JP2015092705A - Dynamic image decoding method and image coding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technology of reducing information amount for describing prediction information of a macro block.SOLUTION: To lower a compression ratio of image coding accompanying magnification/reduction of a block that becomes a coding unit, part of a predicted image that is predicted in a size larger than a macro block size to be coded is used as a predicted image of a macro block to be coded.

Description

  The present invention relates to a moving image signal encoding technique.

  In a moving picture coding system represented by ITU-T H.264, coding is performed by dividing the entire screen into coding units called macroblocks of 16 pixels × 16 pixels.

  In H.264, the prediction value of the pixel value in the target macroblock is determined for the macroblock to be encoded using surrounding pixels and previous and subsequent pictures, and the prediction error between the encoding target pixel and the prediction value is determined. Entropy encoding.

  When predicting the pixel values in the macroblock, intra prediction predicted from neighboring pixels and inter prediction predicted from pixels in the previous and subsequent pictures are selected for each macroblock according to the pattern in the macroblock. can do. Moreover, it can divide | segment into a prediction block smaller than 16 pixels x 16 pixels, and can perform prediction.

  For example, as shown in FIG. 2, in the H.264 intra prediction, a macro block of 16 pixels × 16 pixels is divided into 16 4-pixel × 4-pixel prediction blocks, and each prediction block is divided into those shown in FIG. By copying neighboring pixels in nine directions indicated by indexes 0 to 8, pixels in the prediction block can be predicted. In addition, as shown in FIG. 3, the prediction is performed by copying the surrounding pixels in the four types of directions indicated by indexes 0 to 3 in the prediction block of 16 pixels × 16 pixels without dividing the macroblock. Pixels in the block can be predicted.

  Similarly, with respect to H.264 inter prediction, a motion vector can be set for each prediction block by dividing the macroblock into smaller prediction blocks. For example, as shown in FIG. 4, when motion from a past picture is predicted, 16 pixels × 16 pixels, 16 pixels × 8 pixels, 8 pixels × 16 pixels, 8 pixels × 8 pixels (in this case, each individual prediction The block can be divided into prediction blocks of 8 pixels × 4 pixels, 4 pixels × 8 pixels, 4 pixels × 4 pixels), and different motion vectors can be set for the prediction blocks.

  As described above, by dividing the interior of the macroblock into prediction blocks and performing prediction individually, prediction accuracy is improved when there is a boundary between different patterns in the macroblock, and the compression rate is improved.

  However, in the conventional techniques represented by H.264, the size of the macroblock is limited to 16 pixels × 16 pixels, and it cannot be predicted in a unit larger or smaller than this. .

  In addition, since selection between intra prediction and inter prediction can be set only in units of macroblocks, selection can be made only in units smaller than 16 pixels × 16 pixels.

  In view of such a problem, Patent Document 1 can divide a block of 16 pixels × 16 pixels into any of 8 pixels × 8 pixels, 4 pixels × 4 pixels, and 2 pixels × 2 pixels according to a quadtree structure. The prediction mode can be changed according to these block sizes (Patent Document 1).

Special table 2007-503784

  In the moving picture coding technique described in Patent Document 1, when a coding unit block is divided, a prediction process is performed using the divided macroblock as a coding unit. Therefore, when the number of divided macroblocks having a quadtree structure increases, the code amount of prediction information increases correspondingly, and the compression rate decreases.

  The object of the present invention has been made in view of such circumstances, and provides a technique for reducing the amount of information for describing prediction information of macroblocks.

  Before describing the means for achieving the above objectives, terms are defined. In this specification, a block having a variable block size for which a prediction mode can be selected is referred to as a CU (Coding Unit) in order to distinguish it from a macroblock of a conventional technique (such as H264AVC).

  As a means for achieving the above-mentioned problem, when a prediction process is performed on a certain CU on the encoding side, a prediction image of a higher-order CU (hereinafter referred to as a parent CU) that is larger than the target CU is encoded. This is achieved by making it possible to select whether to use a part of the code as it is or to perform a prediction process for each encoding target CU.

  Then, the flag information indicating which one is selected is stored in the encoded stream, and the decoding side reads the flag information, thereby performing the prediction process on a certain CU to be encoded. This is achieved by selecting whether to use a part of the predicted image of a higher-order CU larger than that (hereinafter referred to as a parent CU) as it is or to perform prediction processing on each of the encoding target CUs. .

  For example, in the prior art, when the encoding target CU is divided into four CU1 to CU4, but only the prediction error of CU1 is small and the prediction accuracy is low for CU2 to 4, a prediction image of CU is generated using the prediction result of CU, An image that is a part of the predicted image of CU and corresponds to the region of CU1 is extracted as a predicted image. By doing so, information on prediction processing for the encoding target CUs 2 to 4 becomes unnecessary, and thus the amount of information can be reduced.

  In the conventional technique, when encoding using the encoding target CU, since the encoded data of the parent CU was not generated, even if the image only needs to be predicted separately for one encoding target CU, Forecasted by individual CU units. However, as described above, if a part of the predicted image of the upper CU is used, the amount of information describing the prediction process of the CU can be reduced and the compression rate can be improved.

  According to the present invention, in an image encoding / decoding method that employs a variable CU having a plurality of prediction unit block sizes, the compression rate is improved by reducing the amount of information describing the prediction process of the CU. Can be made.

It is a figure which shows schematic structure of the image coding apparatus by Example 1. FIG. It is a figure for demonstrating an example of the intra prediction process of a prior art. It is a figure for demonstrating an example of the intra prediction process of a prior art. It is a figure for demonstrating an example of the inter prediction process of a prior art. It is a figure for demonstrating the concept of CU division | segmentation. It is a figure for demonstrating an example of CU division | segmentation of a quadtree structure. It is a figure for demonstrating an example of the syntax in the encoding stream by CU division | segmentation of a prior art. It is a figure for demonstrating an example in which this invention becomes effective. It is a figure for demonstrating an example of CU division | segmentation by Example 1. FIG. FIG. 6 is a diagram for describing an example of syntax in an encoded stream by CU partitioning according to the first embodiment. It is a figure for demonstrating an example of the synthesis | combination of the prediction image at the time of CU division | segmentation by Example 1. FIG. It is a figure for demonstrating another example of the synthesis | combination of the prediction image at the time of CU division | segmentation by Example 1. FIG. FIG. 10 is a diagram for describing processing in intra prediction in the processing for synthesizing a predicted image at the time of CU division according to the first embodiment. It is a figure which shows schematic structure of the prediction mode determination part by Example 1. FIG. It is a figure which shows schematic structure of the image decoding apparatus by Example 1. FIG. It is a figure which shows schematic structure of the prediction selection part by Example 1. FIG.

  The present invention uses a prediction image of a parent CU before division when performing encoding with expansion or reduction of a coding unit block (hereinafter referred to as CU, Coding Unit). By omitting the prediction processing of the divided CUs, the amount of prediction information is reduced.

  Embodiments will be described below with reference to the accompanying drawings. However, it should be noted that the present embodiment is merely an example for realizing the present invention and does not limit the technical scope of the present invention. In each drawing, the same reference numerals are assigned to common components.

<Configuration of Image Encoding Device>
FIG. 1 is a diagram illustrating a schematic configuration of an image encoding device according to the first embodiment.

  In FIG. 1, the image encoding device includes a CU dividing unit 100 that determines a CU size, a difference unit 102 that generates a prediction difference image between a prediction image stored in a prediction image storage unit 107 and an input image 114, and the prediction It has a transforming unit 102 that performs orthogonal transform such as DCT on the difference image, a quantizing unit 103 that quantizes the transformed signal, and a variable length coding unit 104 that encodes the quantized signal, and an encoded stream 115 Is output.

  The moving picture coding apparatus according to the present embodiment has two types of prediction processing systems in order to generate the predicted image. The first system is based on inter prediction, and in order to obtain a reference image for the next input image, an inverse quantization unit 109 that inversely quantizes the quantized signal output from the quantization unit 103, and an inverse quantization signal Inverse transform unit 108 that obtains a prediction difference image by inverse transform, adder 111 that adds the prediction difference image after the inverse transform and the prediction image from prediction image storage unit 107, and a reference obtained by removing block noise from the added image A deblocking processing unit 112 for obtaining an image is included. Then, it includes a reference image storage unit 113 that stores the obtained reference image, and an inter prediction unit 106 that performs motion prediction between the reference image and the input image 114. The second system is based on intra prediction, and has an intra prediction unit 105 that performs intra prediction from the input image 114.

  As will be described later, the processing of the prediction mode determination unit 110 is the most predicted using the above two prediction processing systems, that is, the inter prediction image from the inter prediction unit 106 and the intra prediction image from the intra prediction unit 105. A prediction process that is estimated to be highly efficient is determined. Here, as an index of the prediction efficiency, for example, prediction error energy and the like can be mentioned, but in addition, a prediction image (that is, a prediction image (that is, inter-screen prediction or intra-screen prediction) in consideration of the similarity with the prediction method of the neighboring CU (inter-screen prediction or intra-screen prediction) (Prediction method) may be selected.

  A predicted image obtained by the determined prediction method is stored in the predicted image storage unit 113, and is used to generate a predicted difference image from the input image 114. Information regarding the prediction mode (that is, inter prediction or intra prediction, and the size of the prediction unit block in each case) selected by the prediction mode determination unit 110 is sent to the variable length encoding unit 104, and the encoded stream 115 Stored in part.

  In the present embodiment, the prediction process determined by the prediction mode determination unit 110 is characterized. However, since the CU division pattern is related to the determination of the prediction process, the processing contents of the CU division unit will be described below. .

<Processing content of CU partitioning unit (encoding side)>
Hereinafter, the processing contents of the CU dividing unit 100 will be described in detail with reference to the drawings.

FIG. 5 is a diagram for explaining the concept of the CU. In this embodiment, a coding processing unit block corresponding to a macroblock of the prior art is referred to as CU (Coding Unit). In this embodiment, the following properties are assumed for the CU. However, the application of this embodiment is not limited to this assumption.
(1) The CU is a square. (2) The maximum size and the minimum size of the CU are described in the encoded stream or defined as a standard. (3) The maximum CU is a child CU by a quadtree structure. In FIG. 5, where the division is hierarchically divided into four, the maximum size CU is denoted as LCU (Largest Coding Unit), and the size (the number of pixels in the vertical or horizontal direction of the LCU) is denoted as LCU size. In this embodiment, the LCU size is assumed to be a power of 2. However, the application of the present embodiment is not limited to being a power of 2.

  As shown in FIG. 5, one picture is divided in units of LCUs. A group of consecutive LCUs is defined as a slice. This concept corresponds to a prior art macroblock. Each LCU is hierarchically divided into four by a quadtree structure.

FIG. 6 is a diagram illustrating an example of CU partitioning configured by a quadtree structure. As shown in the figure, the LCU is divided into four CU ₀ , CU ₁ , CU ₂ , and CU ₃ . CU ₀ is not divided and is finally determined as CU. CU ₁ is divided into CU ₁₀ , CU ₁₁ , CU ₁₂ , and CU ₁₃ , CU ₂ is divided into CU ₂₀ , CU ₂₁ , CU ₂₂ , and CU ₂₃ , and CU ₃ is divided into four parts of CU ₃₀ , CU ₃₁ , CU ₃₂ , and CU _33. Has been. Of these, CU ₁₁ is further CU ₁₁₀ , CU ₁₁₁ , CU ₁₁₂ , CU ₁₁₃ , CU ₁₂ is CU ₁₂₀ , CU ₁₂₁ , CU ₁₂₂ , CU ₁₂₃ , CU ₃₀ is CU ₃₀₀ , CU ₃₀₁ , CU ₃₀₂ , CU They are respectively divided into four to _303, and the other CU are determined finally as CU. In this way, the LCU can be divided into four hierarchically, and the division can be performed until the size of the CU becomes the minimum size. In this specification, it referred to CU _10, CU _11, CU _12, CU ₁₃ obtained by dividing the CU ₁ and children CU of CU _1. Conversely, CU ₁ is described as the parent CU of CU ₁₀ , CU ₁₁ , CU ₁₂ , and CU ₁₃ .

  Note that the CU indicates a coding unit. Strictly speaking, prediction processing and conversion processing are performed for each CU. However, when it is described as a parent CU in this specification, it is necessary for this CU. Note that only the prediction process is performed according to the above, and the conversion process is not performed.

  In the case of the above quadtree structure, if the ratio of the maximum size / minimum size is 2 ^ N (2 to the Nth power), the flag indicating whether or not to divide individual CUs as in the prior art is 1 bit. The division pattern can be expressed by notation.

  An example of the syntax of the CU coded stream according to the prior art will be described with reference to FIG. In the figure, the function coding_unit () indicates the encoding syntax of the CU having the pixel position (x0, y0) and the size of currCUSize. Note that PicWidth is the picture width (number of pixels), PicHeight is the picture height (number of pixels), and MinCUSize is the minimum size of the CU.

  split_flag is a 1-bit flag indicating whether the current CU is divided into four (1) or not (0) (L700).

When split_flag is 1, the current CU is divided into four. In this case, the split CU size splitCUSize is 1/2 of the current CU size currCUSize, and the horizontal division position x1 and the vertical division position y1 are x1 = x0 + splitCUSize, y1 = y0 + splitCUSize, respectively. (L702). The four divided CUs (CU _{0 to} CU ₃ ) are stored by recursively calling coding_unit () (L703 to L706). Even in each of the four divided CUs, whether or not to further divide is specified by split_flag. Such a recursive call is performed as long as the CU size is equal to or greater than MinCUSize.

  If split_flag is 0, this CU is determined as a coding unit, and information on prediction processing (function prediction_unit ()) (L707), which is the main processing of coding, and orthogonal transformation information on prediction errors ( The function transform_unit ()) (L708) is stored. In the present specification, the orthogonal transform process is not directly related to the present invention, and is therefore omitted.

  As information of prediction processing (prediction_unit ()) in L707, for example, an identifier of the intra prediction or inter prediction, information indicating the prediction direction in the case of intra prediction (see FIG. 2 and FIG. 3), and inter prediction In some cases, division information and motion vector information (see FIG. 4) inside the CU are stored. However, the present invention is not limited to the prediction processing method and the content of the information.

  As the CU is divided more finely, the prediction process can be performed with a smaller size. However, since the prediction information is required for the number of divided CUs, the code amount increases.

  Therefore, in this embodiment, the prediction mode determination unit 110 includes the parent CU prediction unit 1400 to reduce the amount of prediction information when the number of CU divisions increases. Hereinafter, processing contents in the prediction mode determination unit 110 will be described.

<Processing content of prediction mode determination unit>
Next, processing contents of the prediction mode determination unit 110 according to the first embodiment will be described.
(1) Overview of Overall Processing FIG. 14 is a configuration diagram of the prediction mode determination unit 110.

  The prediction mode determination unit 110 includes a parent CU prediction unit 1400 and a prediction cost comparison unit 1401. As will be described later, the parent CU prediction unit 1400 stores the prediction image of the parent CU of the encoding target CU, and calculates the prediction cost when the prediction process of the current CU is replaced with a part of the prediction image of the parent CU. calculate.

The prediction cost comparison unit 1401 compares a plurality of intra prediction processes in a plurality of CU sizes, an inter prediction image, and a prediction cost from the parent CU prediction unit 1400, and determines a prediction process that minimizes the prediction cost. The predicted image obtained by this prediction process is stored in the predicted image storage unit 107. In the present invention, the calculation method of the prediction cost is not limited, but may be defined by, for example, the sum of absolute differences between the input image 114 and the prediction image and the weighted sum of the total bit amount required for the prediction information. According to this definition, the closer the predicted image is to the input image, and the smaller the amount of bits required for the prediction information, the higher the prediction efficiency of the encoding process.
(2) Details of parent CU prediction unit In the parent CU prediction unit 1400, a prediction image in the parent CU of the encoding target CU is generated and stored in advance, and the prediction process of the encoding target CU is performed on the prediction image of the parent CU. Calculate the estimated cost when the part is replaced. A scene in which such replacement of a parent CU with a predicted image is effective will be described with reference to FIG.

  As shown in FIG. 8, the LCU (X) to be encoded of a certain picture to be coded and the area Y of the picture immediately before it have almost the same background, and there is an object that moves only inside Assume a case. In this case, as the prediction processing of the LCU (X), it is estimated that prediction with high accuracy is performed when the prediction processing is divided into the prediction processing for the entire background and the object portion having the internal motion. Therefore, the LCU (X) may be divided into a background CU and a moving object CU, and individual prediction processing may be specified for each CU.

  However, in the case of CU partitioning using the quadtree structure as described above, the number of segmented CUs may increase depending on the position of the moving object in the LCU, and as a result, prediction information may increase. Such a case will be described with reference to FIG.

  As shown by (A) in FIG. 9, when there is an object moving near the center of the LCU, it is considered that the background and the object part are divided so as to be included in different CUs. First, by dividing the LCU of FIG. 10A once, four CUs (1 to 4) as shown in FIG. In FIG. 5B, since CU (1) to CU (4) include many objects and backgrounds, CU (1) to CU (4) are divided. This causes CU (A to D) from CU (1), CU (E to H) from CU (2), CU (I to L) from CU (3), and CU (3) CU (M ~ P) is created from each. Of these 16 CUs, CU (D), CU (G), CU (J), and CU (M) still contain both an object and a background, and thus further divide them. As a result, CU (D1-D4) from CU (D), CU (G1-G4) from CU (G), CU (J1-J4) from CU (J), and CU (M) CU (M1 to M4) are respectively created ((D) in the figure). Of these, CU (D4), CU (G3), CU (J2), and CU (M1) contain many objects only, and other CUs contain many backgrounds only. Therefore, for CU (D4), CU (G3), CU (J2), and CU (M1), perform prediction processing that takes into account object movement, and other CUs to perform prediction processing that takes into account background movement Thus, it is considered that highly accurate prediction processing can be realized.

  However, if the CU is divided finely as described above, it is necessary to store prediction processing information for all 24 CUs as shown in FIG. 4D, and the prediction processing information increases.

  Therefore, the prediction mode determination unit 110 according to the first embodiment does not necessarily store the prediction process information for all the individual CUs. Whether the prediction image obtained in advance by the prediction process of the parent CU is used as the prediction result. It is possible to select one of the prediction processes performed by individual CUs.

  The parent CU prediction unit 1400 calculates a prediction cost when replacement of the former, that is, the prediction image of the parent CU is selected, and passes the prediction cost result to the prediction cost comparison unit 1401. The prediction cost comparison unit 1401 compares the latter normal prediction processing, that is, the prediction cost of normal inter prediction or intra prediction, with the prediction cost from the former parent CU prediction unit 1400, and makes a prediction with a low prediction cost. Select a process.

Hereinafter, an example of the CU syntax of the encoded stream according to the first embodiment will be described.
(3) Example of CU Syntax An example of the CU syntax of the encoded stream according to the first embodiment will be described with reference to FIG.

  As a feature different from the conventional CU syntax (Fig. 7), when split_flag = 1, that is, when the current CU is divided into four child CUs, it has 1-bit parent_pred_unit_flag, Whether to store prediction processing information (1) or not (0) is specified (L1000), and when parent_pred_unit_flag == 1, prediction processing information is stored (in the figure, parent_prediction_unit () function) (L1001).

  Also, when split_flag == 0, that is, when the current CU is determined by the current size without being divided and becomes a CU to be encoded, it has a 1-bit parent_pred_flag and is a prediction image of the parent CU, that is, parent_prediction_unit Whether to replace a part of the prediction image obtained by the prediction process specified in (1) or to perform another prediction process (0) is specified (L1002).

  When parent_pred_flag == 0, information of another prediction process is stored in the prediction_unit () function.

  When parent_pred_flag == 1, among the predicted images of the parent CU, an image at a position corresponding to the position of the encoding target CU is set as a predicted image of the encoding target CU. Information on prediction processing in the current CU is unnecessary. Therefore, the more CUs that have parent_pred_flag == 1, the more information can be expected to be reduced.

  Hereinafter, a specific example of processing in the prediction mode determination unit 110 and CU syntax will be described with reference to FIG.

  The division pattern of the CU is the same as that in FIG. First, the parent CU prediction unit 1400 determines the prediction process based on the LCU size as shown in FIG. The method of determining the prediction process is not limited in the present invention. For example, in order to describe the difference between the prediction image obtained as a result of performing a plurality of intra predictions and inter predictions, and the input image 114, and the prediction process The cost value defined by the weighted sum of the bit amounts of the prediction information may be calculated, and the prediction process that minimizes the cost value may be determined. The prediction image obtained by this prediction processing is stored in the parent CU prediction unit 1400 as a prediction image of the parent CU. In the LCU syntax, parent_pred_unit_flag = 1 is set, and the determined prediction processing information is stored in parent_prediction_unit ().

  Whether or not the prediction image of the parent CU (LCU) is used as a prediction result is determined for all CUs obtained by dividing the LCU as in FIG. 9D. In the present invention, although this determination process is not limited, as an example, the prediction cost when the prediction image of the parent CU is used as the prediction result, and the case where intra prediction is performed individually or when inter prediction is performed The prediction cost values obtained by a plurality of prediction processes may be compared by the prediction cost comparison unit 1401 and a prediction process having a small prediction cost value may be selected.

With this prediction process selection process,
(1) CU using predicted image of parent CU as prediction result:
CU (A), CU (B), CU (C), CU (D1), CU (D2), CU (D3), CU (E), CU (F), CU (G1), CU (G2), CU (G4), CU (H), CU (I), CU (J1), CU (J3), CU (J4), CU (M2), CU (M3), CU (M4), CU (N), CU (O), CU (P)
(2) CU performing another prediction process:
CU (D4), CU (G3), CU (J2), CU (M1)
Is determined.

  In this case, parent_pred_flag = 1 is set for the CU of (1), and the parent CU prediction unit 1400 uses the predicted image of the location corresponding to the position of each CU from the predicted image of the parent CU (LCU). CU prediction image.

  For the CU of (2), parent_pred_flag = 0 is set, and prediction processing information is stored in parent_prediction_unit () for each CU.

  As described above, since the amount of information of the prediction processing for the CU of (1) can be reduced as compared with the prior art, an improvement in compression rate can be expected.

  In the embodiment, the number of parent CUs is not necessarily limited to one. As shown in Fig. 12, LCU and CU (D) ("D" in Fig. 9 (C), that is, the parent CU of CU (D1) to CU (D3)) are specified as parent_pred_unit_flag = 1 When the prediction process is stored in parent_prediction_unit (), the result of overwriting only the part corresponding to the position of CU (D) in the LCU prediction image from the inclusion relationship between LCU and CU (D) The prediction image of the parent CU. The predicted image of the parent CU is applied to the CU (1) (see FIG. 12).

  In the case as shown in FIG. 12, the information amount of the prediction process is increased by an amount corresponding to parent_prediction_unit () of CU (D) as compared with the case of FIG. However, since the prediction process with higher accuracy can be selected separately from the LCU for the location of CU (D), the prediction accuracy is improved, and the prediction difference information is reduced, so that the compression rate can be expected to be improved.

  In this embodiment, for each CU, it is possible to specify whether the prediction process is performed individually or the prediction image of the parent CU is used as it is, and the prediction process of the child CU and the prediction process of the parent CU The combination of methods is not limited, and any combination of inter prediction and intra prediction can be applied. In inter prediction, various prediction methods, such as forward prediction using only a temporally previous picture as a reference picture and bidirectional prediction using temporally forward and backward pictures, can be applied.

  However, as shown in FIG. 13, when intra prediction is performed with CU (D), and when encoded images (before deblocking) around encoded images around CU (D) are used. The surrounding CU (A), CU (B), and CU (C) encoding processing (but before deblocking processing) must be completed.

As described above, the prediction mode determination unit 110 in the image encoding device according to the present embodiment selects whether to use a prediction image of a parent CU or another prediction process for a prediction process of a certain CU. The prediction process information is stored in the encoded stream only when another prediction process is performed. Thereby, the compression rate can be improved by reducing the predicted information amount of the CU.
<Configuration of image decoding apparatus>
FIG. 15 is a diagram illustrating a schematic configuration of an image decoding device according to the embodiment. In FIG. 15, an image decoding apparatus receives an encoded stream 1500, a variable length decoding unit 1501 that decodes the encoded stream 1500, and a CU that divides the CU based on CU size information obtained by the variable length decoding unit 1501 A division unit 1502, an inverse quantization unit 1503 that inversely quantizes the transform-quantized prediction error image in the CU, an inverse transform unit 1504 that inversely transforms the obtained transformed prediction error image, An adder 1505 that adds the prediction image stored in the prediction image storage unit 1508 and the prediction error image output from the inverse conversion unit 1504, and a deblocking processing unit 1506 that performs a deblocking process on the addition result image. And an output image 1512 is output.

The moving picture decoding apparatus according to the present embodiment has two types of prediction processing systems in order to generate the predicted image. The first system is based on intra prediction, and includes an intra prediction unit 1507 that performs intra prediction using decoded CU image signals (before deblocking processing) that are sequentially stored in CU units. The second system is based on inter prediction, and uses a reference image storage unit 1510 for storing an output image, a reference image stored in the reference image storage unit 1510, and a motion vector decoded by the variable length decoding unit 1501. An inter prediction unit 1511 is provided that performs motion compensation and obtains an inter prediction image. The prediction selection unit 1509 generates a prediction image according to the prediction processing information of the CU decoded by the variable length decoding unit 1501, and stores the prediction image in the prediction image storage unit 1508.
<Processing content of prediction selection unit (decoding side)>
The processing contents of the prediction selection unit 1509 on the image decoding side will be described below with reference to the drawings.

  FIG. 16 is a diagram showing an internal configuration of the prediction selection unit 1509. As shown in FIG. The prediction switching unit 1601 switches prediction processing based on the prediction processing information of each CU decoded by the variable length decoding unit 1501, generates a prediction image, and stores the prediction image in the prediction image storage unit 1508.

  Specific examples of the CU prediction process information include information on parent_pred_unit_flag, parent_prediction_unit (), parent_pred_flag, and prediction_unit () in FIG. The meaning of the syntax of the encoded stream in FIG. 10 and the processing contents of the parent CU prediction unit 1600 corresponding to these syntaxes are the same as those of the parent CU prediction unit 1400 in the encoding device, and thus description thereof is omitted.

  As described above, the prediction selection unit 1509 in the image decoding apparatus according to the present embodiment can use the prediction image of the parent CU as the prediction result of the encoding target CU according to the prediction processing information of the CU of the encoded stream. As a result, the prediction processing information of the encoding target CU in the encoded stream can be reduced, so that the compression rate can be improved.

  As described above, according to the present invention, it is possible to select whether to use a prediction image of the parent CU or perform another prediction process as the prediction process of the encoding target CU. If it is selected to use the predicted image of the parent CU, the image encoding apparatus performs the same parent CU prediction process in the image decoding apparatus without sending the prediction process information of the encoding target CU. Thus, a prediction image of the encoding target CU can be generated, and the information amount of the prediction process can be reduced.

  Note that the present invention can also be realized by a program code of software that implements the functions of the embodiments. In this case, a storage medium in which the program code is recorded is provided to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing it constitute the present invention. As a storage medium for supplying such program code, for example, a flexible disk, CD-ROM, DVD-ROM, hard disk, optical disk, magneto-optical disk, CD-R, magnetic tape, nonvolatile memory card, ROM Etc. are used.

  Also, based on the instruction of the program code, an OS (operating system) running on the computer performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing. May be. Further, after the program code read from the storage medium is written in the memory on the computer, the computer CPU or the like performs part or all of the actual processing based on the instruction of the program code. Thus, the functions of the above-described embodiments may be realized.

  Further, by distributing the program code of the software that realizes the functions of the embodiment via a network, the program code is stored in a storage means such as a hard disk or memory of a system or apparatus, or a storage medium such as a CD-RW or CD-R And the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage means or the storage medium when used.

100 ... CU division
110 ... Prediction mode determination unit
105 ... Intra prediction unit
106… Inter prediction part
102: Conversion unit
103 ... Quantization part
104 ... Variable length coding unit
1400 ... Parent CU prediction part
1401 ... Forecast cost comparison section
1501 ... Variable length decoding unit
1502… CU division
1503… Inverse quantization part
1504 ... Inverse converter
1507 ... Intra prediction unit
1511 ... Inter prediction section
1509 ... Prediction selection part
1600 ... Parent CU prediction part
1601 ... Prediction switching part

Claims (1)

Video decoding that variable-length-decodes the input encoded stream, dequantizes and inversely transforms the encoded stream into a prediction difference image, adds the prediction difference image and the prediction image, and outputs a moving image In the device
The encoded stream to be decoded includes both a first encoding unit and an encoding unit that is larger than the first encoding unit and includes an upper second encoding unit that includes the first encoding unit. If the encoded stream is encoded with
Generating a predicted image generated in the first coding unit and a predicted image generated in the second coding unit;
A moving picture decoding method, wherein a part of a predicted image generated in the second coding unit is used as the predicted image in the first coding unit.

Images