and the inter prediction encoding mode by forward prediction and backward prediction. Coding is also performed depending on the mode.

  In the H.264 / AVC standard, different predictive coding modes can be mixed for each slice. However, in the following description, the slice type in one screen is described as one type for the purpose of explicitly expressing the present invention. Therefore, in the following, it is assumed that one screen is composed of one of the I, P, and B slices, that is, slices of the same slice type.

  In moving picture coding, a high compression rate is realized by techniques such as orthogonal transform such as variable length coding and discrete cosine transform (DCT), quantization of transform coefficients, and motion compensation prediction. However, since these employ complex algorithms, a large amount of data processing capacity and memory capacity are required for processing in real time. For example, in video decoding, in order to perform bidirectional prediction from past and future screens by the above motion compensation prediction, at least two reference frame memories and one decoding frame memory are required. Need. This is a frame memory capacity of about 1.6 MByte for a standard resolution (SDTV) video signal and about 9.4 Mbyte for a high resolution (HDTV) video signal.

As one of the techniques for solving the problem that requires a large memory capacity in moving picture decoding as described above, the decoded image data is compressed and stored in the memory, and is read out from the memory as a reference image. A technique for reducing the memory size by expanding this is proposed (for example, see Patent Document 1).
JP 2000-50272 A

  By the way, when decoding the image data encoded according to the H.264 / AVC standard, if the reference image is compressed / decompressed according to Patent Document 1, the screen existing in the inter-screen predictive encoded portion of the screen In a block that has been subjected to intra prediction encoding, image quality degradation is greater than that of a portion that has been inter prediction encoded, and this may be displayed prominently on the screen. Hereinafter, this point will be described.

  In the H.264 / AVC standard, various encoding tools are newly introduced as compared with conventional encoding methods such as the MPEG-2 standard and the MPEG-4 standard. In the intra prediction encoding which is one of them, an already encoded block adjacent to a block which is the target of the intra prediction encoding (hereinafter referred to as an intra prediction encoding target block) is decoded, and this A prediction value is generated using a predetermined pixel in the decoding block, and a difference between the pixel value and the prediction value is obtained for each pixel of the intra prediction encoding target block (hereinafter referred to as an intra prediction encoding pixel). This is a method for encoding. In the intra prediction encoding in the H.264 / AVC standard, there are two parameters, an encoding mode for determining a block size to be predicted encoding and a prediction mode for determining a prediction direction. There are three types of intra coding modes: Intra_4 × 4 mode, Intra_8 × 8 mode, and Intra_16 × 16 mode.

  MB is a block composed of 16 × 16 pixels, and this is divided into 16 blocks of 4 × 4 pixel size, and the encoding mode for intra prediction encoding for each block is Intra_4 × 4 mode. Intra_8 × 8 mode is the mode that divides the block into 4 blocks of 8 × 8 pixel size and performs intra prediction encoding for each block, and MB of 16 × 16 pixel size is subject to intra prediction encoding The mode for intra prediction encoding as a block is the Intra_16 × 16 mode. One of these intra coding modes is selected for each MB.

  The prediction direction refers to a pixel to be subjected to intra prediction encoding from a pixel for which a prediction value in a decoded block adjacent to this block is calculated in the intra prediction encoding target block (that is, encoding). This is the direction in which the target pixel is arranged. The prediction direction depends on the intra coding mode, but a plurality of prediction directions are prepared, and any one of them can be selected. The method for obtaining the difference between the intra-frame prediction encoding target pixel and the prediction value (this is called an intra prediction mode) differs depending on the prediction direction.

  FIG. 7 is a diagram showing a prediction mode (intra prediction mode) for the intra coding mode. As shown in the figure, in the Intra_4 × 4 mode, nine types of intra prediction modes 0 to 8 are determined. One of them can be selected. In Intra_8 × 8 mode, nine types of intra prediction modes of intra prediction modes 0 to 8 are determined. In Intra_16 × 16 mode, four types of intra prediction modes of intra prediction modes 0 to 3 are determined. One of them can be selected.

  FIG. 8 is a diagram showing an intra prediction mode in the Intra_4 × 4 mode, in which FIG. 8A shows the intra prediction mode 0. Hereinafter, FIG. 8B, FIG. 8C, and FIG. ,..., (I) indicate intra prediction modes 1, 2, 3,.

In FIGS. 8A to 8I, white circles ○ are pixels in the intra prediction encoding target block (herein simply referred to as encoding target block) (that is, intra prediction encoding target pixel: It is simply referred to as a pixel to be encoded) and has an array of 4 rows × 4 columns. In FIG. 8 (a), for the convenience of explanation, reference numerals 1, 2, 3,..., 16 are assigned to the respective encoding target pixels in the horizontal direction from the upper left corner to the lower right corner in the arrangement order. A black circle ● indicates a pixel (hereinafter referred to as a prediction value pixel) of another block (hereinafter referred to as a peripheral block) for obtaining a prediction value for encoding the encoding target pixels 1, 2,. In addition, symbols A, B, C,... Are attached to the respective prediction value pixels, and the peripheral blocks are clearly shown in FIG. 8A, and the prediction value pixels A, B, C,. It is clear. That is,
Predicted value pixels to D: the upper coded block lowermost adjacent blocks BR _U
4 pixel prediction value pixels arranged in the horizontal direction on the side E-H: rightmost neighbor blocks BR _L on the left side of the encoding target block
4 pixel prediction value pixels arranged in the vertical direction on the side I: right adjacent blocks BR _UL of the upper left corner side of the encoding target block
This is the pixel at the bottom corner. The relationship between the predicted value pixels A, B, C,... And the peripheral blocks is the same as in FIGS.

  In FIGS. 8A, 8B, and 8D to 8I, the direction indicated by the arrow indicates the direction of prediction, and the prediction value obtained using the prediction value pixel obtains this prediction value. This is used to encode the encoding target pixels arranged in the prediction direction from the prediction value pixels used in the above.

When the intra prediction mode 0 shown in FIG. 8 (a) is described as an example, the prediction direction is the downward direction, in this case, upper side viewed from the encoding target block adjacent blocks BR _U of pixels A, B, C , D are prediction value pixels. Then, from the prediction direction, the prediction value of the encoding target pixels 1, 5, 9, and 13 in which the pixel value of the prediction value pixel A is arranged in the downward direction of the prediction value pixel A (this is the prediction value A). For each of the encoding target pixels 1, 5, 9, and 13, the difference between the pixel value and the predicted value A is obtained and encoded.

  Similarly, the pixel value of the prediction value pixel B becomes the prediction value B of the encoding target pixels 2, 6, 10, and 14 arranged below, and the pixel value of the prediction value pixel C is arranged below it. The prediction value C of the encoding target pixels 3, 7, 11, and 15 becomes the prediction value C, and the pixel value of the prediction value pixel D becomes the prediction value C of the encoding target pixels 4, 8, 12, and 16 arranged below, respectively. The difference between the encoding target pixel and the predicted value is obtained, and the difference is encoded by DCT transform, quantization, and variable length encoding.

  The same applies to the intra prediction modes 1 to 8 shown in FIGS. 8B to 8I, and encoding is performed from the pixel values of the pixels of the adjacent blocks that emit the arrow along the indicated arrow line indicating the prediction direction. A predicted value of the target pixel is generated.

  In the intra prediction mode 2 shown in FIG. 8C, the prediction values of all the encoding target pixels are set to the average pixel values of the pixels A to H adjacent to the encoding target block as the prediction value pixels. To do.

  As described above, pixels of adjacent blocks adjacent to the upper side, left side, and upper left corner of the encoding target block are set as predicted value pixels, and a predicted value is obtained from the pixel value of the predicted value pixel by a predetermined calculation formula, and the prediction is performed. The difference between the value and the pixel value of the encoding target pixel in the encoding target block is obtained, and the difference value is encoded.

  In the Intra_8 × 8 mode (FIG. 7), one block is set to 8 × 8 pixels, and nine types of intra prediction modes similar to the Intra_4 × 4 mode shown in FIG. 8 can be selected.

  Intra_16 × 16 mode (FIG. 7) is an encoding target block in which an MB of 16 × 16 pixels is an encoding target block, and can be selected from the four types of intra prediction modes 0 to 3 shown in FIG. Yes. These intra prediction modes 0 to 3 are also arranged in the prediction value and the prediction direction obtained from the pixels of the peripheral blocks of the encoding target block for each encoding target pixel of the encoding target block, as in the intra prediction mode of FIG. A difference from the pixel value of the encoding target pixel is obtained, and the difference is encoded.

  Note that, in the intra prediction mode 0 shown in FIG. 9A, the pixels arranged in the lowermost column of the upper adjacent block are predicted value pixels in the same manner as the intra prediction mode 0 shown in FIG. The pixel value is used as a predicted pixel value, and a difference from the pixel value of the encoding target pixel arranged in the prediction direction below the predicted value pixel is obtained. 9B is the same as the intra prediction mode 1 shown in FIG. 8B, and the intra prediction mode 2 shown in FIG. 9C is the intra prediction mode 1 shown in FIG. This is the same as in prediction mode 2. In the intra prediction mode 3 shown in FIG. 9D, the prediction direction is inclined, and the encoding process of the prediction value and the encoding target pixel is the same as described above.

  The image decoding apparatus decodes each pixel in which the difference from the predicted value is encoded in this way into a pixel having the original pixel value.

  FIG. 10 is a diagram illustrating an example of decoding operation of intra-screen predictive-coded MBs in one screen composed of slices (that is, P slices or B slices) in which the inter-screen predictive coding mode is used.

FIG. 10A shows, for example, one screen made up of P slices, and FIG. 10B shows an enlarged area ER made up of 2 MB × 2 MB in this screen. .

  In FIG. 10B, the upper left 16 × 16 pixel MB is MB-0, the upper right MB is MB-1, the lower left MB-2, and the lower right MB is MB-3. MB-0 to MB-2 indicated as "inter prediction" are blocks obtained by decoding inter-frame prediction encoding, and MB-3 indicated as "intra prediction" is illustrated in FIG. 9A, for example. It is assumed that the encoding is performed in the intra prediction mode 0 and the intra prediction encoding mode. Accordingly, FIG. 10B shows a state where..., MB-0, MB-1,... Are decoded in order,..., MB-2 is decoded, and then MB-3 is decoded. It shall be in The decoded inter-screen predictive coding block is hereinafter referred to as an inter-screen predictive decoding block, and MB-3 which is the target of intra-screen predictive decoding is hereinafter referred to as an intra-screen predictive decoding target block.

  Then, in MB-3 as the intra prediction decoding target block at this time, the difference between the pixel (intra prediction decoding target pixel) and the prediction value in the intra prediction mode 0 shown in FIG. Here, it is assumed that each pixel of MB-3 is subjected to variable length decoding, inverse quantization, and IDCT conversion, and the pixel value is represented by a difference value from the predicted value. .

Therefore, in the MB-3 that is the intra-frame predictive decoding target block, the pixel value of the lowermost pixel of the MB-1 that has already been decoded as the inter-screen predictive decoding block is used as a reference value, and this is used as the MB-3. The original pixel value is restored by adding to the pixel value represented by the above difference in the corresponding intra prediction decoding target pixel. In this way, a pixel whose pixel value is a reference value in an intra-screen predictive decoding target block is hereinafter referred to as a reference pixel, and a decoded block including this reference pixel is hereinafter referred to as a reference block.

  FIG. 11 is an enlarged view of the portion of MB-3 in FIG. 10B, and the white circles are the intra-screen prediction decoding target pixels, and from these 16 × 16 intra-screen prediction decoding target pixels. This part is MB-3 as the intra-frame predictive decoding target block in FIG. Further, pixels A, B, C,... Indicated by black circles are reference pixels for this MB-3, and these are arranged on the lowermost side of MB-1 as a reference block in FIG. Is. Furthermore, the arrows from the reference pixels A, B, C,... Indicate the prediction direction of decoding of the intra-screen predictive decoding target block, and this coincides with the prediction direction at the time of encoding.

  In FIG. 11, the prediction direction in this case is the downward direction indicated by the arrow from the reference pixels A, B, C,... By adding the pixel value of the reference pixel A as the reference value to the pixel value represented by the difference of the decoding target pixels, the pixel values of the 16 decoding target pixels are converted into the original values before encoding. The pixel value is restored.

  Similarly, for the 16 decoding target pixels arranged downward from the reference pixel B, the original pixel value is restored by using the pixel value of the reference pixel B as a reference value, and the reference pixel For the 16 decoding target pixels arranged in the downward direction from C, the original pixel value is restored by using the pixel value of the reference pixel C as a reference value, and thus intra-screen predictive decoding is performed. The pixel values of all the decoding target pixels in MB-3 that is the conversion target block are restored to the original values.

  This means that, when encoding, the intra prediction decoding target block using the intra prediction mode shown in FIGS. 9B to 9D is also used as a prediction value at the time of encoding to a prediction value pixel. Obtaining a reference value for reproducing the predicted value from a corresponding pixel (reference pixel), and using this to restore the pixel value of the decoding target pixel in the intra prediction decoding target block to the original value it can. The same applies when the intra prediction mode of Intra_4 × 4 mode shown in FIG. 8 is used or when the intra prediction mode of Intra_8 × 8 mode is used.

  In this way, image data encoded according to the H.264 / AVC standard is decoded, but in order to secure a reference block for decoding the inter-screen prediction encoded block, the decoded screen is Accumulated and saved in memory. Then, when decoding the inter-screen predictive coding block to be decoded, the decoded screen including the block serving as the reference block is read from the memory, and this block is the reference block of the inter-screen predictive decoding target block. Used as

  The image decoding apparatus according to the present invention has a down-decode function for reducing the memory capacity for storing the decoded screen in the memory. With this down-decode function, when the decoded screen is written to the memory, the decoded screen is subjected to pixel thinning to reduce the screen size (compression: down-conversion), and when the decoded screen is read from the memory, pixel interpolation is performed. The processing returns to the original screen size (up-conversion). This makes it possible to use a memory having a smaller capacity than an image decoding apparatus that does not have a down-decoding function.

  However, when the decoding screen is down-decoded in this way, the pixels that are thinned out, interpolated and restored cannot accurately have the original pixel values, and become pixel values including errors. Accordingly, in FIG. 11, each pixel of MB-1 that is a reference block for MB-3 that is the intra-frame predictive decoding target block has a pixel value including an error, and the reference pixel A in this MB-1 , B, C,... Include errors. Therefore, when the pixel value of the intra-frame predictive decoding target pixel of MB-3 is restored with the reference value of the reference pixels A, B, C,... Including such an error, the restored pixel value includes an error. As a result, the influence of errors propagates in the spatial direction to the entire MB-3 block.

  In the inter-frame predictive decoding block that becomes a reference block of the intra-frame predictive-coded block, the down-conversion / up-conversion is performed by the down-decoding function, thereby causing a drift error in the time axis direction.

  Further, when a superimposed drift error related to the decoded pixel of the inter-screen prediction decoding block that becomes the reference pixel of the intra-screen prediction decoding target block is included, the predetermined pixel of the inter-screen prediction decoding block is referred to as a reference pixel. In the intra-frame predictive decoding target block that is decoded in this manner, the same error spills over in the spatial direction in the block, and the image quality deterioration is further increased as compared with other blocks.

  As described above, in the conventional image decoding device that reduces the size of the memory for down-decoding by performing the above-described down-decoding on the inter-screen predictive decoded screen, the intra-screen predictive coding is performed. When a block is decoded, the above-described superimposed error included in the reference pixel propagates in the spatial direction in the intra-frame predictive decoding target block, but this point has not been considered conventionally. For this reason, the image quality degradation of the intra prediction decoding block portion existing between the inter prediction decoding blocks in the decoded screen is larger than the other portions, and subjective image quality degradation is noticeable. There was a problem that it was easy.

  The present invention has been made in view of such a problem, and an object thereof is to reduce image quality deterioration in an intra-screen predictive decoded block existing between inter-screen predictive decoded portions in a screen. It is an object of the present invention to provide an image decoding apparatus and an image decoding method that can perform the above.

In order to achieve the above object, the present invention provides an I slice consisting of a block according to an intra prediction encoding mode consisting of a predetermined number of pixels encoded in an intra prediction encoding mode as an encoding unit. The predetermined number of blocks encoded by the intra prediction encoding mode and the block by the intra prediction encoding mode consisting of the predetermined number of pixels encoded by the intra prediction encoding as the encoding unit. An image decoding apparatus that decodes encoded image data including P and B slices including blocks in an inter-screen predictive encoding mode including pixels, wherein the I, P, and B slices The pixel in the block in the intra prediction encoding mode is set as the intra prediction decoding target pixel, and the pixel in the other block adjacent to the block in the same screen as the block is the reference pixel. Other pixels that precede or follow the screen containing the block as the inter-screen decoding target pixel, using the pixel of the block in the inter-screen predictive coding mode in the P and B slices. A decoding unit that performs decoding using a pixel of another block at a position corresponding to the block on the screen as a reference pixel;
An image correction unit to which a decoding block based on a decoded pixel of the intra-screen prediction decoding target pixel from the decoding unit is supplied, and an intra-screen decoding target from the image data of the decoding block output from the image correction unit Intra-screen prediction unit that generates the reference pixel for decoding the pixel, the decoding block output from the image correction unit, and the decoding of the inter-screen decoding target pixel output from the decoding unit An image reduction unit that reduces the image size of the image data by the block, a storage unit that holds the image data with the reduced image size, and an image that expands the image data read from the storage unit to the original image size A decompression unit, and a motion compensation unit that creates the reference pixel for decoding the inter-picture prediction decoding target pixel from the image data from the image decompression unit, wherein the image correction unit is the decoding unit Of the decoded blocks in the intra prediction encoding mode supplied, the intra prediction encoding mode decoded in the P and B slices and decoded with the pixel of the decoding block in the inter prediction encoding mode as a reference pixel The decoding block is a decoding block to be corrected, and image correction is performed to remove a drift error in the time-axis direction due to decoding using the decoding block pixel in the inter-frame prediction encoding mode as a reference pixel. The correction target decoding block of the image correction unit includes a reference screen target block corresponding to the correction target block on the inter-screen prediction decoding reference screen used for inter-screen prediction decoding, and the correction target decoding block. The average sum of absolute differences of pixel values between the reference block and the reference screen target block is equal to or less than a preset threshold value, correlated, and active The image correction unit has an average pixel value of pixels of the reference screen target block and an average of pixel values of the pixels of the correction target decoding block. A value obtained by multiplying the difference from the difference by a coefficient corresponding to the average value of the sum of absolute differences of the pixel values of the decoding block to be corrected and the reference screen target block is used as the drift error, and each pixel of the decoding block And the image correction of the decoding block to be corrected is performed.

In order to achieve the above object, the present invention provides an I slice consisting of a block according to an intra prediction encoding mode consisting of a predetermined number of pixels encoded in an intra prediction encoding mode as an encoding unit. The predetermined number of blocks encoded by the intra prediction encoding mode and the block by the intra prediction encoding mode consisting of the predetermined number of pixels encoded by the intra prediction encoding as the encoding unit. An image decoding apparatus that decodes encoded image data including P and B slices including blocks in an inter-screen predictive encoding mode including pixels, wherein the I, P, and B slices The pixel in the block in the intra prediction encoding mode is set as the intra prediction decoding target pixel, and the pixel in the other block adjacent to the block in the same screen as the block is the reference pixel. Other pixels that precede or follow the screen containing the block as the inter-screen decoding target pixel, using the pixel of the block in the inter-screen predictive coding mode in the P and B slices. A decoding unit that decodes using a pixel of another block at a position corresponding to the block on the screen as a reference pixel, and a decoding block by a decoded pixel of the intra-screen prediction decoding target pixel from the decoding unit; A supplied image correction unit, an intra-screen prediction unit that generates the reference pixel for decoding the intra-screen decoding target pixel from the image data of the decoded block output from the image correction unit, and the image correction An image reduction unit for reducing an image size of image data by the decoding block output from the decoding unit and a decoding block by the decoding pixel of the inter-screen decoding target pixel output from the decoding unit; A storage unit that holds image data with a reduced image size, an image expansion unit that expands the image data read from the storage unit to the original image size, and the inter-screen prediction from the image data from the image expansion unit A motion compensation unit that creates the reference pixel for decoding the pixel to be decoded, and the image correction unit includes a decoding block of the intra prediction encoding mode supplied from the decoding unit, A decoding block in the intra prediction encoding mode that belongs to the P and B slices and is decoded using a pixel in the decoding block in the inter prediction encoding mode as a reference pixel is set as a correction target decoding block, and the correction target decoding block On the other hand, image correction is performed to remove the drift error in the time axis direction due to the decoding of the pixel of the decoding block in the inter prediction coding mode as the reference pixel, and the correction of the image correction unit is performed. The decoding block of the positive target is preset with an adjacent block temporally preceding the decoding block of the correction target, and an average sum of absolute differences of pixel values of the decoding block of the correction target and the adjacent block. The decoded block that is less than or equal to a threshold value and has a correlation and the activity is less than or equal to a predetermined threshold value and image quality degradation is conspicuous, and the image correction unit includes an average value of pixel values of pixels of the adjacent block A value obtained by multiplying the difference from the average value of the pixel values of the pixels of the decoding block by the coefficient corresponding to the average value of the sum of absolute differences of the pixel values of the decoding block to be corrected and the adjacent block is used as the drift error. The image correction of the decoding block to be corrected is performed by adding to each pixel of the decoding block.

In order to achieve the above object, the present invention provides an I slice consisting of a block according to an intra prediction encoding mode consisting of a predetermined number of pixels encoded in an intra prediction encoding mode as an encoding unit. The predetermined number of blocks encoded by the intra prediction encoding mode and the block by the intra prediction encoding mode consisting of the predetermined number of pixels encoded by the intra prediction encoding as the encoding unit. An image decoding method for decoding encoded image data consisting of P and B slices consisting of blocks in an inter-screen predictive encoding mode consisting of pixels, wherein the I, P and B slices The pixel in the block in the intra prediction encoding mode is set as the intra prediction decoding target pixel, and the pixel in the other block adjacent to the block in the same screen as the block is the reference pixel. Other pixels that precede or follow the screen containing the block as the inter-screen decoding target pixel, using the pixel of the block in the inter-screen predictive coding mode in the P and B slices. Decoding processing is performed using a pixel in another block at a position corresponding to the block on the screen as a reference pixel, and P and B slices among the decoded blocks of the decoded block in the intra prediction encoding mode. And a decoding block in the intra prediction encoding mode decoded using a pixel in the decoding block in the inter prediction encoding mode that has been subjected to decoding processing as a reference pixel, and the decoding block to be corrected On the other hand, image correction is performed to remove the drift error in the time axis direction due to the decoding of the pixel of the decoding block in the inter prediction coding mode as the reference pixel. A decoded block subjected to block decoding processing in the intra prediction encoding mode, a decoded block subjected to image correction in the block corresponding to the intra prediction encoding mode, and a block decoding processing in accordance with the inter prediction encoding mode. The reference pixel for decoding the intra-screen decoding target pixel is generated from the image data of the decoded block, and the decoded block of the block encoded by the intra-screen prediction encoding or the inter-screen prediction is generated. The decoded block of the block encoded by the encoding is stored in the storage unit with the screen size reduced, the decoded block read from the storage unit is expanded to the original image size, and the original image The pixel of the decoded block expanded to the size is used as a reference pixel for decoding the pixel of the block in the inter-picture prediction encoding mode, and the decoding block to be corrected for image correction is used. The reference screen target block corresponding to the correction target block on the inter-screen predictive decoding reference screen used for inter-screen predictive decoding, and the pixel value difference absolute value between the correction target decoding block and the reference screen target block This is a decoded block in which the average value of the sum is less than or equal to a preset threshold value, there is a correlation, the activity is less than the predetermined threshold value, and image quality degradation is conspicuous. The value obtained by multiplying the difference from the average pixel value of the pixels of the target decoding block by the coefficient corresponding to the average value of the sum of absolute differences of the pixel values of the correction target decoding block and the reference screen target block is the drift error. As described above, it is added to each pixel of the decoding block to perform image correction of the decoding block to be corrected .

In order to achieve the above object, the present invention provides an I slice consisting of a block according to an intra prediction encoding mode consisting of a predetermined number of pixels encoded in an intra prediction encoding mode as an encoding unit. The predetermined number of blocks encoded by the intra prediction encoding mode and the block by the intra prediction encoding mode consisting of the predetermined number of pixels encoded by the intra prediction encoding as the encoding unit. An image decoding method for decoding encoded image data consisting of P and B slices consisting of blocks in an inter-screen predictive encoding mode consisting of pixels , wherein the I, P and B slices The pixel in the block in the intra prediction encoding mode is set as the intra prediction decoding target pixel, and the pixel in the other block adjacent to the block in the same screen as the block is the reference pixel. Other pixels that precede or follow the screen containing the block as the inter-screen decoding target pixel, using the pixel of the block in the inter-screen predictive coding mode in the P and B slices. Decoding processing is performed using a pixel in another block at a position corresponding to the block on the screen as a reference pixel , and P and B slices among the decoded blocks of the decoded block in the intra prediction encoding mode. And a decoding block in the intra prediction encoding mode decoded using a pixel in the decoding block in the inter prediction encoding mode that has been subjected to decoding processing as a reference pixel, and the decoding block to be corrected On the other hand, image correction is performed to remove the drift error in the time axis direction due to the decoding of the pixel of the decoding block in the inter prediction coding mode as the reference pixel. A decoded block subjected to block decoding processing in the intra prediction encoding mode, a decoded block subjected to image correction in the block corresponding to the intra prediction encoding mode, and a block decoding processing in accordance with the inter prediction encoding mode. The reference pixel for decoding the intra-screen decoding target pixel is generated from the image data of the decoded block, and the decoded block of the block encoded by the intra-screen prediction encoding or the inter-screen prediction is generated. The decoded block of the block encoded by the encoding is stored in the storage unit with the screen size reduced, the decoded block read from the storage unit is expanded to the original image size, and the original image The pixel of the decoded block expanded to the size is used as a reference pixel for decoding the pixel of the block in the inter-screen predictive coding mode, and the correction target of the image correction is decoded. In the block, the average value of the sum of absolute differences of adjacent pixel blocks of the correction target decoding block and the pixel values of the correction target decoding block and the adjacent block is equal to or less than a preset threshold value. Thus, the decoded block has a correlation and the activity is not more than a predetermined threshold value, and the image quality deterioration is conspicuous . A value obtained by multiplying the difference by a coefficient corresponding to the average value of the sum of absolute differences of the pixel values of the decoding block to be corrected and the adjacent block is added to each pixel of the decoding block as a drift error, thereby decoding the correction target. The image correction of the block is performed.

  According to the present invention, since the decoded block of the intra prediction coded block is subjected to image correction, the decoded block decodes the inter prediction coded block and uses this as a reference block. The image deterioration due to the correction is corrected, and subjective image quality deterioration due to the decoded block in the decoded screen can be reduced.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a configuration block diagram showing a first embodiment of an image decoding apparatus and an image decoding method according to the present invention, where 1 is a decoding unit (a portion surrounded by a one-dot chain line) and 2 is a variable length decoding. , 3 is an inverse quantization unit, 4 is an IDCT (Inverse Discrete Cosine Transform) unit, 5 and 6 are adders, 7 is an image correction unit, 8 is a storage unit, and 9 is in the screen. A prediction unit, 10 is a motion compensation unit, 11 is a deblocking filter unit, 12 is an image reduction unit, 13 is a storage unit, and 14 is an image expansion unit.

  In the figure, the image decoding apparatus according to the first embodiment includes a decoding unit 1, an image correction unit 7, an image reduction unit 12, a storage unit 13, and an image expansion unit 14. The decoding unit 1 includes a variable length decoding unit 2, an inverse quantization unit 3, an IDCT unit 4, adders 5 and 6, a storage unit 8, a deblocking filter unit 11, an intra-screen prediction unit 9, and a motion compensation unit. 10 is provided. Each of these units may be configured by hardware or software. Further, it may be a module combining hardware and software.

  Here, the decoding unit 1, the image reduction unit 12, the storage unit 13, and the image expansion unit 14 have the same configuration as that of a conventional image decoding device using a down-decoding function, and this first embodiment is In this configuration, an image correction unit 7 is added. Note that the image reduction unit 12, the storage unit 13, and the image expansion unit 14 provide the image decoding apparatus with the down-decoding function described above.

  Next, the operation of the first embodiment will be described.

  The input image data is subjected to intra-screen or inter-screen predictive encoding as described with reference to FIGS. 8 and 9 in an image encoding device (not shown), and then DCT (Discrete Cosine Transform: two-dimensional discrete) for each block. The DCT coefficient is quantized by the quantization unit, and further converted into a variable length code by the variable length encoding unit and compressed and encoded. This input image data is processed by a processing unit (not shown), then supplied to the variable length decoding unit 2 and subjected to variable length decoding. Further, the input image data is supplied to the inverse quantization unit 3 to obtain image data composed of DCT coefficients. can get. This image data is supplied to the IDCT unit 4 and is subjected to inverse DCT conversion using the supplied DCT coefficients in units of blocks which are encoding units in the image encoding device, thereby depending on the intra prediction encoding mode. Image data of a screen composed of P slices and B slices in I-slice and inter-screen predictive coding mode is obtained.

  Here, as described above, the P and B slices in the inter prediction encoding mode are divided into blocks encoded by the intra prediction encoding mode and blocks encoded by the inter prediction encoding mode. As described with reference to FIGS. 8 and 9, the blocks encoded by the IDCT unit 4 in the intra prediction encoding mode that is the unit of the inverse DCT transform are the original pixel value and the predicted value. It is a block which consists of a pixel (difference value pixel) which makes a difference value of a pixel value.

  The coding block in the intra prediction encoding mode that has been subjected to inverse DCT conversion by the IDCT unit 4 is supplied to the adder 5, and the encoding block that has been subjected to the inter DC prediction encoding that has been subjected to inverse DCT conversion is supplied to the adding unit 6. The

  In the adder 5, an intra-screen prediction unit 9 is added to each difference value pixel (that is, the intra-screen predictive decoding target pixel) in the coding block obtained by intra-screen predictive coding as the intra-screen predictive decoding target block. The corresponding reference pixels generated in (1) are added to obtain a decoded block composed of pixels of the original pixel value. The addition process of the reference pixel and the difference value pixel in the coding block in this case is the same as that described in FIG. 11, and the difference value pixel arranged in the prediction direction is added to the reference pixel. . The decoded block generated by the adder 5 is subjected to correction processing described later in the image correction unit 7 and then supplied to and stored in the storage unit 8.

  In addition, in addition unit 6, the corresponding reference generated in motion compensation unit 10 for each difference value (difference value with a corresponding prediction value in a past or future screen) in a coding block by inter-screen prediction encoding. The images are added to obtain a decoded block made up of pixels of the original pixel value. The decoded block generated by the adding unit 6 is supplied to and stored in the storage unit 8.

The storage unit 8 is provided in order to obtain a reference pixel corresponding to a difference pixel of an encoded block by intra prediction encoding from the IDCT unit 4 to the adding unit 5. Therefore, the decoded block is stored in the storage unit 8 until the pixel is not used as a reference pixel for the difference value pixel of the encoded block by intra prediction encoding. When the difference value pixel of the coding block by the prediction prediction in the screen is supplied from the IDCT unit 4 to the adding unit 5 from the IDCT unit 4, the decoding block including a pixel serving as a reference pixel for the difference value pixel. Is read from the storage unit 8 as the above reference block, a reference value of the difference value pixel supplied to the adding unit 5 is created based on the pixel serving as the reference pixel, and supplied to the adding unit 5. Thereby, the reference value corresponding to this is added to the pixel value of the difference value pixel, and a pixel restored to the original pixel value is obtained.

For example, when the decoding process shown in FIG. 11 is performed, when MB-3, which is an encoded block by intra prediction encoding, is used as the intra prediction decoding target block, and the difference value pixel is supplied to the adding unit 5. The decoded block MB-1 subjected to the inter-screen predictive decoding including the pixels A, B, C,... Serving as reference pixels for this is read from the storage unit 8 and supplied to the intra-screen prediction unit 9 to be the pixel. Reference values are created from A, B, C,. Of course, even when encoding is performed as shown in FIGS. 8 and 9, similarly, a prediction value pixel that forms a prediction value is set as a reference value pixel.

  In this way, the encoded block in the intra prediction encoding mode is decoded and stored in the storage unit 8 as a decoded block. Note that the decoding block of the coding block in the inter-picture prediction coding mode is also stored in the storage unit 8 because this decoding block can also be a reference block for a coding block in another intra-picture prediction coding mode. . For example, in FIG. 10B, MB-1 is a block encoded by the inter-picture prediction encoding mode, but is stored in the storage unit 8 because it is a reference block for MB-3 decoding.

  From the storage unit 8, a block that is not used as a reference block or a block that is no longer used as a reference block for a decoded block by intra prediction encoding is read and supplied to the deblocking filter unit 11. The deblocking filter unit 11 reduces block distortion such as a difference in color and luminance between blocks due to encoding and decoding by dividing each block, and performs filter processing so that images are smoothly connected between blocks. This is one of the encoding tools newly added in the H.264 / AVC standard, similar to the above-mentioned intra prediction encoding.

  The image data of each block subjected to the filtering process by the deblocking filter unit 11 is supplied to the image reduction unit 12, and a pixel thinning process is performed to convert a preset image size into a predetermined format at a lower level (down-conversion). And stored in the storage unit 13. In the storage unit 13, each block is stored in an array constituting one screen. Further, the image data of the screen is read from the storage unit 13, and pixel interpolation processing is performed for each block by the image expansion unit 14 (up-conversion) and converted to the original format. The image data output from the image decompression unit 14 is output as a decoded image, and is supplied to the motion compensation unit 10 in order to form a reference block of the inter-screen predictive coding block.

  Next, the image correction unit 7 will be described.

  FIG. 2 is a flowchart showing a specific example of the processing operation of the image correction unit 7.

  In FIG. 1, as described above, the intra-frame prediction decoding target pixel (difference value pixel) of the intra-frame prediction decoding target block subjected to intra-frame prediction encoding is supplied to the adder 5, and the pixel value (difference value) is supplied. ) Is added to the reference value supplied from the in-screen prediction unit 9 to obtain a decoded block composed of pixels of the original pixel value. The image correction unit 7 determines whether or not the decoded block supplied from the adding unit 5 belongs to the I slice (S101 in FIG. 2), and when it belongs to the I slice ( In step S101 of FIG. 2, “YES”), the process is terminated, the block of this I slice is sent to the storage unit 8, and the process waits until the next slice is processed. In the case of an I slice, as described above, each block is encoded only in the intra prediction encoding mode, and in the case of decoding this, the intra prediction prediction decoding of the same I slice is performed. This is because the block is decoded as a reference block without down-decoding, so there is no deterioration in image quality, and therefore there is no need for image correction.

If the decoded block supplied from the adder 5 is a block in P slice or B slice (“NO” in S101 of FIG. 2), then the reference block for the decoded block from the adder 5 is appropriate. It is determined whether or not it is a block (S102 in FIG. 2). This is because the screen including the P or B slice of the decoded block supplied from the adder 5 is P ₁ (hereinafter referred to as a decoding target screen), and the P or B slice of the decoding target screen P ₁ If a past or future screen including a block that becomes a reference block with respect to an inter-screen predictive encoded block is P ₂ (hereinafter referred to as an inter-screen predictive decoding reference screen), these screens P ₁ and P _{2 are used.} In the meantime, for example, when there is a large change in the image due to a scene (scene) change or the like, the decoding block supplied from the adding unit 5 is subjected to image correction. It is what.

In order to make such a determination, the pixel value of the pixel of the decoding block supplied from the adding unit 5 on the decoding target screen P ₁ is set to DEC _ij (where the size of this decoding block is m pixels × n pixels, i = 1, 2,..., M, j = 1, 2,..., N), corresponding to the spatial position of the decoding block on the decoding target screen P ₁ in the inter-screen predictive decoding reference screen P _2. If the pixel value of the pixel of the block in the spatial position and the same size as this decoding block (hereinafter referred to as the reference screen target block) is REFTMP _ij , the sum of absolute differences (SAD) of these pixel values Average value, ie

This determination is made based on whether or not the average value DIFF (DEC _ij , REFTMP _ij ) exceeds a preset threshold TH_tmp. That is, when the average value DIFF (DEC _ij , REFTMP _ij ) exceeds the threshold value TH_tmp (“NO” in S102 in FIG. 2), it is determined that there is no correlation between the screens P ₁ and P ₂ and the processing is performed. And the decoding block supplied from the adding unit 5 of the decoding target screen P ₁ is sent to the storage unit 8, and then waits until the decoding block is supplied from the adding unit 5. If the average value DIFF (DEC _ij , REFTMP _ij ) is equal to or less than the threshold value TH_tmp (“YES” in S102 of FIG. 2), the process proceeds to S103 next in FIG.

It should be noted that the determination in S102 for each decoded block is not limited to the scene change, and there may be a partial correlation between the screens P ₁ and P _{2. In} such a case, there is no correlation. The processing in the image correction unit 7 is not performed only for the portion. For the decoded block in the correlated portion, the correction processing is performed in the image correction unit 7 assuming that the reference pixel at the time of decoding is appropriate. Candidates to do.

The decoding block from the adding unit 5 in which the average value DIFF (DEC _ij , REFTMP _ij ) is equal to or less than the threshold value TH_tmp (“YES” in S102 in FIG. 2) is next an image edge or a complicated pattern. It is determined whether or not the block is a substantially uniform block of information content (S103 in FIG. 2).

Such determination is to evaluate seeking activities decoded block, this activity is the pixel value of the pixel of the decoded blocks as DEC _ij, the following evaluation formula ACT (DEC _ij), i.e.,

It is requested from. Here, DEC _ave is the average pixel value of the decoded block composed of m × n pixels, and the activity calculated by the evaluation formula ACT (DEC _ij ) is the pixel value DEC _ij of the decoded block and its average pixel value DEC _ave Is the average value of the sum of square differences (SSD).

S103 in FIG. 2 compares this activity ACT (DEC _ij ) with a preset threshold TH_act, and if the activity ACT (DEC _ij ) exceeds this threshold TH_act, a decoding block that has almost no uniform information content. (NO in S103 of FIG. 2), the process ends, this decoded block is sent to the storage unit 8, and it waits until the next decoded block is supplied from the adding unit 5. If the activity ACT (DEC _ij ) is less than or equal to this threshold value TH_act, it is determined that the block is a decoding block having almost uniform information content (“YES” in S103 of FIG. 2), and image correction of this decoding block is performed. (S104 in FIG. 2).

  As described above, in the image correction unit 7, the decoded block of the intra prediction encoding block supplied from the adding unit 5 does not belong to the I slice (S101 in FIG. 2). A block corresponding to a spatial position in the decoded block in a screen that becomes a reference block at the time of inter-screen predictive decoding on a decoding target screen including a P or B slice to which the block belongs (that is, an inter-screen predictive decoding reference screen) 2 (S102 in FIG. 2) and almost uniform information content (S103 in FIG. 2), the drift error of the reference block decoded by inter-screen prediction is calculated in S104 in FIG. The image correction to be removed is performed.

  The image correction of S104 in FIG. 2 is to correct the pixel value of each pixel in the decoded block using the pixel value of the pixel corresponding to the spatial position of this decoded block on the inter-screen predictive coding reference screen. .

This image correction is performed by the following equation (3):

Here, in parentheses in the second term on the right side of Equation 3 is a difference (average error) between the average value of the pixel value REFTMP _ij in the reference screen target block and the average value of the pixel value DEC _{ij in} the decoded block, A value obtained by multiplying the average error by the weighting coefficient α is added to each pixel DEC _ij of the decoding block as the drift error to perform image correction.

FIG. 3 is an explanatory diagram showing the setting of the weighting coefficient α in Equation 3, and the horizontal axis indicates the correlation between the decoding block of the decoding target screen shown in Equation 1 and the reference screen target block of the inter-screen predictive decoding reference screen. Represents the value of the evaluation formula DIFF (DEC _ij , REFTMP _ij ) of the above equation 1 and the vertical axis represents the weighting coefficient α.

In the figure, the weighting coefficient α corresponds to the value of the evaluation expression DIFF (DEC _ij , REFTMP _ij ) used in S102 of FIG. 2, and the value of the calculation result of the evaluation expression DIFF (DEC _ij , REFTMP _ij ) is As the value increases, the weighting coefficient α is set larger. As a result, the larger the difference between the pixel value DEC _ij in the decoding block and the pixel value REFTMP _ij in the corresponding reference screen target block, the stronger the image correction is performed, and the difference between these pixel values DEC _ij and REFTMP _ij The correction is controlled so as to be small.

  As described above, in the first embodiment, the decoded block is decoded with respect to the decoded block (that is, the decoded block) of the intra-frame predictive coded block in the inter-frame predictive coded slice. A block in a spatial position corresponding to the decoding block on the screen (that is, the inter-screen prediction decoding reference screen) used for the inter-screen prediction decoding on the screen including the decoding screen (that is, the decoding target screen) (that is, the reference screen target block) Therefore, subjective image quality degradation in the decoded block can be reduced.

  For such image correction, when the decoded block belongs to the I slice, the difference between the pixel values of the decoded block and the reference screen target block is larger than a predetermined threshold, and the activity of the decoded block is larger than the predetermined threshold. In such a case, by setting a condition not to perform image correction, for example, when the image is greatly different between the decoded block and the reference screen target block due to a scene change or the like, or when the spatial frequency in the decoded block is high, image quality deterioration is conspicuous. If it is difficult, the image correction can be applied only to a region that is excluded from the image correction target and the image quality deterioration is conspicuous.

  Next, a second embodiment of the image decoding apparatus and the image decoding method according to the present invention will be described. The configuration is the same as that in FIG. 1, and the description thereof will be omitted.

  FIG. 4 is a flowchart showing a specific example of the operation of the image correction unit 7 (FIG. 1) in the second embodiment.

  In the first embodiment, in the image correction shown in FIG. 2 in the image correction unit 7, the decoding block to be corrected and image processing are determined for the screen (decoding target screen) including the decoding block. A block of a past or future screen (that is, an inter-screen predictive coding reference screen) including a block that becomes a reference block of the inter-screen predictive coding block is used. In the second embodiment, this decoding block is used. Specifically, the left block adjacent to the decoded block (for example, MB-2 with respect to MB-3 as the decoded block in FIG. 10) is used. Hereinafter, the operation of the image correction unit 7 in the second embodiment will be described with reference to FIGS. 1 and 4. The operation of the second embodiment other than this is the same as that of the first embodiment, and the description thereof is omitted.

  In FIG. 1, when a decoding block is supplied from the adder 5 to the image correction unit 7 in the same manner as in the first embodiment, the image correction unit 7 firstly has this decoding block belonging to the I slice. (S201 in FIG. 4) and if it belongs to the I slice (“YES” in S201 in FIG. 4), the process is terminated and the block of this I slice is sent to the storage unit 8. Wait until the next slice is processed. When the decoded block supplied from the adding unit 5 is a block in a P slice or a B slice (“NO” in S201 in FIG. 4), a block temporally preceding the decoded block from the adding unit 5; That is, the correlation between the block adjacent to the left side of the decoded block (hereinafter referred to as the left adjacent block) and the decoded block is determined (S202 in FIG. 4). In this determination, if there is a large change in the image content between these blocks, even if a drift error is mixed in the decoded block, the image is not particularly noticeable. For such a decoded block, image correction is performed. This is what you do not have to do.

In order to make such a determination, the pixel value of the pixel of the decoded block supplied from the adding unit 5 is set to DEC _ij (where the size of the decoded block is m pixels × n pixels, as in the first embodiment, i = 1, 2,..., M, j = 1, 2,..., N), and assuming that the pixel value of the pixel in the left adjacent block adjacent thereto is REFTSP _ij , the sum of absolute differences of these pixel values Average value of (Sum of Absolute Difference: SAD), that is,

This determination is made based on whether or not the average value DIFF (DEC _ij , REFTSP _ij ) exceeds a preset threshold TH_sp. That is, when the average value DIFF (DEC _ij , REFTSP _ij ) exceeds the threshold value TH_sp (“NO” in S202 of FIG. 4), it is determined that there is no correlation between the decoded block and the left adjacent block. The process ends and this decoded block is sent to the storage unit 8 and then waits until the next decoded block is supplied from the adding unit 5. If the average value DIFF (DEC _ij , REFTSP _ij ) is less than or equal to the threshold value TH_sp (“YES” in S202 of FIG. 4), the process proceeds to S203 next in FIG.

  S203 in FIG. 4 is similar to S103 in FIG. 2 with respect to the first embodiment, and it is determined whether or not the block has almost uniform information contents without any image edges or complicated patterns. It is.

This determination in S203 is performed by obtaining and evaluating the activity of the decoded block. As described above, this activity uses the pixel value of the pixel of the decoded block as DEC _ij and the following evaluation expression ACT (DEC _ij ), That is,

It is requested from. This number 5 is the same as the number 2 in the first embodiment.

In S203 of FIG. 4, if the same as S103 in FIG. 2, and compares the activity ACT (DEC _ij) with a preset threshold Th_act, activity ACT (DEC _ij) exceeds this threshold Th_act is It is determined that the decoded block is not substantially uniform information content (“NO” in S203 in FIG. 4), the process is terminated, this decoded block is sent to the storage unit 8, and the next decoded block is supplied from the adding unit 5 Wait until If the activity ACT (DEC _ij ) is less than or equal to this threshold TH_act, it is determined that the block is a decoding block with almost uniform information content (“YES” in S203 in FIG. 4), and image correction of this decoding block is performed. (S204 in FIG. 4).

  As described above, also in the image correction unit 7 in the second embodiment, the decoding block of the intra prediction encoding block supplied from the addition unit 5 does not belong to the I slice (S201 in FIG. 4). When this decoded block has a correlation with its left adjacent block (S202 in FIG. 4) and has almost uniform information content (S203 in FIG. 4), the inter-screen prediction is performed by S204 in FIG. Image correction is performed to remove the drift error of the decoded reference block.

  The image correction in S204 of FIG. 4 approximates the drift error in the reference block using the pixel value of the pixel in the left adjacent block of the decoded block, and corrects the pixel value of each pixel in the decoded block. is there.

This image correction is performed by the following equation 6, that is,

Here, in parentheses in the second term on the right side of Equation 6 is the difference (average error) between the average value of the pixel value REFTSP _ij in the left adjacent block and the average value of the pixel value DEC _{ij in} the decoded block. A value obtained by multiplying the average error by the weighting coefficient β is added to each pixel DEC _ij of the decoding block as the above drift error to perform image correction.

FIG. 5 is an explanatory diagram showing the setting of the weighting coefficient β in Equation 6, wherein the horizontal axis indicates the correlation between the decoded block shown in Equation 4 and the left adjacent block, and the evaluation equation DIFF (DEC _ij , REFTSP _ij ), and the vertical axis represents the weighting coefficient β.

In the figure, the weighting coefficient β corresponds to the value of the evaluation expression DIFF (DEC _ij , REFTSP _ij ) used in S202 of FIG. 4, and the value of the calculation result of the evaluation expression DIFF (DEC _ij , REFTSP _ij ) is As the value increases, the weighting coefficient β is set larger. As a result, the larger the difference between the pixel value DEC _ij in the decoding block and the corresponding pixel value REFTSP _ij in the left adjacent block, the stronger the image correction is performed, and the difference between these pixel values DEC _ij and REFTSP _ij becomes smaller. Correction control is performed to make it smaller.

  As described above, also in the second embodiment, the left side of the decoded block (ie, the decoded block) of the intra prediction encoded block in the inter prediction predictive slice Since the image correction is performed based on the adjacent block (that is, the left adjacent block), the draft error in the decoded block is reduced and subjective image quality deterioration can be reduced.

  Such image correction is performed when the decoded block belongs to the I slice, when the difference between the pixel values of the decoded block and the left adjacent block is larger than a predetermined threshold, and when the activity of the decoded block is larger than the predetermined threshold. By providing a condition that does not perform image correction, for example, when the image differs greatly between the decoded block and the left adjacent block depending on the design, or when the spatial frequency in the decoded block is high and image quality degradation is not noticeable, This method can be applied only to a region that is excluded from image correction targets and in which image quality degradation is conspicuous.

In the second embodiment, the left adjacent block is used for image correction of the decoded block. However, the upper adjacent block (for example, MB-1 in FIG. 10) may be used. Also, the left adjacent block and the upper adjacent block can be used, and the adjacent block with the smaller value of the calculation result of the evaluation expression DIFF (DEC _ij , REFTSP _ij ) expressed by Equation 4 may be used. .

  Further, by combining the processing of the image correction unit 7 in the first and second embodiments, a temporal block (that is, a reference screen target block of the inter-screen prediction decoding reference screen) or a spatial block (that is, Any of the blocks adjacent to the decoded block) may be used to perform image correction of the decoded block, and the same effect can be obtained.

  The image decoding apparatus and the image decoding method according to the present invention can be applied to an image recording / reproducing apparatus such as a DVD player, a television receiver, and a portable terminal such as a mobile phone that receives image information. For example, in a video camera equipped with an image recording / playback apparatus such as a DVD player, an encoding apparatus based on the H.264 / AVC standard and a decoding apparatus according to the present invention can be provided. In this case, the compressed and encoded image data recorded on the DVD on which the captured image data is compression-encoded based on the H.264 / AVC standard and read from the DVD is decoded according to the present invention. Decoded by the device.

1 is a configuration block diagram illustrating a first embodiment of an image decoding apparatus and an image decoding method according to the present invention. 2 is a flowchart illustrating a specific example of processing operation of an image correction unit in FIG. 1. It is explanatory drawing of a specific example of the setting of the weighting coefficient weighting coefficient (alpha) by several 3 which shows the correction process in S104 of FIG. It is a flowchart which shows the image correction process of 2nd Embodiment of the image decoding apparatus and image decoding method by this invention. It is explanatory drawing of a specific example of the setting of the weighting coefficient (beta) of 2nd Embodiment of the image decoding apparatus and image decoding method by this invention. It is a figure which shows typically the data structure by the slice of 1 screen. It is a figure which shows the list of the encoding modes and prediction modes in the prediction encoding in a screen in a H.264 / AVC specification. FIG. 8 is a diagram specifically illustrating an intra prediction mode in the Intra4 × 4 mode in FIG. 7. FIG. 8 is a diagram specifically illustrating an intra prediction mode in the Intra16 × 16 mode in FIG. 7. It is a figure which shows an example of the decoding operation | movement of the prediction encoding block in a screen which used the inter prediction encoding block as the prediction block. It is a figure which expands and shows the part of MB-3 of FIG.10 (b).

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Decoding part 2 Variable length decoding part 3 Inverse quantization part 4 IDCT part 5,6 Adder 7 Image correction part 8 Storage part 9 In-screen prediction part 10 Motion compensation part 11 Deblocking filter part 12 Image reduction part 13 Image Storage unit 14 Image expansion unit

Claims (4)

An I-slice consisting of a block according to an intra-screen prediction encoding mode consisting of a predetermined number of pixels encoded in one intra-screen prediction encoding mode as an encoding unit and an intra-screen prediction encoding as an encoding unit A block in the intra prediction encoding mode consisting of the predetermined number of pixels, and a block in the inter prediction encoding mode consisting of the predetermined number of pixels encoded in the inter prediction encoding as a coding unit. An image decoding apparatus for decoding encoded image data consisting of P and B slices consisting of:
The pixel of the block according to the intra prediction encoding mode in the I, P, and B slices is used as the intra prediction prediction target pixel, and the pixel of another block adjacent to the block in the same screen as the block is referred to Other pixels preceding or following the screen including the block, using the pixel of the block in the inter-frame predictive coding mode in the P and B slices as the inter-screen decoding target pixel A decoding unit that performs decoding using a pixel of another block at a position corresponding to the block on the screen as a reference pixel;
An image correction unit to which a decoded block of decoded pixels of the intra-screen predictive decoding target pixel from the decoding unit is supplied;
An intra-screen prediction unit that generates the reference pixel for decoding the intra-screen decoding target pixel from the image data of the decoded block output from the image correction unit;
An image reduction unit for reducing the image size of the image data by the decoding block output from the image correction unit and the decoding block output by the decoding pixel of the inter-screen decoding target pixel output from the decoding unit;
A storage unit for holding image data in which the image size is reduced;
An image expansion unit that expands the image data read from the storage unit to the original image size;
A motion compensation unit that creates the reference pixel for decoding the inter-picture prediction decoding target pixel from the image data from the image decompression unit,
The image correcting unit belongs to the P and B slices of the decoded block in the intra prediction encoding mode supplied from the decoding unit, and the pixel of the decoding block in the inter prediction encoding mode is a reference pixel. The decoding block in the intra prediction encoding mode decoded as a decoding block to be corrected is decoded, and the decoding block in the inter prediction encoding mode is decoded as a reference pixel for the correction decoding block Perform image correction to remove drift errors in the time axis direction ,
The decoding block to be corrected of the image correction unit includes a reference screen target block corresponding to the correction target block on the inter-screen prediction decoding reference screen used for inter-screen predictive decoding, the correction target decoding block, The average value of the sum of absolute differences of the pixel values from the reference screen target block is equal to or less than a preset threshold value, and there is a correlation.
The image correcting unit calculates the difference between the average value of the pixel values of the pixels of the reference screen target block and the average value of the pixel values of the pixels of the correction block to be corrected and the reference screen target block. A value obtained by multiplying the coefficient corresponding to the average value of the sum of absolute differences of the pixel values with the value is added to each pixel of the decoding block as the drift error, and the image correction of the decoding block to be corrected is performed. An image decoding apparatus characterized by that. An I-slice consisting of a block according to an intra-screen prediction encoding mode consisting of a predetermined number of pixels encoded in one intra-screen prediction encoding mode as an encoding unit and an intra-screen prediction encoding as an encoding unit A block in the intra prediction encoding mode consisting of the predetermined number of pixels, and a block in the inter prediction encoding mode consisting of the predetermined number of pixels encoded in the inter prediction encoding as a coding unit. An image decoding apparatus for decoding encoded image data consisting of P and B slices consisting of:
The pixel of the block according to the intra prediction encoding mode in the I, P, and B slices is used as the intra prediction prediction target pixel, and the pixel of another block adjacent to the block in the same screen as the block is referred to Other pixels preceding or following the screen including the block, using the pixel of the block in the inter-frame predictive coding mode in the P and B slices as the inter-screen decoding target pixel A decoding unit that performs decoding using a pixel of another block at a position corresponding to the block on the screen as a reference pixel;
An image correction unit to which a decoded block of decoded pixels of the intra-screen predictive decoding target pixel from the decoding unit is supplied;
An intra-screen prediction unit that generates the reference pixel for decoding the intra-screen decoding target pixel from the image data of the decoded block output from the image correction unit;
An image reduction unit for reducing the image size of the image data by the decoding block output from the image correction unit and the decoding block output by the decoding pixel of the inter-screen decoding target pixel output from the decoding unit;
A storage unit for holding image data in which the image size is reduced;
An image expansion unit that expands the image data read from the storage unit to the original image size;
A motion compensation unit that creates the reference pixel for decoding the inter-picture prediction decoding target pixel from the image data from the image decompression unit;
With
The image correcting unit belongs to the P and B slices of the decoded block in the intra prediction encoding mode supplied from the decoding unit, and the pixel of the decoding block in the inter prediction encoding mode is a reference pixel. The decoding block in the intra prediction encoding mode decoded as a decoding block to be corrected is decoded, and the decoding block in the inter prediction encoding mode is decoded as a reference pixel for the correction decoding block Perform image correction to remove drift errors in the time axis direction,
The decoding block to be corrected of the image correction unit includes an adjacent block temporally preceding the decoding block to be corrected, and an average sum of absolute differences of pixel values of the decoding block to be corrected and the adjacent block. A decoding block whose value is equal to or less than a preset threshold value, which is correlated, and whose activity is equal to or less than a predetermined threshold value and image quality degradation is conspicuous;
The image correction unit calculates a pixel value between the decoding block to be corrected and the adjacent block based on a difference between an average value of pixel values of the pixels of the adjacent block and an average value of pixel values of the pixels of the decoding block to be corrected. A value obtained by multiplying a coefficient corresponding to an average value of the sum of absolute differences is added to each pixel of the decoding block as the drift error to perform the image correction of the decoding block to be corrected. Image decoding device. An I-slice consisting of a block according to an intra-screen prediction encoding mode consisting of a predetermined number of pixels encoded in one intra-screen prediction encoding mode as an encoding unit and an intra-screen prediction encoding as an encoding unit A block in the intra prediction encoding mode consisting of the predetermined number of pixels, and a block in the inter prediction encoding mode consisting of the predetermined number of pixels encoded in the inter prediction encoding as a coding unit. An image decoding method for decoding encoded image data consisting of P and B slices consisting of :
The pixel of the block according to the intra prediction encoding mode in the I, P, and B slices is used as the intra prediction prediction target pixel, and the pixel of another block adjacent to the block in the same screen as the block is referred to Other pixels preceding or following the screen including the block, using the pixel of the block in the inter-frame predictive coding mode in the P and B slices as the inter-screen decoding target pixel Decoding using a pixel in another block at a position corresponding to the block on the screen as a reference pixel ;
Of the decoded blocks of the block in the intra prediction encoding mode that has been decoded, the pixels belonging to the P and B slices and decoded in the inter prediction encoding mode in the inter picture prediction encoding mode are decoded as reference pixels. The decoding block in the intra prediction encoding mode is set as a correction target decoding block, and the decoding target decoding block is decoded in the time axis direction by decoding the decoding block pixel in the inter prediction encoding mode as a reference pixel. Perform image correction to remove drift error,
Decoded block subjected to decoding process of block by intra prediction encoding mode, decoded block of image block corrected by intra prediction encoding mode, and decoding block decoded by inter prediction encoding mode Generating the reference pixel for decoding the intra-screen decoding target pixel from the image data of the block;
The decoded block of the block encoded by the intra prediction encoding or the decoded block of the block encoded by the inter prediction encoding is reduced in the screen size and stored in the storage unit. Reference is made to store the decoded block read from the storage unit to the original image size, and decode the pixel of the decoded block expanded to the original image size to decode the pixel of the block in the inter-screen predictive coding mode Pixels,
The decoding block to be corrected for the image correction includes a reference screen target block corresponding to the correction target block in the inter-screen prediction decoding reference screen used for inter-screen predictive decoding, the decoding block to be corrected, and the reference The average value of the sum of absolute differences of pixel values from the screen target block is equal to or less than a preset threshold value, and there is a correlation.
The difference between the average value of the pixel values of the pixels of the reference screen target block and the average value of the pixel values of the pixels of the decoding block to be corrected is the difference between the pixel values of the decoding block to be corrected and the reference screen target block. A value obtained by multiplying a coefficient corresponding to an average value of the sum of absolute values is added to each pixel of the decoding block as the drift error, and the image correction of the decoding block to be corrected is performed. Decryption method. An I-slice consisting of a block according to an intra-screen prediction encoding mode consisting of a predetermined number of pixels encoded in one intra-screen prediction encoding mode as an encoding unit and an intra-screen prediction encoding as an encoding unit A block in the intra prediction encoding mode consisting of the predetermined number of pixels, and a block in the inter prediction encoding mode consisting of the predetermined number of pixels encoded in the inter prediction encoding as a coding unit. An image decoding method for decoding encoded image data consisting of P and B slices consisting of :
The pixel of the block according to the intra prediction encoding mode in the I, P, and B slices is used as the intra prediction prediction target pixel, and the pixel of another block adjacent to the block in the same screen as the block is referred to Other pixels preceding or following the screen including the block, using the pixel of the block in the inter-frame predictive coding mode in the P and B slices as the inter-screen decoding target pixel Decoding using a pixel in another block at a position corresponding to the block on the screen as a reference pixel ;
Of the decoded blocks of the block in the intra prediction encoding mode that has been decoded, the pixels belonging to the P and B slices and decoded in the inter prediction encoding mode in the inter picture prediction encoding mode are decoded as reference pixels. The decoding block in the intra prediction encoding mode is set as a correction target decoding block, and the decoding target decoding block is decoded in the time axis direction by decoding the decoding block pixel in the inter prediction encoding mode as a reference pixel. Perform image correction to remove drift error,
Decoded block subjected to decoding process of block by intra prediction encoding mode, decoded block of image block corrected by intra prediction encoding mode, and decoding block decoded by inter prediction encoding mode Generating the reference pixel for decoding the intra-screen decoding target pixel from the image data of the block;
The decoded block of the block encoded by the intra prediction encoding or the decoded block of the block encoded by the inter prediction encoding is reduced in the screen size and stored in the storage unit. Reference is made to store the decoded block read from the storage unit to the original image size, and decode the pixel of the decoded block expanded to the original image size to decode the pixel of the block in the inter-screen predictive coding mode Pixels,
The decoding block to be corrected in the image correction includes an adjacent block temporally preceding the decoding block to be corrected, and an average value of sums of absolute differences between pixel values of the decoding block to be corrected and the adjacent block. Is a decoding block that is less than or equal to a preset threshold, has a correlation, and activity is less than or equal to a predetermined threshold and image quality degradation is conspicuous,
The difference between the average value of the pixel values of the pixels of the adjacent block and the average value of the pixel values of the pixels of the decoding block to be corrected is the sum of absolute differences of the pixel values of the decoding block to be corrected and the adjacent block. An image decoding method , wherein a value obtained by multiplying a coefficient corresponding to an average value is added to each pixel of the decoding block as the drift error to perform the image correction of the decoding block to be corrected .

Images