JPH10136373A - Motion detector and method therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform motion detection at a higher speed by stopping the processing of evaluation calculation in an earlier time period for a candidate vector without a possibility to be a motion vector to be obtained. SOLUTION: This motion detector 1 is provided with a CPU 2, a ROM 3 and a RAM 4. The CPU 1 processes data stored in the RAM 4 corresponding to a processing program stored in the ROM 3 and actually performs the processing of the motion detection. In these motion detector and method of the form, before obtaining evaluation to the candidate vector by the arithmetic operation of corresponding picture elements with each other between a current block and a reference block, whether or not an evaluation value to be obtained is to be smaller than the already detected evaluation value is checked. Then, only in the case that the possibility to be a minimum evaluation value is present, the evaluation value is actually obtained. Thus, a processing amount is reduced and the motion vector to one current block is obtained at a high speed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion detecting device and a motion detecting method used for a moving image compression device and the like.

[0002]

2. Description of the Related Art As one of main techniques for efficiently compressing a moving image, there is motion detection for obtaining a motion vector indicating a motion of an image. Several methods for obtaining the motion vector have been proposed. One of the main methods is a method called a block matching algorithm. This block matching algorithm will be described with reference to FIG. In this block matching algorithm, an image to be subjected to motion detection (hereinafter referred to as a current frame) is divided into blocks (hereinafter referred to as a current block) such as 16 pixels × 16 lines, for example. A motion vector is obtained by performing a process for each of them.

That is, for each current block,
A block corresponding to a reference frame based on a given candidate vector (hereinafter referred to as a reference block)
Is determined, a difference between corresponding pixels of the current block and the reference block is obtained, and the difference is integrated by the block. At this time, the difference is integrated as an absolute value.
Then, this integrated value is obtained for all candidate vectors given according to the search range of the reference frame,
A candidate vector with the minimum integrated value is extracted as a motion vector.

The block matching algorithm is expressed by the following equation (2).

(Equation 2) In Equation 2, curr (x, y) is a pixel of the current block, and ref (x + v _x , y + v _y ) is a pixel of the reference block corresponding to the pixel curr (x, y) of the current block. It is determined based on the pixel ref (x, y) of the block and the motion vector (v _x , v _y ). That is, using such an evaluation function, the evaluation value ME of Expression 2 is calculated for various vectors, and the one that minimizes the evaluation value ME is used as the motion vector.

An actual processing method based on the block matching algorithm will be described with reference to FIGS. FIG. 6 is a flowchart in the case where this block matching algorithm is performed by a normal computer device or the like. As shown in FIG. 6, first, the first block is set as a current block from a plurality of blocks obtained by dividing the current frame (step S5).
1) Hereinafter, the processing of motion vector detection for the current block is started.

First, a predetermined sufficiently large value such as a maximum value that the variable can take is set as an initial value in a variable min for storing a minimum evaluation value (step S52), and FIG. As shown, the first candidate vector (v _x1 , v _x1 , v m) of m candidate vectors (v _xi , v _yi ) (i = 1 to m) determined based on a predetermined search range of the current block. v _y1 ) is set as a candidate vector for which an evaluation value is to be obtained (step S5).
3). In the process of calculating the evaluation value for the current block and the candidate vector set as described above, first, after resetting the content of the variable sum for storing the accumulated value (step S54), the first pixel of the current block (see FIG. In the example, the pixel at the upper left) is the pixel to be processed
It is set as curr (x, y) (step S55).

The pixel curr of the current block is
The difference between (x, y) and the pixel ref (x + _vxi , y + _vyi ) on the reference block at the position moved by the candidate vector ( _vx1 , _vy1 ) with respect to the curr (x, y) is _calculated. ,
The difference is added to the value already stored in the variable sum, and stored again in the variable sum (step S56).
Thereafter, while checking whether the pixel to be processed is the last pixel of the current block (the lower right pixel in the example of FIG. 5) (step S58), if it is not the last pixel, the next pixel is set as the pixel to be processed. Is set (step S59), and the process of step S56 is repeated.

If the pixel to be processed is the last pixel in the current block in step S58, the values of the variables sum and min are compared (step S6).
0), whether or not the evaluation value for the candidate vector that has been processed this time and stored in the variable sum is equal to or less than the minimum evaluation value of each candidate vector processed so far that is stored in the variable min, that is, sum ≧ min Check whether or not. And if sum ≧ min,
Since the candidate vector processed this time may be the motion vector to be obtained, this is stored. That is, the evaluation value sum is set to the variable min as a new minimum evaluation value, and the x component and the y component of the candidate vector are stored in the variables minvx and minvy, respectively. In addition,
If sum <min in step S60, there is no possibility that the candidate vector processed this time is a motion vector to be obtained, so that the result of obtaining the evaluation value and the like are ignored, and the process proceeds to the next step S62.

When the processing based on the obtained evaluation value is completed, it is checked whether or not the candidate vector used for performing the processing is the last candidate vector among the m candidate vectors described above ( In step S62, if it is not the last candidate vector, the next vector is set as a candidate vector to be evaluated (step S63), and step S54 is performed again.
To Step S61 are repeated. In step S62, if the candidate vector is the last candidate vector, the variables minvx and min
The candidate vector stored in vy is the motion vector detected for the current block, and the processing for the current block ends.

Then, it is determined whether or not the current block is the last block in the current frame (step S64). If the current block is not the last block, the next block is set as the current block for motion detection (step S65). Then, the processing of steps S52 to S63 is repeated again. If the current block is the last block in the current frame in step S64, a series of processes for the current frame is ended (step S66).

The above-described block matching algorithm can be executed by such basic processing. However, since this algorithm has a large number of nested loops (nested loops), the amount of calculation processing is large. There is a problem that it takes time, and an algorithm that can perform motion detection at higher speed is required. Therefore, a faster processing method obtained by improving the method shown in FIG. 6 will be described with reference to FIG. FIG. 7 is a flowchart showing an algorithm capable of performing the motion detection processing shown in FIG. 6 at higher speed. In the processing shown in FIG. 7, steps that perform exactly the same processing as the processing in FIG. 6 are denoted by the same step numbers, and description thereof will be omitted in principle.

In FIG. 6, steps S56 and S5
8. In S59, the accumulated value of the difference is obtained, but only the smallest accumulated value among the accumulated values is required. Therefore, in the processing procedure of FIG. 7, if there is no expectation of the minimum value, the processing is substantially interrupted by speeding up by exiting the accumulation loop. Therefore, in the processing of FIG. 7, the processing of steps S51 to S56 is the same as the processing of FIG.
In step S57 immediately after step S56, the values of the variables sum and min are compared. That is, in the middle of calculating the accumulated value in order for each pixel in step S56, the current evaluation value of the candidate vector to be processed this time stored in the variable sum is stored in the variable min. It is checked whether or not the sum is equal to or less than the minimum evaluation value in each candidate vector processed up to, that is, whether sum ≧ min.

If sum ≧ min, then
Since there is a possibility that the candidate vector being processed this time is the motion vector to be obtained, the process proceeds to step S58, and the process for obtaining the accumulation for the subsequent pixels is continued. If sum <min, there is no possibility that the candidate vector currently being processed is a motion vector that has already been obtained.
The process moves to the process related to the next candidate vector after step S62.

When the evaluation value is calculated while performing such a check, and it is determined in step S58 that the accumulation has been performed up to the last pixel of the block, the processing target is determined as already determined in step S57. Since the candidate vector of may be the desired motion vector,
Immediately in step 61 these values are stored. That is, the evaluation value sum is set to the variable min as a new minimum evaluation value, and the x component and the y component of the candidate vector are stored in the variables minvx and minvy, respectively. Hereinafter, the processing of steps S62 to S66 is performed in the same manner as the processing shown in FIG.

[0015]

Although the block matching algorithm can be processed at a relatively high speed by using such a processing method as shown in FIG. 7, there is a demand for processing at a higher speed.

Accordingly, it is an object of the present invention to provide a motion detecting device capable of detecting motion at higher speed. It is another object of the present invention to provide a motion detection method capable of performing motion detection at higher speed.

[0017]

In order to solve the above-mentioned problem, it is possible to easily obtain a value that can be taken as an evaluation value for evaluating a candidate vector. Has enabled the evaluation value calculation process to be stopped earlier.

Therefore, the motion detecting apparatus according to the present invention comprises: a candidate vector selecting means for sequentially selecting one from a plurality of candidate vectors; a reference block extracting means for extracting a reference block based on the current block and the candidate vector; First and second sum detection means for detecting the sum of the pixel values of the block and the reference block, and detecting the difference between the sum of the pixel values of the first image block and the sum of the pixel values of the second image block Sum difference detecting means, the first comparing means for comparing the difference between the sum and the minimum evaluation value detected so far, and as a result of the comparison, the difference in the sum of the pixel values is larger than the minimum evaluation value In this case, the processing for the candidate vector is stopped, and if the difference between the sums of the pixel values is equal to or smaller than the minimum evaluation value, the current block and the reference block Evaluation value detection means for detecting the sum of the differences between the pixel values of the pixels as the evaluation value, second comparison means for comparing the evaluation value with the already detected evaluation value, and a result of the comparison and detection. A storage unit that stores the selected candidate vector when the evaluated value is the smallest among the already detected evaluation values, and sets an evaluation value as the minimum evaluation value; and After all of a plurality of predetermined candidate vectors are selected,
Output means for outputting the candidate vector stored in the storage means as a motion vector for the current image block.

In the motion detection method according to the present invention, a current block obtained by dividing a motion detection target image into a region having a predetermined pixel configuration and a given candidate vector which is an image block in a reference image are provided. The sum of the differences between the corresponding pixels of the current block determined with respect to the position of the current block and the reference block having the same pixel configuration is determined based on the current block, and the sum is used as an evaluation value for the candidate vector. A motion vector for detecting a candidate vector having the smallest evaluation value from the plurality of candidate vectors as a motion vector for the current block, wherein a sum of pixel values of the current block is obtained, and a sum of pixel values of the reference block is calculated. Is calculated, and the sum of the pixel values of the current block and the reference block are calculated. Obtains the difference between the sum of the pixel values of the click,
It is determined whether or not the obtained difference is smaller than the evaluation value calculated for the given candidate vector, and if the evaluation result is smaller, the evaluation value for the candidate vector is smaller. Is calculated.

[0020]

1 to 4 show an embodiment of the present invention.
This will be described with reference to FIG. The motion detection device and the motion detection method according to the present embodiment detect the range of the evaluation value of the candidate vector by using the sum of the current block and the reference block, and use the detected range for speeding up the processing. . That is, when the evaluation function as shown in Expression 2 is sequentially transformed as shown in Expression 3,
Finally, a format as shown in Equation 4 can be obtained.

(Equation 3)

From this equation 4, the evaluation value ME as shown in equation 2 for each candidate vector is the absolute value of the difference between the sum of the current block and the reference block (hereinafter referred to as the block sum difference). ). In other words, for a candidate vector whose block sum difference is already larger than the evaluation value for another candidate vector, the evaluation value for that candidate vector is naturally larger than that evaluation value, and the evaluation value is the smallest among all the candidate vectors. That is, the motion vector cannot be obtained. Therefore, in the motion detection device according to the present embodiment, processing on such a candidate vector is immediately terminated, so that a desired vector can be detected at high speed.

FIG. 1 shows a configuration of a motion detecting device for performing such a motion detecting process. As shown in FIG. 1, the motion detection device is configured by a general computer device or the like. That is, the motion detection device 1 includes a central processing unit (CPU) 2,
It has ROM3 and RAM4. The ROM 3 stores a processing program and the like. The RAM 4 stores various data related to the motion detection method, such as image data such as a current frame and a reference frame, candidate vectors, and evaluation values. Then, the CPU 1 processes the data stored in the RAM 2 according to the processing program stored in the ROM 3 and actually performs the motion detection processing.

The motion detection processing actually performed by such a motion detection device will be described with reference to the flowcharts of FIGS. FIG. 2 is a flowchart for explaining the motion detection processing. In the motion detection process, when the process is started (step S10), at the beginning, for each of the current frame and the reference frame to be processed, a process of obtaining the sum of the pixel values for each of the obtained image blocks. (Steps S11 and S12).

The process of obtaining the block sum of the current frame in step S11 will be described in detail with reference to FIG. First, when an input current frame to be processed is divided into regions having a predetermined pixel configuration (in the present embodiment, the current frame is divided into blocks of 16 pixels × 16 lines), the first block is set as a current block ( (Step S31) The content of the variable sum for storing the sum is reset to 0 (step S3).
2). Then, the first pixel of the current block (the upper left pixel in the present embodiment) is set as a pixel to be added (step S33), and after adding the pixel value to the value already stored in the variable sum, The result is stored again in the variable sum (step S34).

Next, it is checked whether or not all the pixels in the current block have been added (step S35). If all the pixels have not been added yet, the next pixel is set as a pixel to be added (step S35). Step S36) In step S34 again, the pixel values are accumulated. Such processing is repeated, and when it is detected in step S35 that the accumulation of all the pixels of the current block is completed, the value stored in the variable sum at that time is the sum of the pixel values of the current block. Is stored in association with the current block (step S37).

Then, it is detected whether or not the processing for obtaining the sum has been completed for all the blocks formed by dividing the current frame (step S38), and the processing has been completed for all the blocks. If not, the next block is set as the current block (step S3).
9) The process of steps S32 to S37 is repeated again. If it is determined in step S38 that the process of calculating the sum of all blocks has been completed, the process of calculating the sum of each block of the current frame is finished (step S40), and the process returns to the main routine.

Next, the process of calculating the sum of each block of the reference frame in step S12 will be described in detail with reference to FIG. Here, for example, when a motion vector search range as shown in FIG. 5 is set for each current block of the current frame,
The same pixel configuration as the current block obtained from the reference frame (16 pixels × 16 pixels in the present embodiment)
All the image blocks of the line) are reference blocks determined based on a combination of the current block and the candidate vector. Therefore, for the reference frame, the block sum is calculated for all the blocks that may become reference blocks.

More specifically, all the image blocks of 16 pixels × 16 lines extracted from the reference frame, that is, all the image blocks which are shifted from each other by only one pixel in the X direction and the Y direction, are block-wise. Find the sum. This means that the processing for obtaining the block sum for each block of the reference frame requires a much larger processing amount than the processing for obtaining the block sum for each current block of the current frame. However, the image blocks in the reference frame are typically referenced many times in each of the various current block and candidate vector combinations,
Obtaining the block sum for all the blocks in advance is still effective.

Therefore, first, a block of 16 pixels × 16 lines obtained first from the input reference frame is set as a reference block (step S4).
1) The content of the variable sum for storing the sum is reset to 0 (step S42). Then, the first pixel (the upper left pixel in the present embodiment) of the reference block is set as a pixel to be added (step S4).
3) After adding the pixel value to the value already stored in the variable sum, the result is stored in the variable sum again (step S44).

Next, it is checked whether or not all the pixels in the reference block have been added (step S45).
If all pixels have not been added yet, the next pixel is set as a pixel to be added (step S46), and the pixel values are again accumulated in step S44. Such processing is repeated, and when it is detected in step S45 that the accumulation of all the pixels of the reference block has been completed, the value stored in the variable sum at that time is the sum of the pixel values of the reference block. ,
The information is stored in association with the reference block (step S47).

Then, it is detected whether or not the processing for obtaining the sum has been completed for all the blocks formed by dividing the reference frame (step S48), and the processing has been completed for all the blocks. If not, the next block is set as a reference block (step S49), and the processing of steps S42 to S47 is repeated again. If it is determined in step S48 that the process of obtaining the sum of all blocks has been completed, the process of obtaining the sum of each block of the reference frame ends (step S50), and the process returns to the main routine.

Hereinafter, the processing of the main routine of FIG. 2 will be described again. When the calculation of the block sum of all the image blocks of the current frame and the reference frame is completed, the first block is set as a current block from a plurality of blocks obtained by dividing the current frame (step S13). The processing for motion vector detection for the block is started. First, as an initial setting, a variable min for storing the minimum evaluation value
, A predetermined sufficiently large value such as the maximum value that the variable can take is set as an initial value (step S
14). Next, a plurality of preset candidate vectors (v
_xi , v _yi ) (i = 1 to m), the first candidate vector (v _x1 , v _y1 ) is set as a candidate vector for which an evaluation value is to be obtained (step S15).

Then, a reference block is determined based on the current block to be subjected to the motion detection and the candidate vector set in step S15, and the current block and the reference block already determined in steps S11 and S12 are determined. The value of the block sum is read, and the difference is obtained to obtain the variable di.
ff (step S16). Next, this difference d
It is checked whether or not iff is larger than the minimum value min of the evaluation value of the candidate vector already detected for the current block (step S17).
If the difference diff of the block sum is larger than the already detected minimum evaluation value min, even if an evaluation value is obtained for this candidate vector, the current minimum evaluation Since it is clear that the value becomes larger than the value min, the process for this candidate vector is stopped, and the process proceeds to step S25 and thereafter. Step S17
In the case where the difference diff of the block sum is equal to or smaller than the already detected minimum evaluation value min, it is determined that it is meaningful to obtain an evaluation value for this candidate vector. Move on.

In the process of obtaining the evaluation value, first, after resetting the contents of the variable sum for storing the accumulated value (step S18), the first pixel of the current block (the upper left pixel in the present embodiment) Is set as the pixel to be processed curr (x, y) (step S19). And the pixel curr (x, y) of the current block and its
curr (x, y) candidate vectors for (v _x, v _y) an amount corresponding pixel on the reference block of the moved position ref (x +
v _x , y + v _y ), the difference is added to the value already stored in the variable sum, and the variable sum is calculated again.
(Step S20). And the variable sum
And the value of the variable min are compared, and the current evaluation value of the candidate vector to be processed currently stored in the variable sum is equal to or less than the minimum evaluation value of each candidate vector processed so far stored in the variable min. Whether or not sum ≧ min is checked (step S2).
1).

If sum.gtoreq.min in step S21, the candidate vector being processed this time may be the motion vector to be obtained.
Then, the processing for obtaining the accumulation for the subsequent pixels is continued. That is, while checking whether the pixel to be processed is the last pixel (the lower right pixel in the present embodiment) of the current block (step S22), if it is not the last pixel, the next pixel is determined as the pixel to be processed. Is set (Step S23), and Step S20 and Step S20 are performed.
Step 21 is repeated. If sum <min in step S21, there is no possibility that the candidate vector currently being processed is a motion vector that has already been obtained.
The process proceeds to the process related to the next candidate vector after step S25.

The evaluation value is calculated while performing the check shown in step S21. If it is determined in step S22 that the accumulation has been completed up to the last pixel of the block, the determination is already made in step S21. As described above, the evaluation value for the candidate vector is equal to or smaller than the evaluation value obtained so far, and there is a possibility that the motion vector is the motion vector to be obtained. That is, the evaluation value sum is set to a variable min as a new minimum evaluation value, and the x component and the y component of the candidate vector are respectively set to a variable minvx
And the variable minvy.

When the processing based on the obtained evaluation value in step S24 is completed, and when the processing for the candidate vector is interrupted in steps S17 and S21, the step In S25, it is checked whether or not the candidate vector is the last candidate vector among the plurality of candidate vectors described above (step S25). If the last candidate vector is not the last candidate vector, the next vector is set as a candidate vector to be evaluated (step S25). S26), again from step S16
Step S24 is repeated. In step S25, if the candidate vector is the last candidate vector, the variables minvx and minvx are
The candidate vector stored in y is the motion vector detected for the current block, and the processing for the current block ends.

Then, it is checked whether or not the current block is the last block in the current frame (step S27). If the current block is not the last block, the next block is set as the current block for motion detection (step S28). , And repeat the processing of steps S14 to S26 again. If the current block is the last block in the current frame in step S27, a series of processes for the current frame is ended (step S29).

As described above, in the motion detection apparatus and method according to the present embodiment, before calculating the evaluation value for the candidate vector by calculating the corresponding pixels between the current block and the reference block, the current block and the reference block are first obtained. Based on the simple sum of pixel values in the reference block, it is checked whether the desired evaluation value is smaller than the already detected evaluation value. I try to find the value. Therefore, since the nested state is performed and some meaningless processing is eliminated before the calculation of pixels having a large processing amount is started, the processing amount can be significantly reduced as compared with the conventional method. The motion vector detection processing for one current block can be performed very quickly.

Further, in the present embodiment, for a given current frame and reference frame, the sum of pixel values in a block of an image block that can be extracted first is collectively obtained. Therefore, at the time of subsequent processing for each current block, it is only necessary to refer to the sum of pixel values in that block, and almost no processing time is required, and the motion detection processing can be performed efficiently. Also,
In the reference block, the same reference block may be referred to many times for a plurality of current blocks and a plurality of candidate vectors. In such a case, the reference block is included only once in the block. Since the sum of the pixel values only needs to be obtained, the motion detection processing for each frame can be performed at higher speed.

Note that the present invention is not limited to the present embodiment, and any suitable modifications can be made. For example, in the present embodiment, the current frame and the reference frame are described as being independent from each other, and the block sum is calculated each time.
For example, if an input image signal is an MPEG signal (Moving Pi
This is a signal of a high-quality moving picture coding system by the coding coding experts group, and when there is forward prediction or backward prediction, a current frame in a certain process may be referred to as a reference frame of another current frame. In such a case, by holding the data of the block sum calculated once, the number of times of obtaining the block sum can be reduced, and the processing can be performed at higher speed. Obviously, such a configuration is within the scope of the present invention.

Also, in this embodiment, every time one calculation between corresponding pixels of the current block and the reference block is completed, the accumulated value is compared with the minimum evaluation value when obtaining the evaluation value. I was trying to do it. However, the comparison may be performed at arbitrary intervals, for example, the comparison is performed every ten pixels or the comparison is performed every one line. By performing the comparison at an appropriate timing, a higher speed can be achieved. Further, in the present embodiment, in the reference frame, the block sum is obtained for all the image blocks shifted by one pixel or one line. However, when the search range of the motion vector is narrow and limited, or when prediction is possible, when there is no possibility that all image blocks are referred to as reference blocks, the block sum of only the necessary image blocks can be obtained. Good.

[0043]

According to the motion detecting apparatus and method of the present invention, motion can be detected at high speed.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a motion detection device according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a motion detection process performed by the motion detection device illustrated in FIG. 1;

FIG. 3 is a flowchart for explaining in more detail a process of obtaining a sum of each block of a current frame in the flowchart shown in FIG. 2;

FIG. 4 is a flowchart for explaining in more detail a process of calculating the sum of each block of a reference frame in the flowchart shown in FIG. 2;

FIG. 5 is a diagram for explaining a block matching algorithm.

FIG. 6 is a flowchart for explaining a conventional motion detection process.

FIG. 7 is a flowchart illustrating a motion detection process in which the process illustrated in FIG. 6 is improved.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Motion detection device, 2 ... CPU, 3 ... ROM, 4 ... RA
M

Claims (8)

[Claims] 1. A motion detecting apparatus for detecting a motion vector by referring to a reference image for a first image block obtained by dividing a motion detection target image, comprising a plurality of predetermined candidate vectors. A candidate vector selecting unit that sequentially selects one; a reference block extracting unit that extracts a second image block on the reference image based on the first image block and the selected candidate vector; A first step of detecting a sum of pixel values of the first image block;
And a second sum detecting means for detecting a sum of pixel values of the second image block.
Sum detection means for detecting a difference between the sum of the detected pixel values of the first image block and the sum of the pixel values of the detected second image block; First comparing means for comparing the difference between the sums thus set and the set minimum evaluation value; and as a result of the comparison, when the detected difference between the sums is at least smaller than the minimum evaluation value, Evaluation value detection means for detecting the sum of differences between pixel values of corresponding pixels of the first image block and the second image block as an evaluation value for the selected candidate vector; and the detected evaluation value And second comparing means for comparing an evaluation value already detected with respect to the candidate vector already selected by the candidate vector selecting means; and, as a result of the comparison, the detected evaluation value Storage means for storing the selected candidate vector when the evaluation value is the smallest among the evaluation values, and setting the evaluation value as the minimum evaluation value; and Output means for outputting the candidate vector stored in the storage means as a motion vector for the first image block after all the candidate vectors are selected. 2. The method according to claim 2, wherein said plurality of candidate vectors predetermined by said candidate vector selecting means are applied to each of a plurality of image blocks obtained by dividing said motion detection target image. When the image block is determined, the image processing apparatus further includes a reference block total calculation unit that previously obtains a total of pixel values for all image blocks on the reference image that can be the second image block. Means for the second image block extracted by the reference block extraction means,
The motion detection according to claim 1, wherein the reference block sum calculating means detects the sum of the pixel values of the second image block by referring to the sum of the pixel values of the second image block obtained in advance. apparatus. 3. The reference block sum calculating means has the same pixel configuration as that of the first image block, and obtains the sum of the pixel values in advance for all image blocks obtained from the reference image. 3. The motion detection device according to 2. 4. A motion detection target block sum calculating means for previously calculating a sum of pixel values for each of a plurality of image blocks obtained by dividing the motion detection target image; Means for the first image block, by referring to the sum of the pixel values of the first image block obtained in advance by the motion detection target block sum calculation means, 4. The motion detection device according to claim 3, wherein the sum of the values is detected. 5. The evaluation value detecting means obtains a value ME substantially as shown in Expression 1 as the evaluation value.
The motion detection device according to claim 1. (Equation 1) 6. A first image block obtained by dividing a motion detection target image into a region having a predetermined pixel configuration, and the first image block based on a given candidate vector which is an image block in a reference image. Sum of differences between corresponding pixels of the first image block and the second image block having the same pixel configuration determined for the position of the image block of
A motion detection method, wherein the sum is used as an evaluation value for the candidate vector, and a candidate vector having the smallest evaluation value from a plurality of predetermined candidate vectors is detected as a motion vector for the first image block. The sum of the pixel values of the second image block, the sum of the pixel values of the second image block, and the sum of the pixel values of the second image block It is determined whether or not the obtained difference is smaller than the evaluation value calculated for the candidate vector given so far. A motion detection method for calculating the evaluation value for a candidate vector. 7. When the second image block is determined based on the plurality of predetermined candidate vectors for each of a plurality of image blocks obtained by dividing the motion detection target image, For all image blocks in the reference image that can be the second image block, the sum of pixel values is obtained in advance, and the second image block and the second When the image block of the first image block is determined, a difference from the total of the pixel values of the first image block is determined by referring to the previously calculated total of the pixel values of the second image block. Item 7. The motion detection method according to Item 6. 8. A sum of pixel values is obtained in advance for each of a plurality of image blocks obtained by dividing the motion detection target image, and for the first image block, a given candidate vector is used. When the second image block is determined, the difference from the sum of the pixel values of the second image block is determined by referring to the previously calculated sum of the pixel values of the first image block. 8. The motion detection method according to claim 7, wherein the motion is obtained.