JP2010273110A - Image encoding apparatus and image encoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable fast processing in an image encoding apparatus for encoding motion picture data using one estimation mode from among a plurality of estimation modes. <P>SOLUTION: In a plurality of divided blocks of macro block, the image encoding apparatus has: a first estimated pixel producing part for producing a first estimated pixel from a reference pixel corresponding to the divided blocks according to a first estimation mode; a second estimated pixel producing part for producing a second estimated pixel in response to a value estimated from the reference pixel corresponding to the divided blocks according to a second estimation mode when the second estimated pixel is produced from the reference pixel corresponding to the divided blocks; and a determination part for deciding one estimation mode by comparing an estimated value responding to the first estimated pixel value with an estimated value responding to the second estimated pixel value. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image encoding device and an image encoding method.

  When processing such as digitizing and storing a moving image is performed, the amount of data becomes enormous, and thus the moving image is compressed. By storing the compressed data in a hard disk or the like, it is possible to store moving image data for a long time. As such an image compression method, a method of using a correlation between adjacent blocks in a frame to increase the compression rate has been studied. For example, the moving picture encoding method H.264 is used. In H.264, intra prediction coding using correlation between blocks is employed.

  FIG. 18 shows a block configuration diagram of the image encoding device 1. As illustrated in FIG. 18, the image encoding device 1 includes a blocking unit 2, a difference calculation unit 3, a DCT unit 4, a quantization unit 5, an encoding unit 6, an inverse quantization unit 7, An inverse DCT unit 8, a reconstruction unit 9, an adjacent pixel storage unit 10, a prediction mode determination unit 11, and an intra prediction unit 12 are included.

  The blocking unit 2 blocks the input image data into a predetermined block unit (color difference signal: 8 pixels × 8 pixels, luminance signal 16 pixels × 16 pixels) for the luminance signal and the color difference signal.

  The intra prediction unit 12 generates prediction data from adjacent pixels accumulated in the adjacent pixel storage unit 10 in the prediction mode determined by the prediction mode determination unit 11.

  The difference calculation unit 3 calculates the difference between the predicted pixel data generated by the intra prediction unit 12 and the pixels of the input image data. The DCT unit 4 performs DCT conversion processing on the difference data from the difference calculation unit 3. The quantization unit 5 quantizes the data after the DCT conversion of the DCT conversion unit 4.

  The encoding unit 6 encodes the output data of the quantization unit 5. The inverse quantization unit 7 performs inverse quantization on the output data of the quantization unit 5. The inverse DCT unit 8 performs an inverse DCT transform process on the data inversely quantized by the inverse quantization unit 7. The reconstruction unit 9 reconstructs image data from the output data from the inverse DCT unit 8 and the prediction data from the intra prediction unit 12. The adjacent pixel storage unit 10 stores adjacent pixel data at the boundary of each block from the output data from the reconstruction unit 9.

  The prediction mode determination unit 11 generates a prediction pixel predicted from the adjacent pixel data stored in the adjacent pixel storage unit 10. Then, for each prediction mode, the difference between the predicted pixel value and the input pixel value is calculated, and the prediction mode with the smallest difference value is determined as the optimum prediction mode. Then, the intra prediction unit 12 again generates a prediction pixel in the determined optimal prediction mode, and the difference calculation unit 3 calculates the difference between the prediction pixel and the input pixel.

  As an example of the prediction mode determination unit 11, there is a technique as described in Non-Patent Document 1. FIG. 19 shows a schematic block diagram of Non-Patent Document 1. Here, it is assumed that DC prediction, vertical prediction, horizontal prediction, and plane prediction are adopted as the types of prediction modes.

  As illustrated in FIG. 19, the prediction mode determination unit 11 of Non-Patent Document 1 includes a divided block vertical prediction calculation unit 21, a divided block horizontal prediction calculation unit 22, a divided block DC prediction calculation unit 23, and divided block plane prediction. The calculation unit 24, the difference calculation units 31 to 34, the evaluation value calculation units 41 to 44, and the comparison unit 50 are included. Note that the evaluation value calculation units 41 to 44 include two-dimensional Hadamard transform value generation units 61 to 64 and absolute value sum calculation units 71 to 74, respectively.

  Here, FIG. 20 illustrates a macroblock used in each prediction mode of general intra prediction. In this description, it is assumed that the prediction mode determination unit 11 performs intra prediction for color difference components. In addition, a macroblock has 8 × 8 pixels as a basic unit.

  As shown in FIG. 20, the macroblock has blocks block0 to block3 made up of 4 × 4 pixels. In the vicinity of such a macroblock, peripheral pixels indicated by shading are adjacent. As the peripheral pixels, there are adjacent pixels p0 to p7 in the horizontal direction of the macroblock and adjacent pixels q0 to q7 in the vertical direction. In addition, there is an adjacent pixel pq at the upper left of the macroblock, that is, at a location in contact with the adjacent pixels p0 and q0.

  The values pq, p0 to p7, and q0 to q7 shown for each pixel indicate the pixel name and the pixel value of that pixel. Further, each pixel value of the macroblock is assumed to be Q (x, y) (x = 0 to 7, y = 0 to 7). Furthermore, in the present specification, the term “divided block” refers to each of the blocks block 0 to block 3 formed by dividing a macro block.

The divided block vertical prediction calculation unit 21 calculates a prediction pixel value predicted by the vertical prediction mode from the adjacent pixels p0 to p7 of the macroblock. FIG. 21 shows a conceptual diagram of intra prediction in the vertical prediction mode. As shown in the conceptual diagram of FIG. 21, in the vertical prediction mode, the predicted pixel value of each pixel is determined by the adjacent pixel values p0 to p7 in the horizontal direction. Therefore, the predicted pixel value P (x, y) by vertical prediction is
P (0, y) = p0 (y = 0-7)
P (1, y) = p1 (y = 0-7)
P (2, y) = p2 (y = 0-7)
P (3, y) = p3 (y = 0-7)
P (4, y) = p4 (y = 0-7)
P (5, y) = p5 (y = 0-7)
P (6, y) = p6 (y = 0-7)
P (7, y) = p7 (y = 0-7)
It becomes.

  Here, attention is focused on the blocks block 0 and block 2. FIG. 22 shows prediction pixel matrices P_VERT (0) and P_VERT (2) of prediction values obtained by vertical prediction of blocks block0 and block2 by the divided block vertical prediction calculation unit 21. Note that the prediction pixel matrices P_VERT (1) and P_VERT (3) of the blocks block1 and block3 are obtained by replacing p0 to p3 of the prediction pixel matrix P_VERT (0) with the values of p4 to p7, respectively.

The divided block horizontal prediction calculation unit 22 calculates a prediction pixel value predicted by the horizontal prediction mode from the adjacent pixels q0 to q7 of the macroblock. FIG. 23 shows a conceptual diagram of intra prediction in the horizontal prediction mode. As shown in the conceptual diagram of FIG. 23, in the horizontal prediction mode, the predicted pixel value of each pixel is determined by the adjacent pixel values q0 to q7 in the vertical direction. Therefore, the predicted pixel value P (x, y) by horizontal prediction is
P (x, 0) = q0 (x = 0 to 7)
P (x, 1) = q1 (x = 0 to 7)
P (x, 2) = q2 (x = 0 to 7)
P (x, 3) = q3 (x = 0 to 7)
P (x, 4) = q4 (x = 0 to 7)
P (x, 5) = q5 (x = 0 to 7)
P (x, 6) = q6 (x = 0 to 7)
P (x, 7) = q7 (x = 0-7)
It becomes. Here, attention is focused on the blocks block 0 and block 1. FIG. 24 shows prediction pixel matrices P_HORI (0) and P_HORI (1) of prediction values obtained by horizontally predicting blocks block0 and block1 by the divided block horizontal prediction calculation unit 22. Note that the predicted pixel matrices P_HORI (2) and P_HORI (3) of the blocks block2 and block3 are obtained by replacing q0 to q3 of the predicted pixel matrix P_HORI (0) with the values of q4 to q7, respectively.

The divided block DC prediction calculation unit 23 calculates a prediction pixel value predicted by the DC prediction mode from the adjacent pixels p0 to p7 and q0 to q7 of the macroblock. FIG. 25 shows a conceptual diagram of intra prediction in the DC prediction mode. As shown in the conceptual diagram of FIG. 25, the divided block DC prediction calculation unit 23 calculates a predicted pixel value of each block based on an average value of adjacent pixel values as DC prediction.
The predicted pixel values P (x, y) of the DC predicted values DC (k) (k = 0 to 3) corresponding to the blocks block 0 to block 3 are as follows.

In the case of the DC prediction value DC (0) of the divided block,
P (x, y) = (p0 + p1 + p2 + p3 + q0 + q1 + q2 + q3 + 4) / 8 (x = 0-3, y = 0-3)
It becomes.

In the case of the DC prediction value DC (1) of the divided block,
P (x, y) = (p4 + p5 + p6 + p7 + 2) / 4 (x = 4-7, y = 0-3)
It becomes.

In the case of the DC prediction value DC (2) of the divided block,
P (x, y) = (q4 + q5 + q6 + q7 + 2) / 4 (x = 0-3, y = 4-7)
It becomes.

In the case of the DC prediction value DC (3) of the divided block,
P (x, y) = (p4 + p5 + p6 + p7 + q4 + q5 + q6 + q7 + 4) / 8 (x = 4-7, y = 4-7)
It becomes.

  FIG. 26 shows a prediction pixel matrix P_DC (k) of prediction values obtained by DC prediction of each block by the divided block DC prediction calculation unit 23.

The divided block plane prediction calculation unit 24 calculates a prediction pixel value predicted by the plane prediction mode from the adjacent pixels p0 to p7, q0 to q7, and pq of the macroblock. FIG. 27 shows a conceptual diagram of intra prediction in the planar prediction mode. In the planar prediction mode, the predicted value P (x, y) (x = 0 to 7, y = 0 to 7) of each pixel is
P (x, y) = (a + b × (x−3) + c × (y−3) +16) / 32
It becomes. However,
a = (p7 + q7)
b = (32 × H + 32) / 64
c = (32 × V + 32) / 64
H = (p7-pq) × 4 + (p6-p0) × 3 + (p5-p1) × 2 + (p4-p2)
V = (q7-pq) × 4 + (q6-q0) × 3 + (q5-q1) × 2 + (q4-q2)
It is. Here, in FIG. 28A and FIG. 28B, prediction pixel matrices P_PLAN (0) and P_PLAN (1) of predicted values obtained by plane prediction of the blocks block0 and block1 by the divided block plane prediction calculation unit 24. Indicates. In addition, FIGS. 29A and 29B show prediction pixel matrices P_PLAN (2) and P_PLAN (3) of prediction values obtained by plane prediction of the blocks block2 and block3. As can be seen from the matrix coefficients obtained by plane prediction in FIGS. 28 (a), 28 (b), 29 (a), and 29 (b), the pixel position that serves as a reference for plane prediction (hereinafter referred to as a reference pixel). (Referred to as position) is Q (3, 3) as shown by the macroblock in FIG.

  The difference calculation units 31 to 34 receive image data (hereinafter referred to as input pixels) inside the divided blocks from the block forming unit 2 in FIG. 18, and pixels stored in the adjacent pixel storage unit 10 (hereinafter referred to as reference pixels). The difference between the prediction pixel values generated by the divided block vertical prediction calculation unit 21, the divided block horizontal prediction calculation unit 22, the divided block DC prediction calculation unit 23, and the divided block plane prediction calculation unit 24 is calculated. An error value is calculated.

  The evaluation value calculation units 41 to 44 calculate evaluation values corresponding to the prediction error values calculated by the difference calculation units 31 to 34, respectively. For example, the evaluation value calculation units 41 to 44 include two-dimensional Hadamard transform value generation units 61 to 64 and absolute value sum calculation units 71 to 74, respectively. Then, the two-dimensional Hadamard transform value generation units 61 to 64 perform Hadamard transform processing on the prediction error values calculated by the difference calculation units 31 to 34, respectively. Then, the absolute value sum calculation units 71 to 74 calculate the absolute value sums of the processing results, and output them as the evaluation values.

  The comparison unit 50 compares the evaluation values output by the absolute value sum calculation units 71 to 74, respectively, and determines the prediction mode having the smallest evaluation value as the optimum prediction mode.

  In addition, there exists a thing like patent document 1 as another technique of the prediction mode determination part 11. FIG.

JP 2008-172581 A

Joint Video Team (JVT) of ISO / IEC MPEG and ITU-T VCEG "Text Description of Joint Model Reference Encoding Methods and Decoding Concealment Methods"

  When the prediction mode determination unit 11 in FIG. 19 calculates the prediction value of the divided block in each prediction mode, if all adjacent reference pixels of the necessary macroblock are not read, the calculation of the evaluation value in each prediction mode is started. Can not. Therefore, there is a drawback that the entire processing time is increased.

  According to one embodiment of the present invention, moving image data is input, and at least one prediction mode of a plurality of prediction modes is used to predict moving image data of a macroblock from reference pixels adjacent to each macroblock. An image encoding apparatus that encodes the input moving image data using the moving image data, wherein each of the divided blocks obtained by dividing the macroblock into at least one of the plurality of prediction modes. According to one first prediction mode, a first prediction pixel generation unit that generates a first prediction pixel from the reference pixels corresponding to the divided blocks, and a second prediction mode of the plurality of prediction modes, When generating the second prediction pixel from the reference pixel corresponding to the divided block, the value estimated from the reference pixel corresponding to the divided block is Comparing the evaluation value according to the first prediction pixel value and the evaluation value according to the second prediction pixel value, the second prediction pixel generation unit that generates the second prediction pixel at the same time, And a determination unit that determines one prediction mode.

  According to another aspect of the present invention, moving image data is input, and at least one prediction mode among a plurality of prediction modes is used to predict moving image data of a macroblock from reference pixels adjacent to each macroblock, An image encoding method for encoding the input moving image data using predicted moving image data, wherein each of the divided blocks obtained by dividing the macroblock into at least one of the plurality of prediction modes. A first prediction pixel is generated from the reference pixel corresponding to the divided block by one first prediction mode, and the second prediction mode of the plurality of prediction modes corresponds to the divided block. When generating the second prediction pixel from the reference pixel, the second prediction pixel is determined according to the value estimated from the reference pixel corresponding to the divided block. And an evaluation value according to the value of the first prediction pixel and an evaluation value according to the value of the second prediction pixel to determine the one prediction mode. .

  The image coding apparatus according to the present invention includes only the reference pixels corresponding to the respective divided blocks obtained by dividing the macroblock by the first prediction pixel generation unit, the variable estimation unit, and the second prediction pixel generation unit. First and second prediction pixels are generated. Then, an optimal prediction mode is determined based on the evaluation values corresponding to the first and second prediction pixels. Thus, when calculating the evaluation value of each prediction mode in the divided block, it is not necessary to read out all the reference pixels adjacent to the macroblock, and a high-speed processing operation is possible.

  The image encoding apparatus according to the present invention can increase the speed of operation processing.

3 is a prediction mode determination unit of the image encoding device according to the first embodiment; FIG. 6 is a diagram for explaining a divided block 0 and its peripheral pixels according to the first embodiment. FIG. 3 is a diagram for explaining a divided block 1 and peripheral pixels according to the first embodiment; FIG. 3 is a diagram for explaining a divided block 2 and peripheral pixels according to the first embodiment; FIG. 3 is a diagram for explaining a divided block 3 and peripheral pixels according to the first embodiment; It is a graph explaining the concept of the calculation method of the variable value of the plane prediction mode in a normal macroblock. 5 is a graph for explaining a concept of a method for calculating a variable value in a planar prediction mode in a divided block according to the first embodiment; 5 is a graph for explaining a concept of a method for calculating a variable value in a planar prediction mode in a divided block according to the first embodiment; 10 is a prediction mode determination unit of the image encoding device according to the second embodiment. It is the macro block and division | segmentation block of a luminance component concerning Embodiment 2. FIG. FIG. 10 is a schematic diagram for explaining a divided block and peripheral pixels according to the second embodiment; FIG. 10 is a schematic diagram for explaining a reference pixel position in a planar prediction mode in a divided block according to a second embodiment; It is the macroblock and division | segmentation block concerning other embodiment. It is the macroblock and division | segmentation block concerning other embodiment. It is the macroblock and division | segmentation block concerning other embodiment. It is the macroblock and division | segmentation block concerning other embodiment. It is the macroblock and division | segmentation block concerning other embodiment. It is a block configuration of an image encoding device. It is the prediction mode determination part of the conventional image coding apparatus. It is a figure for demonstrating a macroblock and its adjacent pixel (color difference component). It is a schematic diagram for demonstrating vertical prediction. This is a matrix of vertical prediction pixels (divided blocks 0, 2). It is a schematic diagram for demonstrating horizontal prediction. This is a matrix of horizontal prediction pixels (divided blocks 0, 1). It is a schematic diagram for demonstrating DC prediction. This is a matrix of DC predicted pixels (divided blocks 0 to 3). It is a schematic diagram for demonstrating plane prediction. This is a matrix of plane prediction pixels (divided blocks 0, 1). It is a matrix of plane prediction pixels (divided blocks 2, 3).

  Embodiment 1 of the Invention

  Hereinafter, a specific first embodiment to which the present invention is applied will be described in detail with reference to the drawings. In the first embodiment, the present invention is applied to a prediction mode determination unit of an image encoding device. The overall configuration of the image encoding device according to the first embodiment is the same as that of the image encoding device 1 shown in FIG. In the first embodiment, the optimum intra prediction mode determined by the prediction mode determination unit is determined from four types of DC prediction, vertical prediction, horizontal prediction, and plane prediction. However, it is not necessarily limited to the above four prediction modes.

  FIG. 1 shows a configuration of a prediction mode determination unit 100 of the image encoding device according to the first embodiment. As shown in FIG. 1, the prediction mode determination unit 100 includes a divided block vertical prediction calculation unit 111, a divided block horizontal prediction calculation unit 112, a divided block DC prediction calculation unit 113, a divided block plane prediction calculation unit 114, a difference Calculation units 121 to 124, evaluation value calculation units 161 to 164, and comparison unit 150 are included.

  The divided block plane prediction calculation unit 114 includes a variable estimation unit 171 and an estimated prediction value calculation unit 172. The evaluation value calculation unit 161 includes a two-dimensional Hadamard transform value generation unit 131 and an absolute value sum calculation unit 141. The evaluation value calculation unit 162 includes a two-dimensional Hadamard transform value generation unit 132 and an absolute value sum calculation unit 142. The evaluation value calculation unit 163 includes a two-dimensional Hadamard transform value generation unit 133 and an absolute value sum calculation unit 143. The evaluation value calculation unit 164 includes a two-dimensional Hadamard transform value generation unit 134 and an absolute value sum calculation unit 144. Note that the following description is based on the assumption that the prediction mode determination unit 100 performs intra prediction for color difference components. In addition, a macro block has 8 pixels × 8 pixels as a basic unit.

  Here, using FIG. 2 to FIG. 5, the divided block vertical prediction calculation unit 111, the divided block horizontal prediction calculation unit 112, the divided block DC prediction calculation unit 113, and the divided block plane prediction calculation unit 114 according to the first embodiment. The adjacent pixels of the macro block used by each will be described. The macro block itself is the same as that shown in FIG.

  As shown in the shaded portions in FIGS. 2 to 5, adjacent pixels used for prediction calculation in each prediction mode are different in each of the divided blocks block 0 to block 3 having 4 pixels × 4 pixels as a basic unit. For example, in the divided block block0, as shown in FIG. 2, adjacent pixels p0 to p3 on the horizontal side and adjacent pixels q0 to q3 on the vertical side are used. Therefore, in the divided block block 0, the “reference pixels immediately above the divided block” illustrated in FIG. 1 are adjacent pixels p0 to p3, and the “reference pixels immediately to the left of the divided block” are adjacent pixels q0 to q3.

  In the divided block block1, as shown in FIG. 3, the adjacent pixels p4 to p7 on the horizontal side and the adjacent pixels q0 to q3 on the vertical side are used. Therefore, in the divided block block 1, the “reference pixels immediately above the divided block” illustrated in FIG. 1 are adjacent pixels p 4 to p 7, and the “reference pixels immediately to the left of the divided block” are adjacent pixels q 0 to q 3.

  In the divided block block2, as shown in FIG. 4, horizontal side adjacent pixels p0 to p3 and vertical side adjacent pixels q4 to q7 are used. For this reason, in the divided block block 2, the “reference pixels immediately above the divided block” illustrated in FIG. 1 are adjacent pixels p 0 to p 3, and the “reference pixels directly to the left of the divided block” are adjacent pixels q 4 to q 7.

  In the divided block block3, as shown in FIG. 5, the adjacent pixels p4 to p7 on the horizontal side and the adjacent pixels q4 to q7 on the vertical side are used. For this reason, in the divided block block 3, the “reference pixels immediately above the divided block” illustrated in FIG. 1 are adjacent pixels p 4 to p 7, and the “reference pixels immediately to the left of the divided block” are adjacent pixels q 4 to q 7.

  Below, the structure and operation | movement of the prediction mode determination part 100 are demonstrated. The divided block vertical prediction calculation unit 111 calculates prediction pixel values predicted by the vertical prediction mode from the adjacent pixels p0 to p3 in the divided blocks block0 and block2, and from the adjacent pixels p4 to p7 in the divided blocks block1 and block3. That is, the predicted pixel value of each pixel is determined by the adjacent pixel values p0 to p7 in the horizontal direction. Therefore, the prediction pixel value P (x, y) by the vertical prediction (P_VERT (0) to P_VERT (3)) of the divided blocks block0 to block3 is as follows.

In the case of P_VERT (0),
P (0, y) = p0 (y = 0-3)
P (1, y) = p1 (y = 0-3)
P (2, y) = p2 (y = 0-3)
P (3, y) = p3 (y = 0-3)
It becomes.

In the case of P_VERT (2),
P (0, y) = p0 (y = 4-7)
P (1, y) = p1 (y = 4-7)
P (2, y) = p2 (y = 4-7)
P (3, y) = p3 (y = 4-7)
It becomes.

In the case of P_VERT (1),
P (4, y) = p4 (y = 0-3)
P (5, y) = p5 (y = 0-3)
P (6, y) = p6 (y = 0-3)
P (7, y) = p7 (y = 0-3)
It becomes.

In the case of P_VERT (3),
P (4, y) = p4 (y = 4-7)
P (5, y) = p5 (y = 4-7)
P (6, y) = p6 (y = 4-7)
P (7, y) = p7 (y = 4-7)
It becomes.

  The divided block horizontal prediction calculation unit 112 calculates predicted pixel values predicted by the horizontal prediction mode from the adjacent pixels q0 to q3 in the divided blocks block0 and block1 and from the adjacent pixels q4 to q7 in the divided blocks block2 and block3. That is, the predicted pixel value of each pixel is determined by the adjacent pixel values q0 to q7 in the vertical direction. Therefore, the prediction pixel value P (x, y) by horizontal prediction (P_HORI (0) to P_HORI (3)) of the divided blocks block0 to block3 is as follows.

In the case of P_HORI (0),
P (x, 0) = q0 (x = 0 to 3)
P (x, 1) = q1 (x = 0-3)
P (x, 2) = q2 (x = 0 to 3)
P (x, 3) = q3 (x = 0-3)
It becomes.

In the case of P_HORI (1),
P (x, 0) = q0 (x = 4 to 7)
P (x, 1) = q1 (x = 4-7)
P (x, 2) = q2 (x = 4-7)
P (x, 3) = q3 (x = 4-7)
It becomes.

In the case of P_HORI (2),
P (x, 4) = q4 (x = 0 to 3)
P (x, 5) = q5 (x = 0 to 3)
P (x, 6) = q6 (x = 0-3)
P (x, 7) = q7 (x = 0-3)
It becomes.

For P_HORI (3),
P (x, 4) = q4 (x = 4-7)
P (x, 5) = q5 (x = 4-7)
P (x, 6) = q6 (x = 4-7)
P (x, 7) = q7 (x = 4-7)
It becomes.

  The divided block DC prediction calculation unit 113 includes adjacent pixels p0 to p3 and q0 to q3 in the divided block block0, adjacent pixels p4 to p5 and q0 to q3 in the divided block block1, and adjacent pixels p0 to p3 and q4 to q7 in the divided block block2. In the divided block block3, predicted pixel values predicted by the DC prediction mode are calculated from the adjacent pixels p4 to p7 and q4 to q7. In the DC prediction mode, the predicted pixel value of each pixel is determined based on the average value of the adjacent pixel values in the vertical and horizontal directions corresponding to each divided block. Therefore, the predicted pixel value P (x, y) by DC prediction (P_DC (0) to P_DC (3)) of the divided blocks block0 to block3 is as follows.

In the case of P_DC (0),
P (x, y) = (p0 + p1 + p2 + p3 + q0 + q1 + q2 + q3 + 4) / 8 (x = 0-3, y = 0-3)
It becomes.

In the case of P_DC (1),
P (x, y) = (p4 + p5 + p6 + p7 + 2) / 4 (x = 4-7, y = 0-3)
It becomes.

In the case of P_DC (2),
P (x, y) = (q4 + q5 + q6 + q7 + 2) / 4 (x = 0-3, y = 4-7)
It becomes.

In the case of P_DC (3),
P (x, y) = (p4 + p5 + p6 + p7 + q4 + q5 + q6 + q7 + 4) / 8 (x = 4-7, y = 4-7)
It becomes.

  The divided block plane prediction calculation unit 114 includes adjacent pixels p0 to p3 and q0 to q3 in the divided block block0, adjacent pixels p4 to p5 and q0 to q3 in the divided block block1, and adjacent pixels p0 to p3 and q4 to q7 in the divided block block2. In the divided block block3, predicted pixel values predicted by the planar prediction mode are calculated from the adjacent pixels p4 to p7 and q4 to q7.

  The variable estimation unit 171 of the divided block plane prediction calculation unit 114 calculates a variable for plane prediction from vertical and horizontal adjacent pixel values corresponding to each divided block.

  The estimated predicted value calculation unit 172 of the divided block plane prediction calculation unit 114 uses the variables calculated by the variable estimation unit 171 to perform prediction in the vertical and horizontal adjacent pixel value plane prediction modes corresponding to each divided block. The predicted pixel value to be calculated is calculated.

The operation and operation theory of the divided block plane prediction calculation unit 114 will be described using mathematical expressions. Here, the conventional block division plane prediction calculation unit 24 has been described, but in the plane prediction mode, as described above, the prediction value P (x, y) (x = 0 to 7, y = 0) of each pixel of the macroblock. ~ 7)
P (x, y) = (a + b × (x−3) + c × (y−3) +16) / 32
It becomes. However,
a = (p7 + q7) × 16
b = (34 × H + 32) / 64
c = (34 × V + 32) / 64
H = (p7-pq) × 4 + (p6-p0) × 3 + (p5-p1) × 2 + (p4-p2)
V = (q7-pq) × 4 + (q6-q0) × 3 + (q5-q1) × 2 + (q4-q2)
It is. Note that the pixel P (3, 3) is P (3, 3) = (a + 16) / 32, which is a reference pixel position for plane prediction of the macroblock.

  Here, the variables b and c are weighted averages according to two pixel values of adjacent pixels on the vertical or horizontal side of the macroblock and the distance between the pixels. In the following description, the variable b calculated from the adjacent pixels pq and p0 to p7 on the horizontal side will be described. The calculation method for the variable c is the same as the variable b except that the adjacent pixels pq and q0 to q7 on the vertical direction side are used, and thus the description thereof is omitted.

  A graph for explaining the concept of the method for calculating the value of the variable b is shown in FIG. In the graph of FIG. 6, the vertical axis represents the pixel value of the adjacent pixel, and the horizontal axis represents the relative pixel position. The relative pixel positions are “−1” to “7” for the pixels pq to p7, respectively. Pixels qq and p7, pixels p0 and p6, pixels p1 and p5, and pixels p2 and p4 as shown in the graph of FIG. 6 are weighted with the relative pixel position difference and averaged. The variable b is calculated.

However, in the first embodiment, only adjacent pixels corresponding to each divided block as shown in FIGS. 2 to 5 are used. Therefore, the variable estimation unit 171 estimates this variable using only adjacent pixels corresponding to each divided block. Variables a (k), b (k), c (k) (hereinafter, estimated variables a (k), b (k), c (k) for plane prediction of each divided block k (k = 0 to 3) ))
a (0) = (p3 + q3) × 16
a (1) = (p7 + q3) × 16
a (2) = (p3 + q7) × 16
a (3) = (p7 + q7) × 16
b (0) = b (2) = (((p3-p0) × 3 + (p2-p1)) × 205 + 32) / 64
b (1) = b (3) = (((p7−p4) × 3 + (p6−p5)) × 205 + 32) / 64
c (0) = c (1) = (((q3-q0) * 3 + (q2-q1)) * 205 + 32) / 64
c (2) = c (3) = (((q7-q4) * 3 + (q6-q5)) * 205 + 32) / 64
It becomes.

In the block block 0, the pixel value of the adjacent reference pixel pq may be estimated from the pixel values of the pixels p0 and q0 as follows.
pq = (p0 + q0 + 1) / 2

And using this estimated pixel value pq,
b (0) = (((p2-pq) × 3 + (p1-p0)) × 205 + 32) / 64
c (0) = (((q2-qq) * 3 + (q1-q0)) * 205 + 32) / 64
It is good.

  FIG. 7 shows a graph for explaining the concept of the method of calculating the values of the estimated variables b (0) and b (1) of the blocks block0 and block1 of the first embodiment. Note that the estimation variables b (2) and b (3) of the divided blocks block2 and block3 are the same as the estimation variables b (0) and b (1), respectively. As shown in this graph, in the divided block block0, the gradients of the straight lines between the pixels p0 and p3 and the pixels p1 and p2 are weighted by the difference between the relative pixel positions, averaged, and estimated variable b (0). Is calculated. Further, in the divided block block 1, the estimation variable b (1) is calculated by weighting and averaging the slopes of the straight lines between the pixels p 4 and p 7 and the pixels p 5 and p 6 with the difference between the relative pixel positions. .

  That is, the estimated variables b (0) and b (1) in FIG. 7 are calculated using the approximate value of the variable b described in FIG. 6 as an estimated value. As for the estimated variable c (k), an approximate value of the variable c is calculated as an estimated value in the same manner as the estimated variable b (k) described above.

  As another method for obtaining the estimated variables b (k) and c (k), an approximate straight line by the least square method of four adjacent reference pixels that can be referred to in each divided block as shown in the conceptual diagram of FIG. 8 is obtained. The estimated variables b (k) and c (k) may be obtained using a scaled slope of the approximate straight line.

  The estimated predicted value calculation unit 172 calculates a predicted pixel value predicted in the planar prediction mode using the estimated variables a (k), b (k), and c (k). Therefore, the prediction pixel value P (x, y) by plane prediction (P_PLAN (0) to P_PLAN (0)) of the divided blocks block0 to block3 is as follows.

In the case of P_PLAN (0),
P (x, y) = (a (0) + b (0) × (x−1) + c (0) × (y−1) +16) / 32 (x = 0 to 3, y = 0 to 3)
It becomes. In the case of PLAN (0), the reference pixel position is P (1, 1).

In the case of P_PLAN (1),
P (x, y) = (a (1) + b (1) × (x−3) + c (1) × (y−1) +16) / 32 (x = 4 to 7, y = 0 to 3)
It becomes. In the case of PLAN (1), the reference pixel position is P (3, 1).

In the case of P_PLAN (2),
P (x, y) = (a (2) + b (2) × (x−1) + c (2) × (y−3) +16) / 32 (x = 0 to 3, y = 4 to 7)
It becomes. In the case of PLAN (2), the reference pixel position is P (1, 3).

In the case of P_PLAN (3),
P (x, y) = (a (3) + b (3) × (x−3) + c (3) × (y−3) +16) / 32 (x = 4 to 7, y = 4 to 7)
It becomes. In the case of PLAN (3), the reference pixel position is P (3, 3).

  The difference calculation units 121 to 124 are the predicted pixel values of the respective divided blocks calculated by the divided block vertical prediction calculation unit 111, the divided block horizontal prediction calculation unit 112, the divided block DC prediction calculation unit 113, and the divided block plane prediction calculation unit 114. The difference between P (x, y) and the pixel value of the input pixel of the divided block is calculated, and the prediction error value of each prediction mode is calculated.

  The evaluation value calculation units 161 to 164 calculate evaluation values corresponding to the prediction error values calculated by the difference calculation units 121 to 124, respectively. The evaluation value calculators 161 to 164 include two-dimensional Hadamard transform value generators 131 to 134 and absolute value sum calculators 141 to 144, respectively.

  The two-dimensional Hadamard transform value generators 131 to 134 perform Hadamard transform processing on the prediction error values calculated by the difference calculators 121 to 124, respectively. Then, the absolute value sum calculation units 141 to 144 respectively calculate the absolute value sum of the processing results and output it as the evaluation value.

  The comparison unit 150 compares the evaluation values output by the absolute value sum calculation units 141 to 144, respectively, and determines the prediction mode having the smallest evaluation value as the optimum prediction mode.

  Here, in the prediction mode determination unit 11 of FIG. 19, when calculating the prediction value of the divided block in each prediction mode, if all the adjacent reference pixels of the necessary macroblock are not read, the evaluation value of each prediction mode Could not start the calculation. However, the prediction mode determination unit 100 of Embodiment 1 does not need to read all the adjacent reference pixels of the macroblock, and only needs to read the adjacent reference pixels corresponding to each divided block. For this reason, the hardware scale of a storage device such as a register holding read data of the adjacent reference pixel value can be reduced.

  Further, since only the adjacent reference pixels corresponding to the divided blocks are read out, the determination process can be started, so that the entire processing time can be shortened. Furthermore, since the adjacent reference pixels referenced in each divided block can be the same in each prediction mode, the prediction mode determination unit 100 can be easily controlled.

  Embodiment 2 of the Invention

  Hereinafter, a specific second embodiment to which the present invention is applied will be described in detail with reference to the drawings. In the second embodiment, as in the first embodiment, the present invention is applied to a prediction mode determination unit of an image encoding device. As in the first embodiment, the overall configuration of the image coding apparatus according to the second embodiment is the same as that of the image coding apparatus 1 shown in FIG. In the second embodiment, as in the first embodiment, the optimum intra prediction mode determined by the prediction mode determination unit is determined from four types of DC prediction, vertical prediction, horizontal prediction, and plane prediction. .

  FIG. 9 illustrates a configuration of the prediction mode determination unit 200 of the image encoding device according to the second embodiment. As illustrated in FIG. 9, the prediction mode determination unit 200 includes a divided block vertical prediction calculation unit 111, a divided block horizontal prediction calculation unit 112, a divided block DC prediction calculation unit 213, a divided block plane prediction calculation unit 214, and a difference. Calculation units 121 to 124, evaluation value calculation units 161 to 164, and comparison unit 150 are included. The evaluation value calculation unit 161 includes a two-dimensional Hadamard transform value generation unit 131 and an absolute value sum calculation unit 141. The evaluation value calculation unit 162 includes a two-dimensional Hadamard transform value generation unit 132 and an absolute value sum calculation unit 142. The evaluation value calculation unit 163 includes a two-dimensional Hadamard transform value generation unit 133 and an absolute value sum calculation unit 143. The evaluation value calculation unit 164 includes a two-dimensional Hadamard transform value generation unit 134 and an absolute value sum calculation unit 144.

  In addition, the structure which attached | subjected the code | symbol same as FIG. 1 among the codes | symbols shown in FIG. 9 has shown the structure similar to or similar to FIG. The difference between the second embodiment and the first embodiment is that the first embodiment performs the intra prediction of the color difference component, whereas the second embodiment performs the intra prediction of the luminance component. It is assumed. In this intra prediction of luminance components, as shown in FIG. 10, the macroblock is 16 pixels × 16 pixels. For this reason, in the second embodiment, a configuration when the present invention is applied to a macroblock of 16 pixels × 16 pixels, which is a difference from the first embodiment, will be described mainly. In addition, when overlapping with Embodiment 1, the description is abbreviate | omitted.

  Here, referring to FIG. 11, each of divided block vertical prediction calculation unit 111, divided block horizontal prediction calculation unit 112, divided block DC prediction calculation unit 213, and divided block plane prediction calculation unit 214 in the second embodiment is used. The adjacent pixels of the macro block to be described will be described. The macroblock itself is the same as that shown in FIG.

  As shown in FIG. 10, the macroblock has 16 divided blocks block (k) (k = 0 to 15) whose basic unit is 4 pixels × 4 pixels. The adjacent pixels used for the prediction calculation in each prediction mode differ corresponding to each of these divided blocks block (k). Therefore, in FIG. 11, adjacent reference pixels in the vertical direction corresponding to an arbitrary divided block block (k) are A to D, and adjacent reference pixels in the horizontal direction are E to H. For this reason, the “reference pixel immediately above the divided block” and the “reference pixel immediately to the left of the divided block” illustrated in FIG. 9 respectively represent the adjacent pixels A to D and E to H in FIG.

  For this reason, the adjacent reference pixels A to D are selected from the adjacent pixels p0 to p15 according to the corresponding divided block block (k). Similarly, the adjacent reference pixels E to H are selected from the adjacent pixels q0 to q15 according to the corresponding divided block block (k). For example, in the divided block block0, adjacent reference pixels A to D are pixels p0 to p3, and adjacent reference pixels E to H are pixels q0 to q3. Further, for example, in the divided block block 10, the adjacent reference pixels A to D are the pixels p4 to p7, and the adjacent reference pixels E to H are the pixels q8 to q11.

  The configuration and operation of the prediction mode determination unit 200 will be described below. The divided block vertical prediction calculation unit 111 performs the same operation as in the first embodiment. For this reason, the prediction pixel value P_VERT (k) predicted by the vertical prediction mode is calculated from the adjacent reference pixels A to D corresponding to the divided block block (k).

  The divided block horizontal prediction calculation unit 112 also performs the same operation as in the first embodiment. For this reason, the prediction pixel value P_HORI (k) predicted by the horizontal prediction mode is calculated from the adjacent reference pixels E to H corresponding to the divided block block (k).

The divided block DC prediction calculation unit 213 includes an estimated prediction value calculation unit 211. The estimated predicted value calculation unit 211 calculates, as estimated values, average values of adjacent reference pixels A to D and E to H corresponding to the divided block block (k), as shown in the following equation. Then, the calculated result is set as an estimated prediction pixel value predicted by the DC prediction mode.
P_DC (k) = (A + B + C + D + E + F + G + H + 4) / 8

The divided block plane prediction calculation unit 214 includes a variable estimation unit 221 and an estimated prediction value calculation unit 222. The variable estimation unit 221 estimates the estimation variables a (k), b (k), and c (k) using the adjacent pixels A to D and E to H corresponding to the divided block block (k). The estimated variables a (k), b (k), and c (k) are expressed as follows.
a (k) = (D + H) × 16
b (k) = (((D−A) × 3 + (C−B)) × 205 + 32) / 64
c (k) = (((H−E) × 3 + (G−F)) × 205 + 32) / 64

The estimated predicted value calculation unit 222 calculates a predicted pixel value predicted by the planar prediction mode using the estimated variables a (k), b (k), and c (k). Therefore, the predicted pixel value P (x, y) by plane prediction of the divided block block (k) is as follows.
P (x, y) = (a (k) + b (k) × (x−xo (k)) + c (k) × (y−yo (k)) + 16) / 32 (x = 0 to 15, y = 0 to 15)

However, xo (k) and yo (k) are respectively in accordance with the divided block block (k),
(Xo (0), yo (0)) = (1,1)
(Xo (1), yo (1)) = (3, 1)
(Xo (2), yo (2)) = (1,3)
(Xo (3), yo (3)) = (3, 3)
(Xo (4), yo (4)) = (5, 1)
(Xo (5), yo (5)) = (7, 1)
(Xo (6), yo (6)) = (5, 3)
(Xo (7), yo (7)) = (7,3)
(Xo (8), yo (8)) = (1,5)
(Xo (9), yo (9)) = (3, 5)
(Xo (10), yo (10)) = (1,7)
(Xo (11), yo (11)) = (3, 7)
(Xo (12), yo (12)) = (5, 5)
(Xo (13), yo (13)) = (7, 5)
(Xo (14), yo (14)) = (5, 7)
(Xo (15), yo (15)) = (7,7)
It becomes. Note that the reference pixel position of each divided block in the planar prediction mode in this case depends on the position of the divided block block (k) in the macro block as shown in the examples of FIGS. 12 (a) to 12 (c). Note that they are different.

  Since the functions and operations of the difference calculation units 121 to 124, the evaluation value calculation units 161 to 164, and the comparison unit 150 are the same as those in the first embodiment, description thereof will be omitted.

  As described above, in the first embodiment, the intra prediction of the color difference component is performed using the present invention. However, the intra prediction of the luminance component can be performed using the present invention as in the second embodiment. Also in the second embodiment, it is not necessary to read all the adjacent reference pixels of the macroblock, and only the adjacent reference pixels corresponding to the divided block need be read. For this reason, the same effect as Embodiment 1 can be acquired.

  Note that the present invention is not limited to the above-described embodiment, and can be changed as appropriate without departing from the spirit of the present invention. For example, in the first embodiment, the divided blocks are 4 pixels × 4 pixels, but may be 8 pixels × 4 pixels as shown in FIG. 13 or 4 pixels × 8 pixels as shown in FIG. Alternatively, when the macro block is 16 pixels × 16 pixels as in the second embodiment, or when the macro block is a unit of pixels larger than that, the divided block is 8 pixels × 8 pixels as shown in FIG. It may be a basic unit. Also, as shown in FIGS. 16 and 17, 16 pixels × 8 pixels or 8 pixels × 16 pixels may be used as a basic unit.

  In this case, in a divided block of n pixels × m pixels (n and m are positive integers), it means that n pixels can be referred to as a reference pixel immediately above the divided block and m pixels can be referred to as a reference pixel immediately to the left of the divided block. In this way, since the number of pixels that can be referred to increases, the estimation accuracy of prediction value calculation at the time of mode determination can be increased.

100, 200 Prediction mode determination unit 111 Division block vertical prediction correction value calculation unit 112 Division block horizontal prediction correction value calculation unit 113 Division block DC prediction correction value calculation unit 114 Division block plane prediction correction value calculation unit 121 to 124 Difference calculation unit 131 ~ 134 2D Hadamard Transform Value Generation Unit 150 Comparison Units 161 to 164 Evaluation Value Calculation Units 171, 221 Variable Estimation Units 172, 222 Estimated Prediction Value Calculation Unit

Claims (15)

Moving picture data is input, using at least one prediction mode among a plurality of prediction modes, the moving picture data of the macroblock is predicted from reference pixels adjacent to each macroblock, and the predicted moving picture data is used. An image encoding device for encoding the input moving image data,
In each divided block obtained by dividing the macroblock into a plurality, a first prediction pixel is generated from the reference pixel corresponding to the division block by using at least one first prediction mode of the plurality of prediction modes. A first predicted pixel generation unit;
When generating the second prediction pixel from the reference pixel corresponding to the divided block by the second prediction mode of the plurality of prediction modes, according to the value estimated from the reference pixel corresponding to the divided block A second predicted pixel generation unit for generating a second predicted pixel;
An image coding apparatus comprising: a determination unit that compares the evaluation value according to the first prediction pixel value with the evaluation value according to the second prediction pixel value and determines the one prediction mode The image encoding apparatus according to claim 1, wherein the plurality of prediction modes include DC prediction, vertical prediction, horizontal prediction, and plane prediction. The first prediction mode is at least one prediction mode of DC prediction, vertical prediction, and horizontal prediction modes;
The image coding apparatus according to claim 2, wherein the second prediction mode is a prediction mode of plane prediction. The second predicted pixel generation unit
A variable estimation unit that generates an estimated variable value for estimating the second predicted pixel using a reference pixel corresponding to the divided block;
4. The image encoding device according to claim 1, further comprising: an estimated predicted value calculation unit configured to estimate a second predicted pixel from the estimated variable value in the second prediction mode. 5. Pq, p0 to pn (n: positive integer) in order from the left in the horizontal direction reference pixels adjacent to the macroblock, and pq, q0 to qm (m in order from the top in the vertical direction. 5 is calculated from the pixel value of the reference pixel corresponding to the divided block among the pixels of p0 to pn and the pixels of q0 to qm. The image encoding device described. Pq, p0 to pn (n: positive integer) in order from the left in the horizontal direction reference pixels adjacent to the macroblock, and pq, q0 to qm (m in order from the top in the vertical direction. : A positive integer), the pixel value of pq is the estimated pq value calculated according to the pixel values of p0 and q0, the estimated pq value, the pixels of p0 to pn, and the pixels of q0 to qm The image coding apparatus according to claim 4, wherein the estimated variable value for plane prediction is calculated from a pixel value of a reference pixel corresponding to the divided block. The first prediction mode is at least one prediction mode of a prediction mode of plane prediction, vertical prediction, and horizontal prediction,
The image coding apparatus according to claim 2, wherein the second prediction mode is a prediction mode of DC prediction. Pq, p0 to pn (n: positive integer) in order from the left in the horizontal direction reference pixels adjacent to the macroblock, and pq, q0 to qm (m in order from the top in the vertical direction. : Is a positive integer), an average value of pixel values of reference pixels corresponding to the divided blocks among the pixels of p0 to pn and the pixels of q0 to qm is set as the estimated value of DC prediction. 8. The image encoding device according to 7. The macroblock is H.264. The image coding apparatus according to any one of claims 1 to 8, comprising a pixel unit according to a H.264 color difference component or luminance component standard. The image coding apparatus according to claim 9, wherein the divided block includes 4 pixels × 4 pixels, 8 pixels × 4 pixels, or 8 pixels × 8 pixels. Moving picture data is input, using at least one prediction mode among a plurality of prediction modes, the moving picture data of the macroblock is predicted from reference pixels adjacent to each macroblock, and the predicted moving picture data is used. An image encoding method for encoding the input moving image data,
In each divided block obtained by dividing the macroblock into a plurality, a first prediction pixel is generated from the reference pixel corresponding to the divided block by at least one first prediction mode of the plurality of prediction modes. ,
When generating the second prediction pixel from the reference pixel corresponding to the divided block by the second prediction mode of the plurality of prediction modes, according to the value estimated from the reference pixel corresponding to the divided block Generating a second predicted pixel;
An image encoding method for comparing the evaluation value according to the value of the first prediction pixel and the evaluation value according to the value of the second prediction pixel to determine the one prediction mode. The image encoding method according to claim 11, wherein the plurality of prediction modes include DC prediction, vertical prediction, horizontal prediction, and plane prediction. The first prediction mode is at least one prediction mode of DC prediction, vertical prediction, and horizontal prediction modes;
The image encoding method according to claim 12, wherein the second prediction mode is a prediction mode of plane prediction. When generating a second prediction pixel from the reference pixel corresponding to the divided block by a second prediction mode of the plurality of prediction modes,
Generating an estimated variable value for estimating the second predicted pixel using a reference pixel corresponding to the divided block;
The image coding method according to claim 13, wherein a second prediction pixel is estimated from the estimated variable value in the second prediction mode. The first prediction mode is at least one prediction mode of a prediction mode of plane prediction, vertical prediction, and horizontal prediction,
The image coding method according to claim 12, wherein the second prediction mode is a prediction mode of DC prediction.

Images