JP2005167852A - Method and apparatus for detecting motion vector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for detecting a motion vector by which a processing time can be shortened without degrading detection accuracy. <P>SOLUTION: The computation of an evaluation value for the integral pixel accuracy of an encoding block and candidate blocks by an integral pixel accuracy motion vector detection section 401 is performed simultaneously with the computation of an evaluation value for the decimal pixel accuracy of the encoding block, the candidate blocks and nearby blocks by a decimal pixel accuracy motion vector detection section 402. With regard to the candidate blocks determined by the evaluation value obtained from the integral pixel accuracy motion vector detection section 401, an integral pixel accuracy motion vector is computed by an optimal motion vector determination section 403, and a decimal pixel accuracy motion vector is computed based on the evaluation value obtained from the decimal pixel accuracy motion vector detection section 402 corresponding to the determined candidate blocks. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a motion vector detection method and a motion vector detection device used in motion compensation encoding in moving image compression encoding, and more particularly to a motion vector detection method and motion vector detection device using a block matching method. It is.

  In order to transfer and store a moving image signal, an image compression technique that reduces the data amount of the image signal having a large amount of data is indispensable. In general, an image signal includes high redundancy, in other words, information that can be reduced. One of the methods for reducing the redundancy is an inter-frame prediction encoding method. However, this method can achieve high image code compression for an image with small motion, but a large amount of code is generated for an image with large motion because the correlation between frames is low. As a method for solving such a problem, there is a motion compensation frame predictive coding method using a motion vector. In this method, there is a block matching method as one method for extracting a motion vector.

  FIG. 5 is a diagram for explaining the block matching method. As shown in FIG. 5, the block matching method is composed of N × M (N and M are integers) pixels of the current frame, that is, the encoded frame 51. The coded block 52 and the coded frame 51 are divided into a plurality of candidate blocks composed of N × M pixels in the search area 54 set on a reference frame 53 that is temporally different from the coded frame 51, Processing for obtaining a correlation between pixels located at the same position in each block, that is, an absolute value of a difference between pixel values, and accumulating these values, with each candidate block 55 composed of N × M pixels Is performed on all candidate blocks 55 in the search area, and the value obtained by cumulative addition, that is, the candidate block with the smallest evaluation value and the encoded block 52 projected onto the reference frame 53 It is a method of detecting a displacement motion vectors between 56.

MPEG2 motion vector detection algorithms include two-stage motion vector detection, ie, motion vector detection with integer pixel accuracy and motion vector detection with decimal pixel accuracy. In this motion vector detection, as a conventional motion vector detection device that realizes a reduction in processing time by performing motion vector detection in two stages, for example, there is a motion vector detection device as disclosed in Patent Document 1. Specifically, the first-stage 1-pixel accuracy search unit performs motion vector detection by the block matching method, and the second-stage decimal pixel accuracy search unit includes a plurality of 1's that are close to the motion vector obtained in the 1st stage. The decimal precision motion vector is determined using a value obtained by multiplying the sum of the prediction error evaluation values of the pixel accuracy candidate vectors by a predetermined coefficient as a prediction error evaluation value of the decimal precision candidate vector.
Japanese Patent No. 2596352

  The conventional motion vector detection method and the motion vector detection apparatus are configured as described above. In motion vector detection using the block matching method, a candidate block at the position with the highest degree of correlation is detected, and this candidate block is detected. By determining the motion vector based on the above, a highly accurate motion vector is detected. In the above-described conventional motion vector detection device, in the decimal pixel accuracy motion vector detection by the second-stage decimal pixel accuracy search unit, block matching is not performed and the processing time until the decimal pixel accuracy motion vector is determined is shortened. However, since the decimal pixel accuracy motion vector is detected based on the candidate block estimated from the first-stage integer pixel accuracy motion vector detection result, block matching is also performed in the decimal pixel accuracy motion vector detection. There is a problem in that a motion vector with poor accuracy may be detected in contrast to the decimal pixel accuracy motion vector detection performed by the original block matching method.

  The present invention has been made to solve the above-described problems, and can reduce the processing time until a decimal pixel precision motion vector is detected and can also detect the decimal pixel precision motion vector. It is an object of the present invention to provide a motion vector detection method and a motion vector detection device that can prevent a decrease in the motion vector.

  The motion vector detection method according to the present invention (Claim 1) shows a degree of correlation with integer pixel accuracy between a coded block of a coded frame image and a plurality of candidate blocks in a search region of a reference frame image. An integer pixel accuracy evaluation value calculating step for calculating an evaluation value; and the encoding block at the pixel positions in the vicinity of the plurality of candidate blocks and the plurality of candidate blocks. Between the coding block based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculating step and the integer pixel accuracy evaluation value calculating step for calculating the evaluation value indicating the degree of correlation of the decimal pixel accuracy. For a candidate block determined to be high, an integer pixel precision motion vector for the coding block is calculated and this determined Calculating a decimal pixel precision motion vector for the coding block based on the evaluation value obtained by the decimal pixel precision evaluation value calculation step corresponding to the complementary block, and determining a motion vector. .

  In addition, the motion vector detection method according to the present invention (Claim 2) provides a correlation degree with integer pixel accuracy between a coded block of a coded frame image and a plurality of candidate blocks in a search region of a reference frame image. An integer pixel accuracy evaluation value calculation step for calculating an evaluation value indicating the value and a representative block consisting of N × M (N and M are integers) pixels in the coding block, which are executed simultaneously with the integer pixel accuracy evaluation value calculation step A decimal pixel accuracy evaluation value calculation step for calculating an evaluation value indicating a degree of correlation of decimal pixel accuracy between the block corresponding to the representative block of the plurality of candidate blocks, and the integer pixel accuracy evaluation value calculation Integer pixel accuracy for the coding block for a candidate block determined to have the highest correlation with the coding block based on the evaluation value obtained in the step A motion vector is calculated, and a decimal pixel accuracy motion vector for the coding block is calculated based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step corresponding to the determined candidate block. Determining a motion vector.

  The motion vector detection method according to the present invention (Claim 3) is the motion vector detection method, wherein the integer pixel accuracy motion vector detection evaluation value is calculated in the integer pixel accuracy evaluation value calculation step. When the evaluation value exceeds a preset threshold value, the step of stopping the calculation of the evaluation value, and the time when the calculation of the evaluation value of the integer pixel accuracy evaluation value calculation step is stopped, the decimal pixel accuracy evaluation And a step of stopping the calculation of the evaluation value in the value calculation step.

  In the motion vector detection method according to the present invention (claim 4), in the motion vector detection method, the step of determining the motion vector is a decimal number calculated from the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step. The number of occurrences in the direction of the pixel accuracy motion vector is integrated, and one of the decimal pixel accuracy motion vectors is selected based on the integration result, thereby obtaining a motion vector with decimal pixel accuracy.

  The motion vector detection method according to the present invention (Claim 5) is characterized in that, in the motion vector detection method, a characteristic pixel value of the coding block is used as a representative block of N × M pixels in the coding block. The block to be included is selected.

  In addition, the motion vector detection device according to the present invention (Claim 6) provides a correlation degree with integer pixel accuracy between a coded block of a coded frame image and a plurality of candidate blocks in a search region of a reference frame image. An integer pixel precision motion vector detection unit that calculates an evaluation value indicating a correlation degree of decimal pixel precision between the plurality of candidate blocks and the pixel block in the vicinity of the plurality of candidate blocks. Based on the evaluation value obtained from the integer pixel accuracy motion vector detection unit and the decimal pixel accuracy motion vector detection unit that executes the calculation of the evaluation value to be performed simultaneously with the calculation of the evaluation value by the integer pixel accuracy motion vector detection unit For a candidate block determined to have the highest correlation with the encoded block, an integer pixel precision motion vector for the encoded block is calculated. And an optimal motion vector determination unit that calculates a motion vector with decimal pixel accuracy for the encoded block based on an evaluation value obtained from the decimal pixel accuracy motion vector detection unit corresponding to the determined candidate block. It is what I did.

  In addition, the motion vector detection device according to the present invention (Claim 7) obtains a correlation degree of integer pixel accuracy between a coded block of a coded frame image and a plurality of candidate blocks in a search region of a reference frame image. Corresponding to an integer pixel precision motion vector detection unit for calculating an evaluation value to be indicated, a representative block composed of N × M pixels (N and M are integers) in the encoded block, and the representative block in the plurality of candidate blocks A fractional pixel precision motion vector detection unit that executes computation of an evaluation value indicating a degree of correlation of decimal pixel precision with the block to be performed simultaneously with computation of an evaluation value by the integer pixel precision motion vector detection unit, and the integer pixel A candidate block that is determined to have the highest correlation with the encoded block based on an evaluation value obtained from an accuracy motion vector detection unit, with respect to the encoded block Calculates a motion vector of several pixel accuracy and calculates a motion vector of decimal pixel accuracy for the encoded block based on an evaluation value obtained from the decimal pixel accuracy motion vector detection unit corresponding to the determined candidate block. And an optimal motion vector determination unit.

  In the motion vector detection device according to the present invention (claim 8), in the motion vector detection device, the integer pixel accuracy motion vector detection unit may be configured to calculate an evaluation value for detecting the integer pixel accuracy motion vector. When the evaluation value exceeds a preset threshold value, the calculation of the evaluation value is stopped, and the decimal pixel accuracy motion vector detection unit calculates the evaluation value by the integer pixel accuracy motion vector detection unit. The evaluation value is stopped when the calculation is stopped.

  In the motion vector detection device according to the present invention (claim 9), in the motion vector detection device, the optimum motion vector determination unit is a decimal pixel calculated from an evaluation value obtained from the decimal pixel precision motion vector detection unit. The number of occurrences in the direction of the precision motion vector is integrated, and one of the decimal pixel precision motion vectors is selected based on the integration result, thereby obtaining a motion vector with decimal pixel precision.

  According to the motion vector detection method of the present invention (Claim 1), the degree of correlation with integer pixel accuracy between an encoded block of an encoded frame image and a plurality of candidate blocks within a search region of a reference frame image An integer pixel accuracy evaluation value calculation step for calculating an evaluation value indicating the same, and the encoding at the pixel positions in the vicinity of the plurality of candidate blocks and the plurality of candidate blocks. A decimal pixel accuracy evaluation value calculation step for calculating an evaluation value indicating a degree of correlation of decimal pixel accuracy with the block, and the encoding block based on the evaluation value obtained by the integer pixel accuracy evaluation value calculation step; For a candidate block determined to have the highest correlation, an integer pixel precision motion vector for the coded block is calculated and this determination is performed. Calculating a decimal pixel accuracy motion vector for the coding block based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step corresponding to the candidate block, and determining a motion vector; By including, there is an effect that it is possible to shorten the processing time required to detect the optimum motion vector with the decimal pixel accuracy without reducing the detection accuracy.

  Further, according to the motion vector detection method of the present invention (Claim 2), an integer pixel accuracy between an encoded block of an encoded frame image and a plurality of candidate blocks in a search region of a reference frame image is obtained. An integer pixel accuracy evaluation value calculation step for calculating an evaluation value indicating the degree of correlation and N × M (N and M are integers) pixels in the coding block are executed simultaneously with the integer pixel accuracy evaluation value calculation step. A decimal pixel accuracy evaluation value calculating step for calculating an evaluation value indicating a degree of correlation of decimal pixel accuracy between a representative block and a block corresponding to the representative block among the plurality of candidate blocks, and the integer pixel accuracy evaluation For a candidate block that is determined to have the highest correlation with the encoded block based on the evaluation value obtained in the value calculation step, an integer image for the encoded block A motion vector with accuracy is calculated, and a motion vector with decimal pixel accuracy for the coding block is calculated based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step corresponding to the determined candidate block. Thus, there is an effect that it is possible to reduce the amount of calculation required for the evaluation value calculation for sub-pixel precision motion vector detection.

  According to the motion vector detection method of the present invention (Claim 3), in the motion vector detection method, during the calculation of the evaluation value for detecting the integer pixel accuracy motion vector in the integer pixel accuracy evaluation value calculation step. When the evaluation value exceeds a preset threshold, the step of stopping the calculation of the evaluation value, and the time when the calculation of the evaluation value in the integer pixel accuracy evaluation value calculation step is stopped, the decimal pixel By including the step of stopping the calculation of the evaluation value in the accuracy evaluation value calculation step, there is an effect that the power consumption of the circuit due to useless calculation can be suppressed.

  According to the motion vector detection method of the present invention (Claim 4), in the motion vector detection method, the step of determining the motion vector is calculated from the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step. The number of occurrences in the direction of the sub-pixel accuracy motion vector is integrated, and one of the sub-pixel accuracy motion vectors is selected based on the integration result, thereby obtaining a sub-pixel accuracy motion vector. There is an effect that the processing time required to detect the optimal motion vector with the decimal pixel accuracy can be shortened without reducing the accuracy.

  According to the motion vector detection method of the present invention (Claim 5), in the motion vector detection method, a characteristic pixel of the coding block is used as a representative block of N × M pixels in the coding block. By selecting a block including a value, there is an effect that the power consumption of the circuit due to useless computation can be suppressed.

  Further, according to the motion vector detection device of the present invention (Claim 6), the integer pixel accuracy between the encoded block of the encoded frame image and the plurality of candidate blocks in the search region of the reference frame image is improved. Correlation with decimal pixel accuracy between the integer pixel accuracy motion vector detection unit for calculating the evaluation value indicating the degree of correlation, and the plurality of candidate blocks and the encoding block at the pixel positions in the vicinity of the plurality of candidate blocks. Based on the evaluation value obtained from the integer pixel accuracy motion vector detection unit and the decimal pixel accuracy motion vector detection unit that executes the calculation of the evaluation value indicating the degree simultaneously with the calculation of the evaluation value by the integer pixel accuracy motion vector detection unit For the candidate block determined to have the highest correlation with the coded block, an integer pixel precision motion vector for the coded block is calculated. And an optimal motion vector determination unit that calculates a motion vector with decimal pixel accuracy for the encoded block based on an evaluation value obtained from the decimal pixel accuracy motion vector detection unit corresponding to the determined candidate block. By doing so, there is an effect that the processing time required to detect the optimal motion vector with decimal pixel accuracy can be shortened without reducing the detection accuracy.

  Further, according to the motion vector detection device of the present invention (Claim 7), the correlation of integer pixel accuracy between the encoded block of the encoded frame image and a plurality of candidate blocks in the search region of the reference frame image. An integer pixel precision motion vector detection unit for calculating an evaluation value indicating degree, a representative block composed of N × M pixels (N and M are integers) in the encoded block, and the representative block in the plurality of candidate blocks A fractional pixel precision motion vector detection unit that executes computation of an evaluation value indicating a correlation degree of fractional pixel precision with the block corresponding to the same as the evaluation value computation by the integer pixel precision motion vector detection unit, and The candidate block determined to have the highest correlation with the encoded block based on the evaluation value obtained from the integer pixel precision motion vector detection unit is assigned to the encoded block. Calculating an integer pixel precision motion vector, and calculating a decimal pixel precision motion vector for the encoded block based on an evaluation value obtained from the decimal pixel precision motion vector detection unit corresponding to the determined candidate block. By including the optimum motion vector determination unit for calculation, it is possible to reduce the amount of calculation required for the evaluation value calculation for sub-pixel accuracy motion vector detection.

  According to the motion vector detection device of the present invention (claim 8), in the motion vector detection device, the integer pixel accuracy motion vector detection unit is calculating an evaluation value for detecting the integer pixel accuracy motion vector. When the evaluation value exceeds a preset threshold value, the calculation of the evaluation value is stopped, and the decimal pixel accuracy motion vector detection unit calculates the evaluation value using the integer pixel accuracy motion vector detection. When the calculation of the evaluation value of the unit is stopped, there is an effect that the power consumption of the circuit due to useless calculation can be suppressed.

  According to the motion vector detection device of the present invention (Claim 9), in the motion vector detection device, the optimum motion vector determination unit calculates from the evaluation value obtained from the decimal pixel precision motion vector detection unit. By integrating the number of occurrences in the direction of the decimal pixel accuracy motion vector and selecting one of the decimal pixel accuracy motion vectors based on the integration result, the detection accuracy is obtained by obtaining the motion vector with the decimal pixel accuracy. There is an effect that it is possible to reduce the processing time required to detect the optimal motion vector with decimal pixel accuracy without dropping the value.

(Embodiment 1)
FIG. 4 is a block diagram showing the configuration of the motion vector detection apparatus according to Embodiment 1 of the present invention. In the figure, the motion vector detection apparatus includes an integer pixel precision motion vector detection unit 401 and a decimal pixel precision motion vector detection. Unit 402 and an optimal motion vector determination unit 403.

  The integer pixel accuracy motion vector detection unit 401 receives a plurality of candidate block image data 405 and encoded block image data 404 in the search region, and calculates an evaluation value for integer pixel accuracy motion vector detection from these data. . The decimal pixel accuracy motion vector detection unit 402 receives a plurality of candidate blocks in the search region and image data (candidate block image data) 405 and encoded block image data 404 in the vicinity of the candidate blocks, and inputs the decimal pixel accuracy motion from these. The evaluation value for vector detection is simultaneously calculated in parallel with the integer pixel precision motion vector detection unit 401. Further, the optimum motion vector determination unit 403 holds the evaluation values obtained from the integer pixel accuracy motion vector detection unit 401 and the decimal pixel accuracy motion vector detection unit 402 in one-to-one correspondence with each candidate block. Then, a candidate block having a high degree of correlation with the encoded block is determined from the evaluation values for all candidate blocks in the search region obtained from the integer pixel precision motion vector detecting unit 401, and the candidate block and the encoded block are An optimal integer pixel precision motion vector is calculated, and an optimal decimal pixel precision motion vector is calculated based on the evaluation value obtained from the decimal pixel precision motion vector detection unit 402 corresponding to the candidate block.

  FIG. 1 is a diagram for explaining a motion vector detection method according to Embodiment 1 of the present invention. In the reference frame, a search region 100, a position 103 corresponding to a coding block, and integer pixel accuracy candidates Blocks 101 and 102 are shown. In the figure, circles indicate integer pixels.

  FIG. 2 is a diagram for explaining a motion vector detection method with decimal pixel accuracy in the motion vector detection method according to the first embodiment of the present invention, in which an integer pixel and a decimal pixel in the vicinity of a coding block (shaded area) are shown. It shows. In the figure, white circles indicate integer pixels, and black circles indicate decimal pixels.

  Hereinafter, the detection method will be described. For the candidate block 101 at the integer pixel accuracy position in the search region 100, a correlation degree for detecting an integer pixel accuracy motion vector, here, an evaluation value A is calculated by block matching with the coding block. Similarly, an evaluation value B for integer pixel precision motion vector detection is also calculated for the candidate block 102. At the same time, in the vicinity of the candidate block 101 and the candidate block 102, as shown in FIG. 2, a block with an encoding block is applied to a pixel position with decimal pixel accuracy generated by interpolation from adjacent pixel data with integer pixel accuracy. Evaluation values for decimal pixel precision motion vector detection are respectively calculated by matching.

  Then, the decimal pixel precision motion vector is determined based on the evaluation value obtained for detecting the integer pixel precision motion vector and the evaluation value obtained for detecting the decimal pixel precision motion vector. That is, if the evaluation value A for determining the integer pixel accuracy motion vector for the candidate block 101 is smaller than the evaluation value B for determining the integer pixel accuracy motion vector for the candidate block 102, that is, the degree of correlation is high, A decimal pixel accuracy motion vector is determined from an evaluation value for detecting a decimal pixel accuracy motion vector for the candidate block 101 obtained by simultaneous calculation.

  These processes are executed for all candidate blocks in the search region 100, and from the results, the evaluation value obtained for determining the integer pixel precision motion vector is the smallest, that is, the optimum from the position of the candidate block having the highest correlation. An integer pixel precision motion vector is determined, and a decimal pixel precision motion vector of the candidate block obtained by simultaneous calculation is determined as an optimal decimal pixel motion vector.

  FIG. 3 is a diagram for explaining a modification of the method for determining a decimal pixel precision motion vector in the first embodiment, and in the figure, circles indicate pixels. In the first embodiment, as shown in FIG. 3, n × m (1 ≦ n ≦ N, 1 ≦ n) in an N × M pixel coding block is used as an evaluation value for decimal pixel precision motion vector detection. m ≦ M n, m is an integer) representative block and a representative block consisting of n × m (1 ≦ n ≦ N, 1 ≦ m ≦ M n, m is an integer) pixels in a candidate block of N × M pixels. You may make it use the evaluation value calculated by the block matching method between the blocks.

  As the representative block of n × m pixels shown in FIG. 3 above, for example, a block including a pixel having the largest pixel value, the smallest pixel value, or a pixel having a large difference in pixel values in adjacent pixels, etc. The block of the characteristic part may be selected as the representative block.

  As described above, according to the first embodiment, in the search region 100, integer pixel precision motion vector detection and decimal pixel precision motion vector detection using the block matching method are simultaneously performed. Decimal pixel precision motion vectors are determined from evaluation values simultaneously calculated for decimal pixel precision motion vector detection for candidate blocks (102, 103) evaluated to have a high degree of correlation with the coding block 103. By doing so, it is possible to prevent a decrease in the precision of decimal pixel precision motion vector detection, and to perform an integer pixel precision motion vector detection and a decimal pixel precision motion vector detection simultaneously to obtain an optimal motion vector with a decimal pixel precision. The processing time required until detection can be shortened.

  In the first embodiment, the optimum motion vector determination unit 403 uses the evaluation value obtained by the decimal pixel accuracy motion vector detection unit 402 corresponding to the evaluation value obtained by the integer pixel accuracy motion vector detection unit 401. Although one-to-one holding is performed, the decimal pixel accuracy motion vector corresponding to the evaluation value obtained by the decimal pixel accuracy motion vector detection unit 402 is integrated for each direction, and finally the direction of the most accumulated direction is accumulated. The evaluation value corresponding to the vector may be determined as the optimum decimal pixel precision motion vector.

(Embodiment 2)
Next, a motion vector detection method and a motion vector detection device according to Embodiment 2 of the present invention will be described. FIG. 6 is a block diagram showing the configuration of the motion vector detection device according to the second embodiment of the present invention. The motion vector detection device according to the second embodiment is the motion vector detection device according to the first embodiment described above. , A threshold value setting unit 606 for setting a threshold value set in advance based on a test or the like is provided, and instead of the integer pixel accuracy motion vector detection unit 401, an integer for which the threshold value is set by receiving the output of the threshold value setting unit 606 is provided. A pixel precision motion vector detection unit 601 is provided, and instead of the decimal pixel precision motion vector detection unit 402, a decimal pixel precision motion that stops computation upon receiving a computation stop signal output from the integer pixel precision motion vector detection unit 601 A vector detection unit 602 is provided.

  More specifically, in FIG. 6, the same reference numerals as those in FIG. 4 indicate the same or corresponding parts, and the integer pixel precision motion vector detection unit 601 includes a plurality of candidate block image data 405 and encoded block images in the search area. The data 404 is input, and an evaluation value for detecting an integer pixel accuracy motion vector is calculated from the data 404. When the evaluation value being calculated exceeds the threshold set by the threshold setting unit 606, the calculation is stopped. To do. At the same time, an operation stop signal is output to the decimal pixel precision motion vector detection unit 602.

  The decimal pixel accuracy motion vector detection unit 602 receives a plurality of candidate blocks in the search region and the image data 405 and encoded block image data 404 in the vicinity thereof, and uses them as evaluation values for detecting the decimal pixel accuracy motion vector. In parallel with the integer pixel precision motion vector detection unit 601, the calculation is performed simultaneously. In addition, when an operation stop signal is given from the integer pixel precision motion vector detection unit 601, the operation is stopped.

  As described above, in the second embodiment, the integer pixel accuracy motion vector detection unit 601 and the fractional pixel accuracy motion vector detection unit 602 each include an integer pixel as in the motion vector detection device according to the first embodiment described above. The evaluation value for detecting the precision motion vector and the evaluation value for detecting the decimal pixel precision motion vector are simultaneously calculated in parallel, and the threshold value setting unit 606 is used when the integer pixel accuracy motion vector detection unit 601 calculates the evaluation value. When the evaluation value higher than the threshold value set by is accumulated, the calculation is stopped, and the integer pixel accuracy motion vector detection unit 601 outputs an operation stop signal to the decimal pixel accuracy motion vector detection unit 602. Thereby, the calculation of the evaluation value by the decimal pixel precision motion vector detection unit 602 is also stopped. As a result, during the calculation of the evaluation value, it can be determined that the evaluation value for the coding block is too high, that is, the degree of correlation is low, and the wasteful computation can be aborted halfway. Can be suppressed.

  INDUSTRIAL APPLICABILITY The motion vector detection method and motion vector detection apparatus according to the present invention are useful for an encoder that encodes an image, an apparatus having an encoder, and the like, and in particular, a DVD video recorder that encodes an image by MPEG2, a hard disk recorder, and the like Useful for.

It is a figure for demonstrating the integer pixel precision motion vector detection method in the motion vector detection method which concerns on Embodiment 1 of this invention. It is a figure for demonstrating the decimal pixel precision motion vector detection method in the motion vector detection method which concerns on the said Embodiment 1. FIG. It is a figure for demonstrating the decimal pixel precision motion vector detection method using the representative block in the motion vector detection method which concerns on the said Embodiment 1. FIG. It is a figure which shows the structure of the motion vector detection apparatus which concerns on the said Embodiment 1. FIG. It is a figure for demonstrating the conventional block matching method. It is a figure which shows the structure of the motion vector detection apparatus which concerns on Embodiment 2 of this invention.

Explanation of symbols

100 Search area 101, 102 Candidate block 103 Coding block position 401, 601 Integer pixel accuracy motion vector detection unit 402, 602 Decimal pixel accuracy motion vector detection unit 403 Optimal motion vector determination unit 404 Encoding block data 405 Candidate block image data 606 Threshold setting unit

Claims (9)

An integer pixel accuracy evaluation value calculating step for calculating an evaluation value indicating a correlation degree of integer pixel accuracy between the encoding block of the encoded frame image and a plurality of candidate blocks in the search region of the reference frame image;
An evaluation value that is executed simultaneously with the integer pixel accuracy evaluation value calculating step and that indicates a correlation degree with decimal pixel accuracy between the plurality of candidate blocks and the pixel block in the vicinity of the plurality of candidate blocks. A decimal pixel accuracy evaluation value calculation step to be calculated;
For a candidate block determined to have the highest correlation with the encoded block based on the evaluation value obtained by the integer pixel accuracy evaluation value calculating step, an integer pixel accuracy motion vector for the encoded block is calculated. Calculating a decimal pixel accuracy motion vector for the coding block based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step corresponding to the determined candidate block, and determining a motion vector; ,
A motion vector detection method comprising: An integer pixel accuracy evaluation value calculating step for calculating an evaluation value indicating a correlation degree of integer pixel accuracy between the encoding block of the encoded frame image and a plurality of candidate blocks in the search region of the reference frame image;
A block corresponding to the representative block of N × M (N and M are integers) in the encoded block and the representative block in the plurality of candidate blocks, which is executed simultaneously with the integer pixel accuracy evaluation value calculation step A decimal pixel accuracy evaluation value calculating step for calculating an evaluation value indicating the degree of correlation of the decimal pixel accuracy between,
For a candidate block determined to have the highest correlation with the encoded block based on the evaluation value obtained by the integer pixel accuracy evaluation value calculating step, an integer pixel accuracy motion vector for the encoded block is calculated. Calculating a decimal pixel accuracy motion vector for the coding block based on the evaluation value obtained by the decimal pixel accuracy evaluation value calculation step corresponding to the determined candidate block, and determining a motion vector; ,
A motion vector detection method comprising: The motion vector detection method according to claim 1 or 2,
During the calculation of the evaluation value for detecting the integer pixel accuracy motion vector in the integer pixel accuracy evaluation value calculation step, a step of stopping the calculation of the evaluation value when the evaluation value exceeds a preset threshold value When,
When the calculation of the evaluation value in the integer pixel accuracy evaluation value calculation step is stopped, the step of stopping the calculation of the evaluation value in the decimal pixel accuracy evaluation value calculation step;
A motion vector detection method comprising: The motion vector detection method according to claim 1 or 2,
Determining the motion vector comprises:
By integrating the number of occurrences in the direction of the decimal pixel accuracy motion vector calculated from the evaluation value obtained in the decimal pixel accuracy evaluation value calculation step, and selecting one of the decimal pixel accuracy motion vectors based on the integration result, A motion vector detection method characterized by obtaining a motion vector with decimal pixel precision. The motion vector detection method according to claim 2, wherein
A motion vector detection method comprising: selecting a block including a characteristic pixel value of the coding block as a representative block of N × M pixels in the coding block. An integer pixel accuracy motion vector detection unit that calculates an evaluation value indicating a degree of correlation of integer pixel accuracy between the encoding block of the encoding frame image and a plurality of candidate blocks in the search region of the reference frame image;
Evaluation by the integer pixel accuracy motion vector detection unit is performed for calculating an evaluation value indicating a degree of correlation of decimal pixel accuracy with the coding block at the plurality of candidate blocks and pixel positions in the vicinity of the plurality of candidate blocks. A decimal pixel precision motion vector detection unit that is executed simultaneously with a value calculation;
For a candidate block determined to have the highest correlation with the coding block based on the evaluation value obtained from the integer pixel precision motion vector detection unit, an integer pixel precision motion vector for the coding block is calculated, and this An optimal motion vector determination unit that calculates a decimal pixel accuracy motion vector for the coding block based on an evaluation value obtained from the decimal pixel accuracy motion vector detection unit corresponding to the determined candidate block;
A motion vector detection device comprising: An integer pixel accuracy motion vector detection unit that calculates an evaluation value indicating a degree of correlation of integer pixel accuracy between the encoding block of the encoding frame image and a plurality of candidate blocks in the search region of the reference frame image;
The degree of decimal pixel accuracy is shown between a representative block composed of N × M (N and M are integers) pixels in the encoded block and a block corresponding to the representative block in the plurality of candidate blocks. A decimal pixel accuracy motion vector detection unit that executes an evaluation value calculation simultaneously with the evaluation value calculation by the integer pixel accuracy motion vector detection unit;
For a candidate block determined to have the highest correlation with the coding block based on the evaluation value obtained from the integer pixel precision motion vector detection unit, an integer pixel precision motion vector for the coding block is calculated, and this An optimal motion vector determination unit that calculates a decimal pixel accuracy motion vector for the coding block based on an evaluation value obtained from the decimal pixel accuracy motion vector detection unit corresponding to the determined candidate block;
A motion vector detection device comprising: The motion vector detection device according to claim 6 or 7,
The integer pixel precision motion vector detection unit includes:
During the calculation of the evaluation value for detecting the integer pixel precision motion vector, when the evaluation value exceeds a preset threshold, the calculation of the evaluation value is stopped.
The decimal pixel precision motion vector detection unit includes:
The calculation of the evaluation value is stopped when the calculation of the evaluation value of the integer pixel precision motion vector detection unit is stopped.
A motion vector detection apparatus characterized by the above. The motion vector detection device according to claim 6 or 7,
The optimal motion vector determination unit includes:
The number of occurrences in the direction of the decimal pixel accuracy motion vector calculated from the evaluation value obtained from the decimal pixel accuracy motion vector detection unit is integrated, and one of the decimal pixel accuracy motion vectors is selected based on the integration result, A motion vector detection apparatus characterized by obtaining a pixel precision motion vector.

Images