JP2005348008A - Moving picture coding method, moving picture coder, moving picture coding program and computer-readable recording medium with record of the program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for coding moving pictures by the intra-frame prediction which reduces a flicker quantity with restraining the coding efficiency from deteriorating. <P>SOLUTION: In selection of predicting methods for indicating an optimum prediction evaluation value, one of predicting methods utilizable for a coding target block in the present frame is selected and set as a tentative predicting method, while the prediction evaluation value of a predicting method utilized for the block at the same position in a coded frame just before the present frame as the coding target block is set as a prediction evaluation value for determining the prediction method. If the prediction evaluation value of the set tentative prediction method is approximately equal to that for determining the set predicting method, the predicting method utilized for the block at the same position in the coded frame just before the present frame is determined to be a prediction method to be utilized for the coding target block in the preset frame. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a moving image encoding method and apparatus for encoding a moving image by intra-frame prediction, a moving image encoding program used for realizing the moving image encoding method, and a computer-readable recording of the program. And a recording medium.

  In general, since the correlation between the pixel values of adjacent pixels in each frame of a moving image is large, reducing the amount of information in the frame using the correlation in this frame is effective for highly efficient compression coding of moving images. It is. In a moving image, since the correlation between pixel values is large between frames, the amount of information can be further reduced by using the correlation between frames.

  However, when encoding is performed using correlation between frames, an arbitrary frame cannot be decoded without referring to another frame at the time of decoding.

  Since encoded data obtained by an encoding method using only correlation within a frame (hereinafter, such an encoding method is referred to as intra-frame limited encoding) can be edited more easily, It is considered important in application.

  In H.264, which is an international standard video encoding system (see, for example, Non-Patent Document 1), an I frame that is a frame to be encoded using a correlation between pixels in a frame, and a correlation between frames. Since the P frame and the B frame, which are frames to be used for encoding, are defined, it is possible to perform encoding using the intra-frame and inter-frame correlation.

  In addition, when realizing intra-frame limited encoding with H.264, all frames of a moving image may be encoded with I frames.

  In H.264, intra-frame prediction is employed in order to increase the coding efficiency by using the correlation between pixel values in a frame. In encoding using intra-frame prediction, a pixel value to be encoded and a pixel value sequence in a block are predicted from adjacent pixel value sequences, and the prediction residual is encoded. If the correlation between the pixels in the frame is large, the average energy of the prediction residual is small, and the coding efficiency is improved.

  In the intra-frame prediction of H.264, two prediction modes are defined. Hereinafter, these two modes are referred to as a 4 × 4 prediction mode and a 16 × 16 prediction mode. H.264 specifies that each macroblock can use these two prediction modes adaptively.

  In the 4 × 4 prediction mode, the macroblock is divided into blocks each having 4 vertical pixels and 4 horizontal pixels (hereinafter, this block is described as 4 × 4 blocks), and prediction is performed in units of 4 × 4 blocks. I do. In the 16 × 16 prediction mode, prediction is performed in units of macroblocks. Specifically, as shown in FIG. 6, a pixel value in the current block is predicted using a decoded pixel value sequence around the current block. A plurality of prediction methods can be executed in any of the 4 × 4 and 16 × 16 prediction modes.

  As specific examples, prediction in the vertical direction (hereinafter referred to as vertical prediction), prediction in the horizontal direction (hereinafter referred to as horizontal prediction), and prediction using an average value (hereinafter referred to as average value prediction), A case where the 16 × 16 prediction mode is used will be described as an example.

  In vertical prediction, the top 16 pixels of the current block (which are already encoded) are decoded and used for prediction. Specifically, a predicted image is created by putting the decoded pixel values above all the pixels in each column of a 16 × 16 block. In the horizontal prediction, the left 16 pixels of the current block (which are already encoded) are decoded and used for prediction. In this case, a predicted image is created by inserting the decoded pixel values on the left of all the pixels in each row of the 16 × 16 block. Also, in the average value prediction, 32 pixels on the left and top of the current block (these are already encoded) are decoded, and these average values are put into all the pixels of the 16 × 16 block to predict the predicted image. Create

  In H.264, encoding is performed by performing orthogonal transformation, quantization, and entropy encoding on a prediction residual image obtained from a prediction image obtained by such a method.

  When intra-frame limited encoding is performed by intra-frame prediction, there is a problem that an artifact called flicker is likely to occur particularly under a low bit rate condition. Here, flicker is defined as a difference between a pixel value difference in the original image and a pixel value difference in the decoded image between pixels at the same position in the current frame and the previous frame.

  The occurrence of flicker means that, for example, a temporal change occurs in the decoded image even in a block where there is no temporal change between successive frames of the original image.

  Specifically, in Non-Patent Document 2 and Non-Patent Document 3 shown below, flicker is defined as one that is quantitatively evaluated by an amount S given by the following equation.

ここで、Ｉ_i,j （ｋ）は動画像中のｉ番目のフレーム内のｊ番目のブロックにおけるｋ個目の画素（ブロック内の画素数はＭで表している）を表し、Ｒ_i,j （ｋ）は復号された動画像中のｉ番目のフレーム内のｊ番目のブロックにおけるｋ個目の画素を表す。

Here, I _{i, j} (k) represents the k-th pixel (the number of pixels in the block is represented by M) in the j-th block in the i-th frame in the moving image, and R _{i, j} (k) represents the k-th pixel in the j-th block in the i-th frame in the decoded moving image.

  FDIF (i, j) represents the square error sum of each pixel between the j-th block in the i-th and i + 1-th frames in the moving image. That is, it is given by the following equation.

また、Σ_FDIF(i,j)<εは、ＦＤＩＦ（ｉ，ｊ）がεより小さいブロックに関して、各フレームについて和を取ることを表している。つまり、εは動画像中で時間的に一定の量より多く変化するブロックに関してはフリッカ量の評価から除外するための値である。このような除外を行うのは、動画像中で時間的に変化の小さい領域においてフリッカが主観的に目立つからである。以下では、動画像中で時間的に変化の小さい領域においてのフリッカを扱っていくこととする。

Further, Σ _{FDIF (i, j) <} ε represents that a sum is obtained for each frame with respect to a block in which FDIF (i, j) is smaller than ε. That is, ε is a value for excluding from the evaluation of the flicker amount a block that changes more than a certain amount in time in the moving image. Such exclusion is performed because flicker is conspicuously conspicuous in an area where the temporal change is small in the moving image. In the following, it is assumed that flicker is handled in an area in which a change in time is small in a moving image.

  In each frame, if the original image is encoded by performing orthogonal transformation and quantization without performing intra-frame prediction, the decoded image does not change in a block that does not change temporally between consecutive frames. Flicker does not occur.

  On the other hand, in encoding using intra-frame prediction, a predicted image may be greatly different even in a block where there is no temporal change between consecutive frames. In such a case, the residual image also has a large temporal change between consecutive frames. If the residual image that has been subjected to orthogonal transform and quantization is added to the predicted image during decoding, the temporal difference between successive frames Changes occur even in blocks where there is no change (flicker occurs).

  Focusing on these points, in Non-Patent Document 2 shown below, a residual image is obtained by subtracting an image obtained by performing orthogonal transform, quantization, inverse quantization, and inverse orthogonal transform on a predicted image from an original image. A method for encoding the residual image has been devised.

  When the quantization used for encoding is uniform quantization (quantization with a uniform quantization step size), the decoded image obtained by the technique of Non-Patent Document 2 is the original image. On the other hand, since it becomes equal to a decoded image obtained by performing orthogonal transform, quantization, inverse quantization, and inverse orthogonal transform, it is possible to eliminate flicker related to a stationary block and reduce flicker.

  Further, Non-Patent Document 3 shown below proposes a technique for further reducing flicker on the premise that the technique of Non-Patent Document 2 is used in combination.

In this method, according to the method of Non-Patent Document 2, the quantized value of the transform coefficient of the prediction residual is
L = Q (I _trans ) −Q (P _trans )
The fact that it can be expressed as

Here, I _trans represents an arbitrary transform coefficient obtained by performing orthogonal transform on a block in the original image, and P _trans is a transform coefficient obtained by performing orthogonal transform on a predicted image corresponding to the block (same as I _trans Frequency component). Q (A) represents the quantized value of value A.

In Non-Patent Document 3, the difference between the quantized value of the transform coefficient corresponding to the block at the same position in the previous frame and the quantized value of the current frame is 1, and the two representative values closest to I _trans are If the absolute value of the difference between the midpoint and I _trans is within a certain threshold, L is expressed as L ′
L ′ = Q (I ′ _trans ) −Q (P _trans )
Replace with However, Q (I ′ _trans ) is the quantized value of the transform coefficient of the block at the same position in the immediately preceding frame.

By doing so, the flicker amount can be reduced even for a block that is not a static region but has only a slight change.
ITU-T Rec.H.264 / ISO / IEC 11496-10, "Advanced Video Coding", Final Committee Draft, Document JVT-E022, September 2002 Xiaopeng Fan, Wen Gao, Yan Lu, Debin Zhao, ITU-T Rec.H.264 / ISO / IEC 11496-10, "Flicking Reduction in All Intra Frame Coding", Document JVT-E070, October 2002 Kazuhisa Iguchi, Makoto Ikeda, Shinichi Sakai, Masahide Naemura, "Intra-mode flicker reduction method in H.264 coding", FIT2003, J-040

  However, the method of Non-Patent Document 2 has a problem that the amount of calculation is enormous because it is necessary to perform orthogonal transformation and quantization on the predicted image every time a predicted image is created in the block to be encoded. . Moreover, the amount of computation increases with the number of available prediction methods and the block size.

When this increase amount is specifically described, when the block size is composed of N pixels both vertically and horizontally, the number of product-sum operations in a general orthogonal transform is N ⁴ , and the number of product-sum operations of quantization is the N ^2. From this, when the number of available prediction methods is Q, the number of product-sum operations that are increased by using the method of Non-Patent Document 2 is approximately Q (N ⁴ + N ² ) per block.

  As can be seen, the technique of Non-Patent Document 2 is not suitable for applications that require encoding in real time.

  Further, since the method of Non-Patent Document 3 is premised on the combined use of the method of Non-Patent Document 2, there is a problem similar to that described above.

  The present invention has been made in view of such circumstances, and has a configuration in which a moving image is encoded by intra-frame prediction, and while the flicker is reduced, the increase in the amount of calculation thereby is small, It is another object of the present invention to provide a new video encoding technique in which the increase in the amount of computation does not depend on the number of prediction methods and the block size.

[1] Basic Configuration of the Present Invention In order to achieve the above object, the moving picture coding apparatus of the present invention performs the process of coding a moving picture by intra-frame prediction. A prediction method setting unit that selects one of the prediction methods that can be used in the current encoding target block and sets the prediction method as a temporary prediction method, and (b) the prediction method set by the temporary prediction method setting unit Of the current frame based on the prediction evaluation value of the current frame and the prediction evaluation value when using the prediction method used in the block at the same position as the block to be encoded in the frame encoded immediately before the current frame A configuration is adopted in which a prediction method determination unit that determines a prediction method used in a block is provided.

  Here, the temporary prediction method setting unit selects one of the prediction methods that can be used in the current frame encoding target block according to some method. For example, a prediction method that shows an optimal prediction evaluation value In some cases, the selection method is selected based on the prediction evaluation value of the prediction method.

  The moving image encoding method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program. This computer program can be provided by being recorded on an appropriate recording medium, or can be connected to a network. The present invention is realized by operating on a control means such as a CPU installed when implementing the present invention.

[2] Specific Configuration of the Present Invention In order to realize the basic configuration described above, the moving image encoding apparatus of the present invention performs (i) a process of encoding a moving image by intra-frame prediction. And (b) a prediction evaluation value calculation unit that calculates a prediction evaluation value for each of the prediction methods that can be used in the current block encoding target block, and (b) a prediction evaluation value calculated by the prediction evaluation value calculation unit. A temporary prediction method setting unit that selects one of the prediction methods available in the encoding target block of the frame and sets the prediction method as a temporary prediction method; and (c) a frame encoded immediately before the current frame. A pre-frame prediction method evaluation value setting unit that sets a prediction evaluation value when a prediction method used in a block at the same position as the encoding target block is used as a prediction evaluation value for determining a prediction method; Forecasting method Prediction that determines a prediction method to be used in the current block encoding target block based on the prediction evaluation value of the prediction method set by the setting unit and the prediction evaluation value set by the previous frame prediction method evaluation value setting unit The method determining unit is provided.

When this configuration is adopted and a mechanism for narrowing down the prediction method for which the prediction evaluation value is to be calculated is not prepared, the video encoding device of the present invention uses all the prepared prediction methods. As a prediction method that can be used in the current frame encoding target block, the prediction evaluation value calculation unit is used in a block at the same position as the encoding target block in the frame encoded immediately before the current frame. Therefore, the previous frame prediction method evaluation value setting unit sets the calculated prediction evaluation value as the prediction evaluation value for determining the prediction method. To process.

On the other hand, when this mechanism is adopted and a mechanism for narrowing down the prediction method for which the prediction evaluation value is to be calculated is prepared, the video encoding device of the present invention uses the narrowed-down prediction method. As a prediction method that can be used in the current frame encoding target block, the prediction evaluation value calculation unit is used in a block at the same position as the encoding target block in the frame encoded immediately before the current frame. If the predicted evaluation value is calculated, the previous frame prediction method evaluation value setting unit may calculate the prediction evaluation value when the predicted evaluation value is calculated. It is set as a prediction evaluation value for determination, and when the prediction evaluation value is not calculated, it is calculated and processed so as to be set as a prediction evaluation value for determining a prediction method.

  The moving image encoding method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program. This computer program can be provided by being recorded on an appropriate recording medium, or can be connected to a network. The present invention is realized by operating on a control means such as a CPU installed when implementing the present invention.

[3] Operation of the present invention In the moving picture coding apparatus of the present invention configured as described above, for example, by selecting a prediction method that shows an optimal prediction evaluation value, it can be used in a current frame encoding target block. Prediction method selected as a temporary prediction method and used in a block at the same position as the current block in the frame encoded immediately before the current frame The prediction evaluation value when using is set as the prediction evaluation value for determining the prediction method.

  Then, based on the set prediction evaluation value of the provisional prediction method and the set prediction evaluation value for determining the prediction method, for example, by performing a threshold process on the ratio of the prediction evaluation values of the two prediction methods, It is determined whether or not the prediction evaluation values of the two prediction methods are substantially equal, and in the case where they are approximately equal, in the frame encoded immediately before the current frame, the same position as the current block is encoded. The prediction method used in the block is determined as the prediction method used in the current block encoding target block.

  As described above, in the present invention, the prediction evaluation value (some amount indicating prediction efficiency) related to the temporary prediction method determined by some method in the current block and the block at the same position in the previous frame are used. When the prediction evaluation value when the prediction method is used in the current block is almost equal, the prediction method used in the block at the same position in the previous frame is used for encoding the current block Take.

  More specifically, when the prediction efficiency of the prediction method used in the block at the same position in the previous frame is almost equal to the prediction efficiency of the prediction method having the maximum prediction efficiency in the current block Is configured to use the prediction method used in the block at the same position in the previous frame for encoding the current block.

  As described above, flicker can be reduced by using the same prediction method in a region where the block of the previous frame and the current block are substantially the same, including adjacent pixels used for prediction. Thus, from the viewpoint of reducing flicker, it is sufficient to adopt a configuration that always uses only one prediction method.

  However, according to this method, although the flicker reduction can be realized, there is a possibility that the encoding efficiency is greatly reduced.

  In contrast, the present invention employs a configuration in which the prediction method for the previous frame is used only when the decrease in prediction efficiency is small. Therefore, flicker can be reduced while suppressing the decrease in encoding efficiency. Can be realized.

  As described above, according to the present invention, the prediction efficiency of the prediction method used in the block at the same position in the previous frame is almost equal to the prediction efficiency of the prediction method having the maximum prediction efficiency in the current block. If they are equal, the prediction method of the previous frame is used for encoding the current block, so that the flicker amount can be reduced while suppressing a decrease in encoding efficiency.

  Further, the increase in the amount of calculation according to the present invention is that the number of product-sum operations per block is one, and this does not depend on the number of prediction methods or the block size. Therefore, the present invention is suitable for real-time encoding.

  Hereinafter, the present invention will be described in detail according to embodiments.

  Here, it is assumed that each macroblock performs three prediction methods: horizontal direction prediction, vertical direction prediction, and average value prediction. In addition, although prediction is described only with respect to the luminance component of the image, the present invention can be realized with a configuration similar to the following also regarding the color difference component.

  FIG. 1 illustrates an embodiment of a moving picture encoding apparatus 1 according to the present invention.

  As shown in this figure, the moving picture encoding apparatus 1 of the present invention includes an intra-frame prediction unit 10 that is configured to realize the present invention and determines a prediction method of a current macroblock to be encoded. The intra-frame prediction unit 10 includes an image information input unit 101, a temporary prediction method setting unit 102, a prediction method evaluation unit 103, a previous frame prediction method evaluation value setting unit 104, a prediction method memory 105, And a prediction method determination unit 106.

  The image information input unit 101 inputs an image of the current macroblock and a decoded image used for prediction.

  The temporary prediction method setting unit 102 obtains a temporary prediction method in the current macroblock based on the prediction evaluation value of the prediction method.

  The prediction method evaluation unit 103 calculates a prediction evaluation value of the prediction method in the current macroblock for the given prediction method.

  The previous frame prediction method evaluation value setting unit 104 sets a prediction evaluation value when the prediction method used in the previous frame is used in the current macroblock as a prediction evaluation value for determining a prediction method.

  The prediction method memory 105 records information on the prediction method used in the previous frame.

  The prediction method determination unit 106 compares the prediction evaluation value of the temporary prediction method set by the temporary prediction method setting unit 102 with the prediction evaluation value set by the previous frame prediction method evaluation value setting unit 104, and compares the comparison result. Based on the result, the prediction method used in the current macroblock is determined.

  FIG. 2 illustrates an example of a flowchart executed by the intra-frame prediction unit 10 configured as described above. Here, this flowchart shows the determination method of the prediction method for a macroblock having a certain frame in the moving image.

  Next, the processing of the present invention will be described in detail according to this flowchart. In the following description, for convenience of explanation, it is assumed that integer indexes of 0 for vertical prediction, 1 for horizontal prediction, and 2 for average prediction are associated with each other.

  When there is a request for determining a prediction method in the current macroblock, the intraframe prediction unit 10 firstly, in step 10, in step 10, the image of the current macroblock and the decoded image used for prediction are shown. (Processing of the image information input unit 101), and an initial value is substituted for the variable pred indicating the index of the prediction method, and an initial value is substituted for the variable minSAD into which the prediction evaluation value is substituted in a later process. (Process of provisional prediction method setting unit 102).

  In the present embodiment example, it is assumed that “0” is substituted as the initial value of pred and the upper limit value “131072” of the predicted evaluation value is substituted as the initial value of minSAD. It is assumed that the pixel value is represented by 8 bits and the macroblock is 16 pixels long and 16 pixels wide.

  Subsequently, in step 11, a prediction image is generated using a prediction method corresponding to the value substituted for pred (processing of the temporary prediction method setting unit 102).

  Subsequently, in step 12, a prediction evaluation value related to the prediction method corresponding to the value substituted for pred is obtained using the generated prediction image (processing of the prediction method evaluation unit 103).

  In this embodiment, SAD (Sum of Absolute Differences) given by the following equation is used as a predicted evaluation value. In the following, SAD is also used as a variable name.

ここで、ａｂｓは絶対値をとることを表し、Ｉ[i][j]は現マクロブロックの画像を表し、ｐ_y [i][j]はステップ１１で生成した現マクロブロックの予測画像を表し、ｉとｊはそれぞれ垂直方向及び水平方向の画素の位置を表すものとする。

Here, abs represents an absolute value, I [i] [j] represents an image of the current macroblock, and p _y [i] [j] represents the predicted image of the current macroblock generated in step 11. I and j represent the positions of the pixels in the vertical and horizontal directions, respectively.

  Subsequently, in step 13, the prediction method index value prevpred used for encoding in the macroblock at the same position in the previous frame is read from the prediction method memory 105, and whether or not the value of pred is equal to prevpred. (Processing of the previous frame prediction method evaluation value setting unit 104).

  When it is determined by this determination processing that the value of pred is equal to prevpred, the process proceeds to step 14, and the variable prevSAD representing the value of SAD in the current macroblock when using the prediction method of the previous frame is set to step After substituting the SAD calculated in step 12, the process proceeds to step 15 described below (process of the previous frame prediction method evaluation value setting unit 104).

  On the other hand, when it is determined in step 13 that the value of pred is not equal to prevpred, the process proceeds to step 15 immediately without performing step 14, and the SAD (pred value calculated in step 12 is set. It is determined whether or not the SAD associated with the corresponding prediction method is smaller than minSAD (processing of the provisional prediction method setting unit 102).

  When it is determined by this determination process that SAD is smaller than minSAD, the process proceeds to step 16, and the value of pred is substituted into variable bestpred representing the index of the prediction method having the smallest SAD so far. After substituting the value of SAD for minSAD, the process proceeds to step 17 described below (process of provisional prediction method setting unit 102).

  On the other hand, if it is determined in step 15 that the SAD calculated in step 12 is larger than minSAD, the process proceeds directly to step 17 without performing step 16, and the value of pred is set to 2. It is determined whether or not they are equal (processing of the provisional prediction method setting unit 102).

  When it is determined by this determination processing that the value of pred is not equal to 2, the process proceeds to step 18 and increments the value of pred by one, and then returns to step 11 to determine the next prediction method. The above processing is repeated (processing of the provisional prediction method setting unit 102).

On the other hand, when it is determined in step 17 that the value of pred has become equal to 2, the process proceeds to step 19, and the preset parameters scaleC and scaleD are used.
scaleC × minSAD> scaleD × prevSAD
It is determined whether or not the following relational expression holds (processing of the prediction method determination unit 106).

  That is, by using this relational expression to evaluate the ratio between minSAD and prevSAD, the prediction efficiency of the prediction method with the highest prediction efficiency in the current macroblock and the code in the macroblock at the same position in the previous frame are encoded. It is determined whether or not the prediction efficiency in the case of using the prediction method used for conversion is substantially equal.

With this decision process,
scaleC × minSAD> scaleD × prevSAD
Is determined, the prediction efficiency of the prediction method having the maximum prediction efficiency in the current macroblock and the prediction method used for encoding in the macroblock at the same position in the previous frame are used. It is determined that the prediction efficiency is almost equal to the case, and the process proceeds to step 20 where the bestpred value is replaced with the prevpred value, the bestpred value is stored in the prediction method memory 105, and the process is terminated. To do.

  On the other hand, when it is determined that this relational expression is not satisfied, the value of bestpred is directly stored in the prediction method memory 105 without changing the value of bestpred by not performing the process of step 20, and the process is terminated. To do.

  In response to the processing of the intra-frame prediction unit 10, an intra encoding unit (not shown) encodes the current macroblock using a prediction method corresponding to the value of bestpred stored in the prediction method memory 105 in this way. Execute.

  In this way, the intra-frame prediction unit 10 has the prediction efficiency of the prediction method used in the macroblock at the same position in the previous frame and the prediction efficiency of the prediction method having the maximum prediction efficiency in the current macroblock. If they are almost equal, processing is performed so that the prediction method of the previous frame is used for encoding the current macroblock.

  According to this configuration, according to the present invention, it is possible to reduce the amount of flicker while suppressing a decrease in encoding efficiency.

  Here, in the flowchart of FIG. 2, in order to compare the prediction evaluation value of the temporary prediction method with the prediction evaluation value of the prediction method used in the previous frame, threshold processing is performed on the ratio. The reason is as follows.

  That is, when the predicted evaluation value is an amount such as residual energy, the magnitude tends to be proportional to the energy of the current image. From now on, if a prediction method to be used for encoding is determined by performing threshold processing on such a difference in prediction evaluation values, the difference in prediction evaluation values is also proportional to the energy of the current image. Decrease in coding efficiency becomes large in a region where is small. On the other hand, if threshold processing is performed for the ratio of predicted evaluation values, a normalized comparison can be made with respect to the energy of the current image, so that such inconvenience does not occur.

  FIG. 3 shows another embodiment of the moving picture encoding apparatus 1 of the present invention.

  The moving picture encoding apparatus 1 according to the present embodiment according to this embodiment employs a configuration in which a prediction method candidate selection unit 20 is provided in addition to the intra-frame prediction unit 10 shown in FIG.

  The prediction method candidate selection unit 20 has a function (selection function) for narrowing down the prediction method used for calculating the prediction evaluation value according to some algorithm when the current macroblock is given. For example, in the case where 10 types of prediction methods are available, it has a function of selecting, for example, 5 types of prediction methods.

  When such a prediction method candidate selection unit 20 is prepared, the number of prediction methods for calculating a prediction evaluation value is reduced, so that the amount of calculation for calculating the prediction evaluation value can be greatly reduced.

  For the present invention, it does not matter what algorithm the prediction method candidate selection unit 20 uses to narrow down the prediction method. However, the prediction method candidate selection unit 20 is described in the following reference, for example. The algorithm will narrow down the prediction method.

[References] P. Feng et.al. "Fast mode decision for intra prediction", JVT-I049, Sept., 2003.
This reference describes a configuration in which an edge of an image is detected using a Sobel filter, and a prediction mode is selected based on the detected edge direction. By using such a configuration, it is possible to narrow down the prediction method.

  FIG. 4 illustrates an example of a flowchart executed by the intra-frame prediction unit 10 when the prediction method candidate selection unit 20 is prepared.

  Next, according to this flowchart, the process of the present invention when the prediction method candidate selection unit 20 is prepared will be described in detail.

  When the prediction method candidate selection unit 20 selects a prediction method candidate for the current macroblock, the intraframe prediction unit 10 first receives the prediction method candidate in step 100.

  Subsequently, in step 101, the prediction method indicated by prevpred described above is specified, and in the subsequent step 102, whether or not the specified prediction method is included in the prediction method candidates given from the prediction method candidate selection unit 20. Determine whether.

  When it is determined that the prediction method indicated by prevpred described above is included in the prediction method candidates given from the prediction method candidate selection unit 20 according to this determination processing, the process proceeds to step 103 and the prediction method candidate selection unit 20 2 is set as a prediction method that can be used for the current macroblock, and the above-described flowchart of FIG. 2 is executed to determine a prediction method for the current macroblock. Then, it is instructed to encode the current macroblock using the prediction method indicated by bestpred to which a value is assigned according to the execution result, and the process is terminated.

  In response to this instruction, an intra coding unit (not shown) performs coding of the current macroblock using the prediction method indicated by bestpred.

  In this way, when the intra-frame prediction unit 10 determines that the prediction method indicated by prevpred described above is included in the prediction method candidates given from the prediction method candidate selection unit 20, the above-described prediction method shown in FIG. By executing the flowchart, the prediction method for the current macroblock is determined.

  On the other hand, when it is determined that the prediction method indicated by prevpred described above is not included in the prediction method candidates given from the prediction method candidate selection unit 20 according to the determination processing in step 102, step 13 in the flowchart of FIG. 14 is not possible to execute, so the process proceeds to step 105, and the prediction method candidate given from the prediction method candidate selection unit 20 is set as a prediction method usable for the current macroblock. 5 is executed to determine a prediction method for the current macroblock, and in subsequent step 106, the current macroblock is encoded using the prediction method indicated by bestpred to which a value is assigned according to the execution result. To end the process.

  In response to this instruction, an intra coding unit (not shown) performs coding of the current macroblock using the prediction method indicated by bestpred.

  Next, the flowchart of FIG. 5 will be described.

  That is, when the process proceeds to step 105 in the flowchart of FIG. 4, the intra-frame prediction unit 10 first uses the current macroblock image and prediction in step 30 as shown in the flowchart of FIG. 5. The decoded image is input, and the initial value “0” is substituted for the variable pred indicating the index of the prediction method, and the initial value “131072” is substituted for the variable minSAD into which the prediction evaluation value is substituted in a later process. To do.

  Subsequently, in step 31, a prediction image is generated using a prediction method corresponding to the value substituted for pred, and in step 32, the prediction corresponding to the value substituted for pred is generated using the generated prediction image. A predicted evaluation value SAD related to the method is obtained.

  Subsequently, in step 33, it is determined whether or not the SAD related to the prediction method corresponding to the calculated pred value is smaller than minSAD, and when it is determined that the SAD is smaller than minSAD, the process proceeds to step 34. Thus, after substituting the value of pred for the variable bestpred representing the index of the prediction method with the smallest SAD so far, and substituting the value of SAD for minSAD, the process proceeds to step 35 described below.

  On the other hand, if it is determined in step 33 that the SAD is larger than minSAD, the process proceeds to step 35 immediately without performing the process in step 34, and the value of pred is the maximum of the index representing the prediction method. It is determined whether or not the value is equal to max.

  When it is determined by this determination processing that the value of pred is not equal to max, the process proceeds to step 36, the value of pred is incremented by one, and then the process returns to step 31, whereby the next prediction method is described above. Repeat the process.

  On the other hand, when it is determined in step 35 that the value of pred is equal to max, the process proceeds to step 37, and the prediction method used for encoding in the macroblock at the same position in the previous frame. Is read from the prediction method memory 105, a prediction image is generated using a prediction method corresponding to the value of the prevpred, and in step 38, the value of the prevpred is set using the generated prediction image. A prediction evaluation value SAD for the corresponding prediction method is obtained.

  Subsequently, in step 39, the SAD calculated in step 38 is substituted for a variable prevSAD representing the value of SAD in the current macroblock when the prediction method for the previous frame is used.

Subsequently, in step 40, using the preset parameters scaleC and scaleD,
scaleC × minSAD> scaleD × prevSAD
It is determined whether or not the relational expression is established.

With this decision process,
scaleC × minSAD> scaleD × prevSAD
Is determined, the prediction efficiency of the prediction method having the maximum prediction efficiency in the current macroblock and the prediction method used for encoding in the macroblock at the same position in the previous frame are used. 4 is judged to be almost equal to the prediction efficiency in the case, the process proceeds to step 41, where the value of bestpred is replaced with the value of prevpred, and then the value of bestpred is stored in the prediction method memory 105. The process of step 105 in the flowchart ends.

  On the other hand, when it is determined that this relational expression is not satisfied, the value of bestpred is stored in the prediction method memory 105 as it is without changing the value of bestpred by not performing the process of step 40. The process of step 105 in the flowchart ends.

  In this manner, the intraframe prediction unit 10 confirms that the prediction method indicated by prevpred described above is not included in the prediction method candidates given from the prediction method candidate selection unit 20 in step 102 of the flowchart of FIG. 4. When making the determination, the process proceeds to step 105, and the flowchart of FIG. 5 is executed to determine the prediction method for the current macroblock.

  Then, when the prediction method for the current macroblock is determined, the intraframe prediction unit 10 uses the prediction method indicated by the bestpred value substituted by the execution result in step 106 of the flowchart of FIG. 4 as described above. In response to this, the intra coding unit (not shown) performs coding of the current macroblock using the prediction method indicated by bestpred.

  Although the present invention has been described according to the illustrated embodiment, the present invention is not limited to this. For example, in the embodiment, it has been described that the prediction method does not have a hierarchical structure, but the present invention can also be applied when the prediction method has a hierarchical structure.

  For example, H.264 intra-frame prediction is hierarchical. Specifically, each macroblock can use two prediction modes, ie, a 4 × 4 prediction mode and a 16 × 16 prediction mode (this means that there are two prediction methods in terms of a macroblock division method). Prediction methods such as vertical prediction and horizontal prediction can be used in each prediction mode. In such a case, for example, when the average prediction of the 16 × 16 prediction mode is used in the previous frame and the prediction efficiency of the current macroblock is the maximum in the 4 × 4 prediction mode, in the present invention, the 4 × 4 prediction mode is used. The prediction efficiency of the current macroblock may be determined by comparing the prediction efficiency when the 4 prediction mode is used with the prediction efficiency of the average value prediction in the 16 × 16 prediction mode.

It is an example of one Embodiment of the moving image encoder of this invention. It is an example of the flowchart which an intra prediction part performs. It is another example of embodiment of the moving image encoder of this invention. It is an example of the flowchart which an intra prediction part performs. It is an example of the flowchart which an intra prediction part performs. It is explanatory drawing of the pixel row | line | column utilized for prediction.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Moving image encoder 10 Intra-frame prediction part 101 Image information input part 102 Temporary prediction method setting part 103 Prediction method evaluation part 104 Previous frame prediction method evaluation value setting part 105 Prediction method memory 106 Prediction method determination part

Claims (16)

A moving image encoding method for encoding a moving image by intra-frame prediction,
A temporary prediction method setting step of selecting one of the prediction methods available in the current block encoding target block and setting it as a temporary prediction method;
Prediction evaluation value of the prediction method set in the temporary prediction method setting step, and prediction evaluation value when using the prediction method used in the block at the same position as the encoding target block in the frame encoded immediately before the current frame And a prediction method determination step for determining a prediction method to be used in the current block encoding target block,
A moving image encoding method as a feature. The moving image encoding method according to claim 1,
In the temporary prediction method setting step, based on the prediction evaluation value of the prediction method, one of the prediction methods that can be used in the encoding target block of the current frame is selected and set as a temporary prediction method. ,
A moving image encoding method as a feature. A moving image encoding method for encoding a moving image by intra-frame prediction,
A prediction evaluation value calculating step for calculating a prediction evaluation value for each of the prediction methods available in the current block encoding target block;
Based on the prediction evaluation value calculated in the prediction evaluation value calculation step, a temporary prediction method for selecting one of the prediction methods available in the current block encoding target block and setting it as a temporary prediction method Configuration steps;
Pre-frame prediction method for setting a prediction evaluation value when a prediction method used in a block at the same position as the encoding target block in a frame encoded immediately before the current frame is used as a prediction evaluation value for determining a prediction method An evaluation value setting step;
Based on the prediction evaluation value of the prediction method set in the provisional prediction method setting step and the prediction evaluation value set in the previous frame prediction method evaluation value setting step, the prediction method used in the current block encoding target block Having a prediction method determination step for determining,
A moving image encoding method as a feature. In the moving image encoding method according to claim 3,
In the previous frame prediction method evaluation value setting step, in the prediction evaluation value calculation step, a prediction evaluation value in the case of using the prediction method used in the block at the same position as the encoding target block in the frame encoded immediately before the current frame Is calculated as the prediction evaluation value for determining the prediction method,
A moving image encoding method as a feature. In the moving image encoding method according to claim 3,
In the previous frame prediction method evaluation value setting step, in the prediction evaluation value calculation step, a prediction evaluation value in the case of using the prediction method used in the block at the same position as the encoding target block in the frame encoded immediately before the current frame If is not calculated, calculate the prediction evaluation value and set it as the prediction evaluation value for determining the prediction method.
A moving image encoding method as a feature. In the moving image encoding method according to any one of claims 2 to 5,
In the provisional prediction method setting step, setting a prediction method that shows the optimal prediction evaluation value as a temporary prediction method,
A moving image encoding method as a feature. In the moving image encoding method according to any one of claims 1 to 6,
In the prediction method determination step, determining a prediction method to be used in the current block encoding target block based on a threshold process for the ratio of the prediction evaluation values of the two prediction methods,
A moving image encoding method as a feature. A video encoding device that encodes a video by intra-frame prediction,
A temporary prediction method setting unit that selects one of the prediction methods available in the current block encoding target block and sets it as a temporary prediction method;
Prediction evaluation value of the prediction method set in the temporary prediction method setting unit and prediction evaluation value when using the prediction method used in the block at the same position as the encoding target block in the frame encoded immediately before the current frame And a prediction method determination unit that determines a prediction method to be used in the current block encoding target block,
A moving image encoding device. The moving image encoding apparatus according to claim 8, wherein
The temporary prediction method setting unit selects one of the prediction methods available in the current block encoding target block based on the prediction evaluation value of the prediction method, and sets it as the temporary prediction method. ,
A moving image encoding device. A video encoding device that encodes a video by intra-frame prediction,
A prediction evaluation value calculation unit that calculates a prediction evaluation value for each of the prediction methods available in the current block encoding target block;
Based on the prediction evaluation value calculated by the prediction evaluation value calculation unit, a temporary prediction method for selecting one of the prediction methods available in the current block encoding target block and setting it as a temporary prediction method A setting section;
Pre-frame prediction method for setting a prediction evaluation value when a prediction method used in a block at the same position as the encoding target block in a frame encoded immediately before the current frame is used as a prediction evaluation value for determining a prediction method An evaluation value setting unit;
Based on the prediction evaluation value of the prediction method set by the provisional prediction method setting unit and the prediction evaluation value set by the previous frame prediction method evaluation value setting unit, a prediction method used in the current block encoding target block Having a prediction method determination unit for determining,
A moving image encoding device. The moving image encoding device according to claim 10,
The previous frame prediction method evaluation value setting unit is a prediction evaluation value calculation unit, and a prediction evaluation value in the case of using a prediction method used in a block at the same position as the encoding target block in a frame encoded immediately before the current frame Is calculated as the prediction evaluation value for determining the prediction method,
A moving image encoding device. The moving image encoding device according to claim 10,
The previous frame prediction method evaluation value setting unit is a prediction evaluation value calculation unit, and a prediction evaluation value in the case of using a prediction method used in a block at the same position as the encoding target block in a frame encoded immediately before the current frame If is not calculated, calculate the prediction evaluation value and set it as the prediction evaluation value for determining the prediction method.
A moving image encoding device. The moving image encoding device according to any one of claims 9 to 12,
The provisional prediction method setting unit sets a prediction method indicating an optimal prediction evaluation value as a provisional prediction method,
A moving image encoding device. The moving image encoding device according to any one of claims 8 to 13,
The prediction method determination unit determines the prediction method to be used in the current block encoding target block based on the threshold processing for the ratio of the prediction evaluation values of the two prediction methods.
A moving image encoding device.   A moving picture coding program for causing a computer to execute processing used to realize the moving picture coding method according to claim 1.   A computer-readable recording medium having recorded thereon a moving picture coding program for causing a computer to execute processing used to realize the moving picture coding method according to claim 1.

Images