JP2005191898A - Motion compensation encoding method and motion compensation encoding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a motion compensation encoding method and a motion compensation encoding device which can remove unnecessary encoding processings and decrease the amount of encoding. <P>SOLUTION: When a motion vector with respect to an input macro block is set to a zero vector, an encoding for determination as to whether the difference between a predicted image, in which the motion vector is set as the zero vector and the input macro block is further encoded is performed. When the difference is equal to or smaller than a certain value, a variety between both is determined to be a noise component. so that the input macro block is set to the predicted image, without performing encoding processing with respect to this difference. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an encoding method for encoding a video signal input from the outside, and an encoding apparatus therefor. The present invention relates to an encoding method having a motion compensation function, represented by 26x, ISO / IEC standard MPEG, and the like, and an encoding apparatus thereof.

  In a conventional motion compensated prediction apparatus, when a video signal including a noise component is input, it may be determined that a stationary video portion is moving and a motion vector may be detected. Therefore, in such a case, useless processing for encoding and outputting an originally unnecessary motion vector and useless information transmission occur. In addition, when a bit stream affected by such noise is decoded, the decoded video signal is affected by a noise component, and the decoded image quality deteriorates.

  To solve the above problem, for example, in Patent Document 1 below, an evaluation value for a motion vector detected for a certain block is compared with an evaluation value for a motion vector indicating that the block is stationary. Describes a method of setting a motion vector indicating that the block is stationary as a motion vector of the block.

  Patent Document 2 below describes a method for detecting and setting a motion vector in accordance with a predetermined priority order. The following Patent Document 3 includes a simple difference calculator that calculates the sum of absolute differences between the current frame and the previous frame in units of blocks, and a motion vector detector that calculates the sum of absolute differences of motion vectors. The difference absolute value sum output from the calculator and the difference absolute value sum output from the motion vector detector are compared, and if the difference between the two is less than the threshold value, a zero vector indicating a stationary state is obtained. It describes how to choose.

JP-A-58-107785 (page 3, upper right column, line 10 to lower right column, line 12) JP-A-6-30399 (paragraphs 0026 to 0034, FIG. 1) Japanese Patent Laid-Open No. 7-99657 (paragraph 0007, FIG. 1)

  By the way, also in the motion compensation encoding method described in the above-mentioned patent document, encoding processing is performed on a difference calculated from an input video signal and a predicted image in which a zero vector is set as a motion vector in the decoded video signal. Is going. When both image contents have changed, encoding processing for the difference between the two is necessary. However, when the change in image contents is caused by noise components, the encoding processing is unnecessary and the amount of codes increases. End up.

  The present invention has been made to solve the above-described problems, and it is an object of the present invention to provide a motion compensation encoding method and a motion compensation encoding device capable of reducing encoding processing and code amount. It is a thing.

This invention
A motion compensation encoding device that divides an input video signal into a plurality of blocks, performs encoding processing in units of the blocks, and outputs a bitstream obtained by the encoding processing,
A motion vector detecting means for detecting a motion vector corresponding to a block in the inputted video signal from the decoded video signal obtained by decoding the already encoded video signal and the inputted video signal;
Detected by the first prediction image obtained when the decoded video signal and the motion vector are zero vectors, a first evaluation value for evaluating a difference between the block, and the decoded signal and the motion vector detecting means A zero vector for determining whether or not the motion vector is a zero vector from a second predicted image obtained when the motion vector is used and a second evaluation value for evaluating a difference between the block and the second predicted image A determination means;
In the zero vector determination means, when it is determined that the motion vector is a zero vector, the zero determination means comprises: an encoding determination means for determining whether or not to encode a difference between the second predicted image and the block;
When the encoding determination means determines that the difference between the second predicted image and the block is not to be encoded, the block is set as a predicted image.

  The present invention can provide a motion compensation coding method and a motion compensation coding device capable of reducing coding processing and reducing the amount of code.

Embodiment 1 FIG.
In the motion compensation encoding apparatus described in the first embodiment, when the motion compensation predictor 11 sets the motion vector to a zero vector, the motion compensation encoding apparatus further sets the motion vector to the zero vector in the input macroblock and the decoded video signal. An encoding determination (hereinafter referred to as “Not_Coded determination”) is performed as to whether or not the difference from the predicted image of 1 is encoded. If the difference value between the two is less than a certain value, it is determined that the change in the image content between the two is a noise component, and the difference is not encoded. Hereinafter, the present invention will be specifically described with reference to the drawings showing embodiments thereof.

  FIG. 1 is a conceptual diagram showing a video signal input from the outside to the motion compensation coding apparatus according to the first embodiment, FIG. 2 is a block diagram showing a configuration of the motion compensation coding apparatus according to the first embodiment, and FIG. FIG. 4 is a diagram showing an encoding process flow in a motion compensation predictor in the motion compensation encoding apparatus, and FIG. 4 is a block diagram showing a configuration of the motion compensation predictor in the first embodiment.

  First, the video signal input to the motion compensation encoding apparatus 20 will be described. As shown in FIG. 1, each frame of the input video signal is divided into a plurality of blocks, and each block is input to the motion compensation encoding apparatus 20. For example, when the input video signal is a 4: 2: 0 signal composed of a luminance signal (Y signal) and two color difference signals (Cb signal and Cr signal), a 16 × 16 pixel block of the luminance signal and the color difference signal An 8 × 8 pixel block has the same size on the screen. Therefore, a macro block is constituted by four blocks of 8 × 8 pixels of luminance signals and two blocks of 8 × 8 pixels of color difference signals, and each macro block is input to the motion compensation coding apparatus 20 as a video signal. It will be.

  In general, an input video signal is a frame, but it may be a VOP (Video Object Plane) such as MPEG-4.

  Next, processing of the video signal input to the motion compensation encoding apparatus 20 will be described. The video signal is first input to the motion compensation predictor 11. The motion compensation predictor 11 compares the input video signal with the past decoded video signal stored in the memory 10, detects a motion vector for each macroblock, and creates a predicted image. Then, the detected motion vector and an encoding determination result to be described later are output to the variable length encoding means 5, and the predicted image is output to the differentiator 2 and the adder 8. Further, with respect to the input video signal, determination of intra-frame encoding (hereinafter referred to as “intra”) / inter-frame encoding (hereinafter referred to as “inter”) is performed, and the determination result is sent to the switches 2 and 9. Output. The detailed operation of the motion compensation predictor 11 will be described later.

  The differentiator 1 outputs the difference between the input video signal and the predicted image output from the motion compensated predictor 11 to the switcher 2. The switcher 2 outputs either the input video signal or the difference input from the differentiator 2 to the DCT conversion means 3 based on the inter / intra determination result output from the motion compensation predictor 11. That is, based on the intra / inter determination result, an input video signal is output in the case of intra, and a difference is output in the case of inter.

  The DCT conversion means 3 performs DCT conversion on the input video signal or difference input from the switch 2. The motion compensated prediction 11 performs processing in units of macroblocks, but the DCT conversion means 3 performs DCT conversion in units of further dividing the macroblock into 8 × 8 pixels.

  The quantizer 4 quantizes the DCT coefficient input from the DCT conversion means 3. Then, the variable length encoding unit 5 encodes the quantized DCT coefficient input from the quantization unit 4, the motion vector input from the motion compensation predictor 11, the intra / inter determination result, and the encoding determination result. , Output a bitstream.

  On the other hand, the quantized DCT coefficient output from the quantizer 4 is also input to the inverse quantizer 6 in addition to the variable length encoding means 5. The quantized DCT coefficient is inversely quantized by the inverse quantizer 6 and further subjected to inverse DCT transform by the inverse DCT transform means 7. If the intra / inter determination input to the switch 2 is intra, the data output from the inverse DCT means 7 is a decoded video signal. The adder 8 outputs a decoded signal obtained by adding the data input from the inverse quantization means 7 and the predicted image input from the motion compensated predictor 11 to the switch 9.

  Based on the intra / inter determination result input from the motion compensation predictor 11, the switch 9 is either a decoded video signal input from the inverse DCT conversion means 7 or a decoded video signal input from the adder 8. One of them is output to the memory 10. That is, when the intra / inter determination result is intra, the decoded video signal input from the inverse DCT conversion means 7 is output, and when it is inter, the decoded video signal input from the adder 8 is output.

  Next, the operation of the motion compensated predictor 11 will be described with reference to FIGS. 3, 4 and 5. As described above, when the motion compensated predictor 11 described in the present embodiment sets the motion vector for the input macroblock to a zero vector in the decoded video signal, the motion compensation predictor 11 further includes the input macroblock and the decoded video signal. The Not_Coded determination is made as to whether or not to encode the difference from the first predicted image in which the motion vector is set to the zero vector. If the difference between the two is less than a certain value, it is determined that the change in the image contents of both is a noise component, and the difference is not encoded.

First, the motion vector detection means 12 detects a motion vector for the input macroblock (step 101). For example, as shown in FIG. 5, the position of the input macroblock in the frame is (n, m), and the luminance signal in the input video signal is FY (n, m) (0 ≦ n <N, 0 ≦ m <M). When the luminance signal of the decoded image stored in the memory 10 is GY (n, m) (0 ≦ n <N, 0 ≦ m <M), the luminance signal of the input macroblock is
FY (n + i, m + j) (0 ≦ i, j <16)
The luminance signal of the predicted image when the vector (x, y) is a motion vector is
GY (n + x + i, m + y + j) (0 ≦ i, j <16)
It is expressed. Therefore, using these luminance signals, an evaluation value SAD (x, y), which is the sum of absolute differences between the predicted image and the input macroblock, is calculated from Equation 1 below.

  Then, a vector (x, y) that minimizes the evaluation value SAD (x, y) is detected as a motion vector.

  After detecting the motion vector in step 101, next, zero vector determination means 13 performs zero vector determination (step 102). Zero vector determination refers to a first evaluation value S1 that evaluates a difference value between a first predicted image when a motion vector of a decoded video signal corresponding to an input macroblock is a zero vector, and the input macroblock. In the decoded video signal, the motion vector is determined as a zero vector from the second predicted image using the motion vector obtained in the motion vector detection step 101 and the second evaluation value S2 for evaluating the difference value between the input macroblock. It is determined whether or not to set.

The first evaluation value S1, which is the difference value between the first predicted image and the input macroblock,
S1 = SAD (0,0)
And the second evaluation value S2, which is the difference value between the second predicted image and the input macroblock,
S2 = SAD (x, y)
It becomes. Note that the second evaluation value S2 is a value obtained by detecting the minimum value in SAD (x, y).
S1 ≧ S2
It becomes.

And using a predetermined fixed value α,
S1 <S2 + α
In other words,
S1-S2 <α
In this case, the correlation between the image content of the first predicted image and the image content of the input macroblock is high. Therefore, in such a case, it is determined that “the motion vector is set to a zero vector”.

  On the other hand, in the cases other than the above, the correlation between the image content of the first predicted image and the image content of the input macroblock is low. Therefore, in such a case, if the motion vector is set to the zero vector, the difference between the first predicted image and the input macroblock increases, and the amount of code increases. Therefore, “the motion vector is not set to the zero vector” Is determined. The fixed value α is a value determined by the code bit rate, the size of the macroblock, and the like.

  If it is determined in step 102 that “the motion vector is set to the zero vector”, the motion vector is set to the zero vector (step 105). On the other hand, if it is determined in step 102 that “the motion vector is not set to the zero vector”, that is, the motion vector detected in step 101 is used, then the intra / inter determination unit 14 performs the processing for the input macroblock. Intra / inter determination is performed (step 103).

  In the intra / inter determination, for example, the third evaluation value V1 using the luminance signal of the input macroblock shown in the following Equation 2 and the luminance signal of the input macroblock and the luminance signal of the predicted image shown in the following Equation 3 are used. And a fourth evaluation value V2 using the above.

ただし、

However,

And using a predetermined fixed value β,
V1 <V2 or V2 <β
In this case, the input macroblock is determined to be inter, and otherwise determined to be intra.

  When it is determined as “inter” in step 103, the predicted image is created by the predicted image creating means 16 using the motion vector detected in step 101 (step 104). On the other hand, when it is determined as “intra” in step 103, neither the motion vector nor the predicted image is necessary, and thus the operation of the motion compensator 11 ends. The fixed value β is a value determined by the code bit rate, the size of the macroblock, and the like.

  If the motion vector is set to a zero vector in step 105, the encoding determination means (hereinafter referred to as “Not_Coded determination means”) 15 encodes the difference between the first predicted image and the input macroblock. It is determined whether or not.

In the Not_Coded determination, first, the difference between the first predicted image and the input macroblock is calculated for the luminance signal (Y signal) and the two color difference signals (Cb signal and Cr signal). That is, in addition to the luminance signals FY (n, m) and GY (n, m), the color difference signals Cb and Cr of the input macroblock are respectively converted to Fb (n, m) and Fr (n, m) (0 ≦ n < N / 2, 0 ≦ m <M / 2), and the color difference signals Cb and Cr of the decoded video signal are respectively Gb (n, m) and Gr (n, m) (0 ≦ n <N / 2, 0 ≦ m < M / 2), the difference in luminance signal between the first predicted image and the input macroblock is
FY (n + i, m + j) −GY (n + i, m + j) (0 ≦ i, j <16)
And the difference between the two color difference signals is
Fb (n / 2 + i, m / 2 + j) -Gb (n / + i, m / 2 + j)
Fr (n / 2 + i, m / 2 + j) -Gb (n / + i, m / 2 + j)
It is expressed. Then, a fifth evaluation value SY, a sixth evaluation value Sb, and a seventh evaluation value Sr, which are sums of absolute differences for the respective signals, are calculated from the following equations. The fifth evaluation value SY is the same value as the SAD (0, 0) in the evaluation value SAD (x, y) where x = 0 and y = 0.

For each evaluation value calculated by the above Equation 4, using predetermined fixed values γ1, γ2,
SY <γ1 and Sb <γ2 and Sr <γ2
In this case, since the difference between the first predicted image and the input macroblock is small, the correlation between the image contents is high, and it is determined that the change between the two is a noise component. Is determined. In this case, the encoding determination result is “Not_Coded”. On the other hand, in other cases, that is, when the encoding determination result is “Not Not Coded”, it is determined that the image content itself has changed because the correlation between the image content of the first predicted image and the input macroblock is low. It is determined that “the difference between the two is encoded”. The fixed values γ1 and γ2 are values determined by the code bit rate, the size of the macroblock, and the like.

  In the determination of Not_Coded in step 106, when it is determined that “not Not Coded”, that is, when it is determined that “the difference between the first predicted image and the input macroblock is encoded”, the predicted image creation unit 16 is notified. Then, a prediction image is created and output using the first prediction image, the luminance signal of the input macroblock, and the two color difference signals (step 104).

  On the other hand, in the Not_Coded determination of step 106, when it is determined that “it is Not_Coded”, that is, when it is determined that “the difference between the first predicted image and the input macroblock is not encoded”, the input macroblock is It sets as a prediction image (step 107).

  As described above, the motion compensated predictor 11 according to Embodiment 1 performs Not_Coded determination as to whether or not to encode the difference from the input macroblock when the motion vector is a zero vector in the decoded video signal. When the difference value is small, it is determined that the change between the two is a noise component, the difference value is not encoded, and the input macroblock is set as a predicted image. Therefore, the process of encoding the noise component is not necessary, and the entire code amount can be reduced. In particular, the effect is great when the motion compensation encoding apparatus 20 including the motion compensation encoder 11 described in the present embodiment is installed in a portable device or the like that strictly requires a code amount restriction.

  In the first embodiment, the motion compensated predictor 11 has been described as performing motion vector detection, intra / inter determination, Not_Coded determination, and prediction image generation for each macroblock unit. It is also possible to do this every time. That is, the processing unit of the motion compensated predictor 11 in Embodiment 1 is not limited to 16 × 16 pixels, and can be applied to blocks of any size.

  In the motion vector detection in step 101, the evaluation value SAD (x, y), which is the sum of absolute differences between the luminance signal of the input macroblock and the luminance signal of the predicted image, is used. Or a sum of absolute differences of color difference signals may be used.

  Similarly, in the zero vector determination in step 102, the first evaluation value S1 = SAD (0,0) and the second evaluation value S2 = SAD (x, y) are used. Or a weighted sum of the sum of absolute differences between the luminance signal and the color difference signal may be used. That is, any value that can evaluate the size of the difference between the input macroblock and the predicted image may be used.

  Further, in the intra / inter determination in step 103, the third evaluation value V1, which is the sum of squares of the difference between the luminance signal in the input macroblock and the average value, the square of the difference between the luminance signal in the input macroblock and the luminance signal in the predicted image. Although the fourth evaluation value V2 that is the sum is used, it may be a sum of absolute differences, or a color difference signal may be used.

  In the Not_Coded determination in step 106, the fifth evaluation value SY that is the sum of absolute differences between the luminance signal of the input macroblock and the luminance signal of the predicted image, the color difference signal between the color difference signal of the input macroblock and the predicted image, and Although the sixth evaluation value Sb and the seventh evaluation value Sr, which are the sum of absolute differences, are used, the sum of squared differences or the maximum value may be used, respectively.

  In the present embodiment, the case where the motion vector is detected in pixel units, that is, in pixel accuracy has been described. However, the motion vector may be detected in half pixel accuracy or ¼ pixel accuracy.

  Further, the decoded video signal used for motion vector detection is not limited to the one obtained by decoding the video signal input to the motion compensation encoding device 20 in time before the input video signal. A video signal input later may be decoded. Further, the present invention is not limited to the one obtained by decoding the video signal immediately before and after the input video signal.

  Note that the basic processing flow of the motion compensated predictor 11 in Embodiments 3 and 5 to be described later is the same as that of the motion compensated predictor 11 in Embodiment 1. Therefore, the motion compensation predictor 11 according to the third and fifth embodiments also has the effects of removing the noise component encoding process and reducing the code amount, which the motion compensated predictor 11 according to the first embodiment has.

Embodiment 2. FIG.
The encoding process flow of the motion compensation encoding method described in the second embodiment is shown in FIGS. 6.1 and 6.2. In the present embodiment, the overall operation of the motion compensation coding apparatus 20 that performs the “Not_Coded determination” described in the first embodiment will be described. In the first embodiment, the case where the motion compensation predictor 11 includes a motion vector detection unit, a zero vector determination unit, and the like has been described. However, these need not be in the motion compensated predictor 11, and the motion compensated encoding apparatus 20 as a whole only needs to satisfy the functions. In FIG. 6.1 and FIG. 6.2, the same reference numerals are assigned to the same processing steps as those in FIG.

  When the motion compensation encoding apparatus 20 in this embodiment determines Not_Coded in step 106, the motion compensation encoding apparatus 20 performs variable-length encoding on the bitstream including the encoding determination result indicating whether or not the encoding process has been performed. Output from the means 5 (steps 118 and 120). That is, using a predetermined fixed code, a bit stream including a flag indicating whether or not “Not_Coded” is output as an encoding determination result is output.

Hereinafter, the operation of the motion compensation coding apparatus 20 will be described with reference to FIGS. 6.1 and 6.2.
First, a motion vector for the input macroblock is detected in the decoded video signal (step 101), and zero vector determination is performed (step 102). Further, when it is determined in step 102 that “the motion vector is not set to the zero vector”, intra / inter determination is performed on the input macroblock (step 103), and “the motion vector is set to the zero vector” is determined. In this case, a Not_Coded determination is performed (step 106). Since the processing contents in these steps are the same as those described in the first embodiment, detailed description thereof is omitted.

  In the first embodiment, when it is determined that “the motion vector is set to the zero vector”, the motion vector is set to the zero vector before the Not_Coded determination (step 105). However, in the present embodiment, the processing flow is performed after it is determined that “not Not Coded” in the Not_Coded determination in Step 106. This is due to the following reason. That is, if it is not Not_Coded, information that the motion vector is a zero vector is necessary for the process in the prediction image creation in step 104. On the other hand, if it is determined that the input macroblock is “Not_Coded”, the information that the motion vector is a zero vector is not used in the processing in steps 118 and 119 and is unnecessary. However, step 105 may exist at a position as shown in FIG. Also in the first embodiment, step 105 may be present at a position as shown in FIG.

  If it is determined in step 103 that the input macroblock is intra, the input macroblock is DCT transformed as it is without creating a prediction screen (step 111). On the other hand, if it is determined in step 103 that the image is inter, a prediction image of the luminance signal and the color difference signal is created using the motion vector detected in step 101 (step 104), and the prediction image and the input macroblock are compared. The difference is calculated (step 110), and the calculated difference is DCT transformed (step 111).

  If it is determined in step 102 that “the motion vector is set to a zero vector”, then whether or not the difference between the first predicted image having the motion vector as the zero vector and the input macroblock is to be encoded is determined. “Not_Coded determination” is performed (step 106). If it is determined that “the difference between the two is encoded”, ie, “not Not Coded”, a prediction image with a motion vector as a zero vector is created (step 104), and the difference between the prediction image and the input macroblock is calculated. Then, the calculated difference is DCT transformed (step 111). The case where it is determined as “Not_Coded” will be described later.

  The DCT coefficient obtained by the DCT transform in step 111 becomes a quantized DCT coefficient by the quantization in step 112. Further, the quantized DCT coefficients are inversely quantized at step 113 and inverse DCT transformed at step 114. In the case of inter, the quantized DCT coefficients are added to the predicted image to obtain a decoded image (steps 115 and 116). In the case of intra, the output of the inverse DCT transform in step 114 is a decoded image (step 115). The obtained decoded image is stored in the memory 10 (step 117), and the obtained quantized DCT coefficient, motion vector, and intra / inter determination result are encoded and output (step 118).

  If it is determined as “Not_Coded” in step 106, the difference between the input macroblock and the first predicted image is not encoded, so the input macroblock is set as a predicted image and written to the memory 10 (step 119). ). Furthermore, a bit stream including a flag indicating “Not_Coded” as an encoding determination result is output from the variable length encoding means 5 using a fixed code (step 120).

  When the motion vector in the decoded video signal is a zero vector in the decoded video signal, the motion compensation encoding apparatus 20 according to the second embodiment further performs Not_Coded determination as to whether or not to encode the difference from the input macroblock. If it is small, it is determined that the change between the two is a noise component, the difference value is not encoded, and the input macroblock is set as a predicted image. Therefore, the process of encoding the noise component is not necessary, and the entire code amount can be reduced. In particular, when the motion compensation encoding apparatus 20 described in the present embodiment is mounted on a portable device or the like that requires a strict code amount restriction, the effect is great.

Embodiment 3 FIG.
FIG. 7 shows an encoding process flow of the motion compensation encoding method described in the third embodiment. In the present embodiment, an operation will be described in which motion vector detection is performed a plurality of times for the purpose of shortening the motion vector detection processing time in the motion compensated predictor 11 described in the first embodiment. That is, first, the operation of the motion compensated predictor 11 that detects a motion vector with half-pixel accuracy or 1 / 4-pixel accuracy after detecting a motion vector with pixel units, that is, with pixel accuracy will be described. In FIG. 7, the same reference numerals are assigned to the same processing steps as those in FIG.

  The motion compensated predictor 11 in the present embodiment first detects a motion vector in pixel units, that is, in pixel accuracy, before the zero vector determination in step 102 (step 101). This processing content is the same as that of the first embodiment. Further, when it is determined that the inter / inter determination in step 103 is inter, a motion vector is further detected with half-pixel accuracy or ¼-pixel accuracy around the motion vector detected in step 101 (step 201). Since the other processing contents in steps 102 to 107 are the same as those in the first embodiment, description thereof is omitted.

  As described above, the motion compensated predictor 11 according to the third embodiment first detects a motion vector in units of pixels, that is, pixel accuracy, is determined to be inter by the intra / inter determination in step 103, and is detected in step 101. Only when using the motion vector thus determined, the motion vector is further detected with half-pixel accuracy or 1 / 4-pixel accuracy, centering on the motion vector detected with pixel accuracy. Therefore, it is possible to shorten the motion vector detection processing time as compared with the case where the motion vector is detected only with half pixel accuracy or 1/4 pixel accuracy. In addition, since the motion compensated predictor 11 according to the third embodiment performs motion vector detection with pixel accuracy and then with half pixel accuracy or 1/4 pixel accuracy, the motion compensation predictor 11 performs motion vector detection only with pixel accuracy. Compared to the case, the motion vector detection accuracy is improved.

Embodiment 4 FIG.
The coding process flow of the motion compensation coding method described in the fourth embodiment is shown in FIGS. 8.1 and 8.2. In the present embodiment, an operation will be described in which motion vector detection is performed a plurality of times for the purpose of reducing the motion vector detection processing time in the motion compensated prediction apparatus 20 described in the second embodiment. That is, first, the operation of the motion compensated predictor 20 that detects a motion vector with half-pixel accuracy or 1 / 4-pixel accuracy after detecting a motion vector in pixel units, that is, with pixel accuracy will be described. In FIG. 8.1 and FIG. 8.2, the same processing steps as those in FIG. 6.1 and FIG.

  The motion compensation encoder 20 in the present embodiment first detects a motion vector in units of pixels, that is, pixel accuracy, before the zero vector determination in step 102 (step 101). Further, when it is determined that the inter / inter determination in step 103 is inter, a motion vector is further detected with half-pixel accuracy or ¼-pixel accuracy around the motion vector detected in step 101 (step 201). Since the other processing contents in steps 102 to 120 are the same as those in the second embodiment, description thereof is omitted.

  As described above, the motion compensation encoding apparatus 20 according to the fourth embodiment first detects a motion vector in pixel units, that is, in pixel accuracy, and is determined to be inter by the intra / inter determination in step 103. Only when the detected motion vector is used, the motion vector is further detected with half-pixel accuracy or ¼-pixel accuracy, centering on the motion vector detected with pixel accuracy. Accordingly, it is possible to shorten the processing time of motion vector detection compared to a case where a motion vector is detected only with half pixel accuracy or 1/4 pixel accuracy. In addition, since the motion compensation coding apparatus 20 according to the fourth embodiment performs the detection of the motion vector with the pixel accuracy, and further performs the detection with the half pixel accuracy or the quarter pixel accuracy, the motion vector is detected only with the pixel accuracy. There is also an effect that the motion vector detection accuracy is improved as compared with the case where it is performed.

Embodiment 5 FIG.
FIG. 9 shows an encoding process flow of the motion compensation encoding method described in the fifth embodiment. In the present embodiment, the motion compensated predictor 11 described in the third embodiment will be described with reference to an operation for performing pre-zero vector determination before motion vector detection for the purpose of reducing the overall processing time. In FIG. 9, the same processing steps as those in FIG. 7 are denoted by the same reference numerals.

  The motion compensated predictor 11 according to the present embodiment first performs pre-zero vector determination in step 301 before the first motion vector detection with pixel accuracy in step 101, and sets the motion vector of the predicted image as a zero vector. It is determined whether or not. If it is determined that “the motion vector is a zero vector”, the processing of steps 101 and 102 is not performed, and the motion vector is set to a zero vector in step 105. Hereinafter, the process of step 301 will be described.

  In the pre-zero vector determination in step 301, the first evaluation value S1 that is the difference value between the first predicted image and the input macroblock is used, and the first evaluation value S1 is smaller than a predetermined fixed value ε. Sometimes, it is determined that “the motion vector is set to a zero vector”, and in other cases, it is determined that “the motion vector is not set to a zero vector”.

That is,
S1 = SAD (0,0)
In the first evaluation value S1 calculated by
S1 <ε
In this case, it is determined that “the motion vector is set to a zero vector”, and the process proceeds to step 105 where the motion vector is set to the zero vector. On the other hand, in cases other than the above, it is determined that “the motion vector is not set to the zero vector”, and the process proceeds to step 101 to perform the first motion vector detection. Note that the fixed value ε is a value determined by the code bit rate, the size of the macroblock, and the like.

  Further, after the motion vector is detected in step 101, zero vector determination is performed in step 102. Since the processing contents after step 102 are the same as those in the third embodiment, the description thereof will be omitted.

  As described above, the motion compensated predictor 11 according to the fifth embodiment first performs the pre-zero vector determination before the first motion vector detection, and determines that “the motion vector is set to the zero vector”. If not, the process proceeds to step 105 without performing the processing of steps 101 and 102, and the zero vector is set. Therefore, the overall processing time can be shortened. In particular, when the motion compensation encoding apparatus including the motion compensation predictor 11 described in the fifth embodiment is applied to an apparatus that transmits a video signal in which most of the image is stationary, such as a video phone, The time can be greatly shortened, and a great effect can be obtained.

  Further, although the fifth embodiment has been described based on the motion compensated predictor 11 described in the third embodiment, it can also be applied to the motion compensated predictor 11 described in the first embodiment. That is, in the processing flow shown in FIG. 3, a processing flow in which step 301 is added before step 101 may be used.

Embodiment 6 FIG.
An encoding process flow of the motion compensation encoding method described in the sixth embodiment is shown in FIGS. 10.1 and 10.2. In the present embodiment, an operation for performing pre-zero vector determination before motion vector detection is described for the purpose of reducing the overall processing time in the motion compensation coding apparatus 20 described in the fourth embodiment. In FIG. 10.1 and FIG. 10.2, the same reference numerals are assigned to the same processing steps as those in FIG. 8.1 and FIG.

  The motion compensation encoding apparatus 20 in the present embodiment first performs the pre-zero vector determination in step 301 before the first motion vector detection with pixel accuracy in step 101, and determines the motion vector of the predicted image as the zero vector. It is determined whether or not to do so. If it is determined that “the motion vector is set to the zero vector”, the processing of steps 101 and 102 is not performed, and the motion vector is set to the zero vector in step 105. Note that the content of the processing in step 301 is the same as that in the fifth embodiment, and thus description thereof is omitted.

  As described above, the motion compensation encoding apparatus 20 according to the sixth embodiment first performs pre-zero vector determination before determining the first motion vector, and determines that “the motion vector is set to the zero vector”. If so, the process proceeds to step 105 without performing the processing of steps 101 and 102, and the zero vector is set. Therefore, the overall processing time can be shortened. In particular, if the motion compensation encoding apparatus 20 described in the sixth embodiment is applied to an apparatus that transmits a video signal in which most of the image is stationary, such as a video phone, the processing time is greatly reduced. Can be obtained, and a great effect can be obtained.

  Further, although the sixth embodiment has been described based on the motion compensation encoding apparatus 20 described in the fourth embodiment, it can also be applied to the motion compensation encoding apparatus 20 described in the first embodiment. That is, in the processing flow shown in FIG. 6.1, a processing flow in which step 301 is added before step 101 may be used.

6 is a conceptual diagram illustrating a video signal input to the motion compensation encoding apparatus according to Embodiment 1. FIG. 1 is a block diagram showing a configuration of a motion compensation encoding apparatus in Embodiment 1. FIG. 6 is a diagram showing a processing flow of a motion compensated predictor in Embodiment 1. FIG. 3 is a block diagram illustrating a structure of a motion compensated predictor according to Embodiment 1. FIG. It is explanatory drawing explaining a motion vector. FIG. 10 is a diagram illustrating a processing flow of the motion compensation encoding apparatus according to the second embodiment. FIG. 10 is a diagram illustrating a processing flow of the motion compensation encoding apparatus according to the second embodiment. FIG. 10 is a diagram showing a processing flow of a motion compensated predictor in the third embodiment. [Fig. 20] Fig. 20 is a diagram illustrating a process flow of the motion compensation encoding apparatus according to the fourth embodiment. [Fig. 20] Fig. 20 is a diagram illustrating a process flow of the motion compensation encoding apparatus according to the fourth embodiment. FIG. 10 is a diagram showing a processing flow of a motion compensated predictor in the fifth embodiment. FIG. 25 is a diagram showing a processing flow of the motion compensation encoding apparatus in the sixth embodiment. FIG. 25 is a diagram showing a processing flow of the motion compensation encoding apparatus in the sixth embodiment.

Explanation of symbols

  1 differencer, 2 switch, 3 DCT conversion means, 4 quantizer, 5 variable length coding means, 6 inverse quantizer, 7 inverse DCT conversion means, 8 adder, 9 switch, 10 memory, 11 motion Compensation predictor, 12 motion vector detection means, 13 zero vector determination means, 14 intra / inter determination means, 15 encoding determination means, 16 predicted image creation means, 20 motion compensation encoder.

Claims (7)

A motion compensation encoding device that divides an input video signal into a plurality of blocks, performs encoding processing in units of the blocks, and outputs a bitstream obtained by the encoding processing,
A motion vector detecting means for detecting a motion vector corresponding to a block in the inputted video signal from the decoded video signal obtained by decoding the already encoded video signal and the inputted video signal;
A first evaluation value for evaluating a difference between a first predicted image generated based on the decoded video signal and the block when the motion vector is a zero vector, and the decoded video signal and the motion vector; A second evaluation value for evaluating a difference between the second predicted image generated based on the motion vector detected by the detection unit and the block is obtained, and the motion vector is calculated based on the two evaluation values. Zero vector determination means for determining whether or not to make a zero vector;
When it is determined that the motion vector is a zero vector, an encoding determination unit that determines whether or not to encode a difference between the first predicted image and the block,
The motion compensation encoding apparatus according to claim 1, wherein, in the encoding determination unit, when it is determined that the difference between the first predicted image and the block is not to be encoded, the block is set as a predicted image. The motion vector detection means detects the motion vector with pixel accuracy,
The zero vector determination means further comprises second motion vector detection means for detecting a motion vector with an accuracy equal to or lower than a pixel accuracy when the motion vector is determined not to be a zero vector. The motion compensation encoding device described. The pre-zero vector detecting means for determining whether or not the motion vector is set to a zero vector from the first evaluation value before detecting the motion vector by the motion vector detecting means. 2. A motion compensation encoding apparatus according to 1. Zero vector determination means
4. The method according to claim 1, wherein the motion vector is determined to be a zero vector when the first evaluation value is smaller than the sum of the second evaluation value and a predetermined value. 5. 2. A motion compensation encoding apparatus according to 1. The pre-zero vector detection means is
The motion compensation coding apparatus according to claim 3 or 4, wherein when the first evaluation value is smaller than a predetermined value, the motion vector is determined to be a zero vector. 6. The motion compensation encoding apparatus according to claim 1, wherein a bit stream including a determination result of the encoding determination unit is output. A motion compensation encoding method that divides an input video signal into a plurality of blocks, performs encoding processing in units of the blocks, and outputs a bitstream obtained by the encoding processing,
A motion vector detection step of detecting a motion vector corresponding to a block in the input video signal from the decoded video signal obtained by decoding the already encoded video signal and the input video signal;
A first evaluation value for evaluating a difference between a first predicted image generated based on the decoded video signal and the block when the motion vector is a zero vector, and the decoded video signal and the motion vector; A second evaluation value for evaluating a difference between the second predicted image generated based on the motion vector detected in the detection step and the block is obtained, and the motion vector is calculated based on the two evaluation values. A zero vector determination step for determining whether to make a zero vector;
An encoding determination step for determining whether or not to encode a difference between the second predicted image and the block when it is determined in the zero vector determination step that the motion vector is a zero vector;
A motion-compensated encoding method, wherein, in the encoding determination step, when it is determined that the difference between the second predicted image and the block is not to be encoded, the block is set as a predicted image.

Images