JPH11113003A - Motion vector detector and motion vector detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To execute detection of a motion vector efficiently with high accuracy from images with various characteristics. SOLUTION: A motion vector detection section 30 detects a motion vector from a coding object block, based on a correlation with respect to object blocks within a search range. A search range storage memory 20 stores pixel data within an original search range and gives the pixel data within the original search range to the motion vector detection section 30 in response to a memory address MA. A control section 40 sets a size and a position of the search range, based on a stored motion vector fed from the motion vector detection section 30 and a correlation evaluation value and generates a pixel width in horizontal and vertical directions of the search range and an offset denoting a relative position with respect to the original search range. An address generating section 10 generates the memory address MA given to the retrieval range storage memory 20, based on the pixel width and the offset of the search range generated by the control section 40.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to motion vector detection used in motion compensated predictive coding, which is a technique for compressing and coding digital video data.

[0002]

2. Description of the Related Art As a method of realizing image coding of a moving image, temporal information is used by using information (movement vector) indicating from where in the previous screen a certain portion of the current screen has moved. There is a method of reducing redundancy. One method for extracting the motion vector is a block matching method.

FIG. 16 shows the principle of the block matching method. A target frame image (target frame) to be encoded is compared with a search frame image (search frame) for searching for a motion vector, and the target frame image most similar to the coding target block of the target frame (that is, having the highest degree of correlation). A block (best match block) is extracted from within the search range of the search frame to detect a motion vector.
This block matching method is widely used for motion compensation prediction in image compression coding.

In the block matching method, the degree of correlation between one encoding target block and a plurality of candidate blocks within a search range of a search frame is calculated, and the candidate block having the highest correlation is determined as the best match block. The difference between the position of the best match block and the current block is detected as a motion vector.

As the evaluation value of the degree of correlation, an accumulated value of square error (Square Error) and an accumulated value of absolute value error (Absolute Error) are generally used. Since the difference value between the current block and the candidate block becomes smaller as the degree of correlation becomes higher, when the accumulated value of the square error or the accumulated value of the absolute value error is used as the evaluation value, the evaluation value is , The higher the correlation, the smaller the value.

The purpose of motion vector detection is to detect, from a search frame, a block in which the amount of code can be reduced most effectively in motion compensated prediction coding. Therefore, the detection accuracy of the motion vector is one of the important performances of the motion vector detection device. However, motion vector detection is a process that requires a particularly large amount of computation among image coding algorithms, and when this is realized by hardware, power consumption is large due to the long processing time and large hardware scale. There are issues. At present, real-time processing by software is extremely difficult. From these backgrounds, motion vector detection with high detection accuracy and high processing efficiency is desired.

As a conventional motion vector detecting device, a search range is reduced for an image having a small motion to shorten the processing time for detecting a motion vector, while a search range is increased for an image having a large motion. By,
In some cases, the processing efficiency is improved without lowering the detection accuracy.

For example, a motion vector detecting device disclosed in Japanese Patent Application Laid-Open No. 8-32969 has a means for performing a correlation evaluation at the same position as a processing block, and a search range size according to the correlation evaluation result at the same position. A means for adaptively controlling the selection and a means for evaluating the correlation between each candidate block in the search range and detecting a motion vector are provided, and the search range is reduced as the correlation at the same position becomes higher. This allows
A motion vector can be accurately detected with a small amount of calculation.

[0009]

However, the conventional motion vector detecting device has the following problems.

[0010] A so-called panned image in which the entire screen moves in a fixed direction is frequently used in an image captured by a camera or the like. In the pan image, the object itself projected on the screen does not change, and its position moves in a certain direction.

In the above-mentioned conventional motion vector detecting device,
Since the size of the search range is set by the correlation evaluation at the same position as the processing block, for example, for a pan image, the degree of correlation at the same position is low, so that a large search range is always set. In addition, since the same position correlation evaluation is performed prior to the motion vector detection, there is a problem that the processing becomes complicated.

In view of the above problems, an object of the present invention is to make it possible to efficiently and accurately execute motion vector detection on an image having various characteristics. Thus, the motion vector detection can be performed accurately and at high speed.

[0013]

Means for Solving the Problems To solve the above-mentioned problems, a solution taken by the invention of claim 1 is to provide a motion vector detecting device as a coding target block of a target frame and a search range of a search frame. Motion vector detecting means for calculating an evaluation value indicating a degree of correlation with a candidate block in the block, and detecting a motion vector for the encoding target block based on the degree of correlation evaluation, and pixel data in an original search range. And an original search range storage means for reading out pixel data in accordance with a given memory address and supplying the pixel data to the motion vector detection means, and a first search width which is a horizontal pixel width of the search range. Control means for generating a second search width, which is a pixel width in the vertical direction, and an offset value indicating a relative position of the search range with respect to the original search range; Based on the first and second search widths and the offset value generated I, wherein is obtained and an address generating means for generating a memory address to be given to the original search range storing means.

According to the first aspect of the present invention, the address generating means generates a memory address corresponding to the search range based on the first and second search widths and the offset value generated by the control means. Using the pixel data read from the original search range storage means according to the memory address, the motion vector detection means calculates an evaluation value indicating the degree of correlation between the current block and the candidate blocks within the search range. , Detect the motion vector. For this reason, the size and position of the search range can be freely set,
Motion vector detection can be performed efficiently and accurately.

According to a second aspect of the present invention, in the motion vector detecting device according to the first aspect, the motion vector detecting means controls the detected motion vector and a correlation evaluation value corresponding to the detected motion vector. The control means generates first and second search widths and offset values based on the motion vector and the correlation evaluation value supplied from the motion vector detection means. .

According to the second aspect of the present invention, the first of the search range is set.
And the second search width and the offset value are generated based on the motion vector and the evaluation value already detected, so that new processing is performed prior to the motion vector detection in order to determine the size and position of the search range. No need. In addition, since the motion vector represents the motion of the image, by setting the search range based on the motion vector and the corresponding correlation evaluation value, particularly for a pan image in which the image moves in a certain direction. Thus, an appropriate search range can be set.

Further, in the invention according to claim 3, the control means in the motion vector detecting apparatus according to claim 2 is arranged such that a block located at the same position as or adjacent to the coding target block in the previous frame of the target frame is set as a range setting reference block. The size of the search range is set based on the correlation evaluation value corresponding to the range setting reference block.

According to the invention of claim 3, the correlation evaluation value corresponding to the range setting reference block located at or near the same position as the current block in the previous frame of the current frame, Since the degree of association with the search frame can be predicted to some extent, setting the size of the search range based on the correlation evaluation value of this range setting reference block makes it possible to set an appropriate value especially for a panned image in which the image moves in a certain direction. The search range can be set.

According to the fourth aspect of the present invention, the second aspect of the present invention is provided.
The control means in the motion vector detection device of the above, the block at the same position as the encoding target block in the previous frame of the target frame or a block in the vicinity thereof as a range setting reference block, based on a motion vector corresponding to the range setting reference block, The position of the search range shall be set.

According to the fourth aspect of the present invention, the motion of the current block to be encoded is controlled to some extent by the motion vector corresponding to the range setting reference block at the same position as or near the current block in the previous frame of the current frame. Since the prediction can be made, by setting the position of the search range by the motion vector of the range setting reference block, it is possible to set an appropriate search range, particularly for a panned image in which the image moves in a certain direction.

According to a fifth aspect of the present invention, the control means in the motion vector detecting apparatus according to the second aspect sets the block at the same position as or adjacent to the encoding target block in the previous frame of the target frame as a range setting reference block. And, when the correlation degree indicated by the correlation degree evaluation value corresponding to the range setting reference block is relatively low, the original search range is set as the search range, while when the correlation degree is relatively high,
The size of the search range is set smaller than the original search range, and the position of the search range is set based on the motion vector corresponding to the range setting reference block.

In the invention according to claim 6, the first aspect is provided.
The offset value in the motion vector detecting device is a component of a vector starting from the search starting point in the original search range and ending at the search starting point in the search range.

According to a seventh aspect of the present invention, in the motion vector detecting device according to the first aspect, the address generating means includes a first counter for counting clocks, a first search width, and a first counter. Comparing the count value, and when they match, a comparator that resets the count value of the first counter and outputs a count-up signal; a second counter that counts the count-up signal; A memory address conversion unit configured to generate a memory address to be supplied to the original search range storage unit in accordance with a predetermined formula based on the count values of the first and second counters.

[0024] In the invention of claim 8, the above-mentioned claim 1 is provided.
In the motion vector detecting device, the control means may include a first and a second search width and an offset value, and a first sub-sample degree in a horizontal direction of the search range.
And the second sub-sample number representing the vertical sub-sample degree is generated by the control unit. The address generation unit generates the memory address to be given to the original search range storage unit by the control unit. The search is performed based on the first and second search widths, the first and second subsample numbers, and the offset value.

According to a ninth aspect of the present invention, a method of detecting a motion vector for a block to be encoded includes:
A first step of determining whether or not there is a similar part to the encoding target block; and determining that there is a similar part in the first step, the search frame includes the similar part, A second step of setting a first area smaller than the predetermined area as a search range for detecting a motion vector;
And setting a region larger than the first region as a search range for detecting a motion vector when it is determined that there is no similar portion in the step.

According to the ninth aspect of the present invention, it is determined whether or not there is a portion similar to the block to be coded in a predetermined area of the search frame. Since a region including a portion and smaller than the predetermined region is set as a search range for motion vector detection, an appropriate search range can be set particularly for a panned image in which an image moves in a certain direction. become. This makes it possible to efficiently and accurately execute motion vector detection.

According to a tenth aspect of the present invention, in the motion vector detecting method according to the ninth aspect, the first step includes the step of detecting a block at the same position as or adjacent to the current block to be coded in a frame preceding the current frame. A range setting reference block, and the determination is performed based on a correlation degree evaluation value corresponding to a motion vector detected for the range setting reference block, and the second step includes: It is set based on the motion vector detected for the range setting reference block.

In the eleventh aspect of the present invention, in the motion vector detecting method according to the ninth aspect, the first step includes the step of: determining a unit area including at least a part of an encoding target block; A predetermined pixel operation is performed on each of a plurality of comparison regions of the same size as the unit region provided near the same position as the target block, and the determination is performed based on the calculation results. In the second step, the most similar to the unit area is obtained from the plurality of comparison areas from the calculation result of the predetermined pixel calculation, and the position of the search range is set based on the position of the comparison area. It shall be.

In a twelfth aspect of the present invention, a motion vector detecting method comprises the steps of: setting a search range for detecting a motion vector; Detecting a motion vector based on the degree of correlation with the block, wherein the search range setting step includes setting a block at the same position as or adjacent to the coding target block in the previous frame of the target frame as a range setting reference. A block is set, and the size of the search range is set based on the correlation evaluation value corresponding to the motion vector detected for the range setting reference block.

[0030] According to the twelfth aspect of the present invention, the correlation evaluation value corresponding to the range setting reference block located at or near the same position as the current block in the previous frame of the current frame,
Since the degree of association with the search frame can be predicted to some extent, setting the size of the search range based on the correlation evaluation value of the range setting reference block makes it possible to set an appropriate search range.

According to a thirteenth aspect of the present invention, there is provided a motion vector detecting method comprising the steps of: setting a search range for detecting a motion vector; Detecting a motion vector based on the degree of correlation with the block, wherein the search range setting step includes setting a block at the same position as or adjacent to the coding target block in the previous frame of the target frame as a range setting reference. A block is set, and the position of the search range is set based on the motion vector detected for this range setting reference block.

According to the thirteenth aspect of the present invention, the motion of the current block to be encoded is controlled to some extent by the motion vector corresponding to the range setting reference block at the same position as or near the current block in the previous frame of the current frame. Since the prediction can be made, it is possible to set an appropriate search range by setting the position of the search range by the motion vector of the range setting reference block.

[0033]

Embodiments of the present invention will be described below with reference to the drawings.

(First Embodiment) FIG. 1 is a diagram showing three temporally continuous frame images, and shows a so-called pan image obtained by moving a camera from right to left. Generally, in a pan image, when viewed in a very short time, it can be said that the object on the screen is moving at a constant speed. Therefore, in the present embodiment, when detecting the motion vector MV for the encoding target block CB of the target frame (frame (1)), the block at the same position as the encoding target block CB in the search frame (frame (0)) AB is a range setting reference block, and the motion vector BM detected for this range setting reference block AB
Based on V and the correlation evaluation value BAE at that time, the size and position of the search range are set. Thereby, as shown in FIG. 1, it is possible to set a search range for detecting a motion vector by narrowing down to a position where it is estimated that an image of the coding target block CB exists in the search frame. Therefore, motion vector detection can be performed accurately and at high speed.

FIG. 2 is a schematic diagram showing a search range for detecting a motion vector according to the present embodiment. In the present embodiment, there are two types of search range sizes. Figure 2 shows MPEG
FIG. 3 shows an example of a search range for an encoding target block CB having a general size (16 pixels in the horizontal direction and 16 pixels in the vertical direction) according to the image encoding standard such as
(A) shows a search range of 46 pixels in the horizontal direction and 46 pixels in the vertical direction as an example of the large search range W, and (b) shows 30 pixels in the horizontal direction and 30 pixels of the vertical direction as an example of the small search range S
3 shows a pixel search range.

The position of the small search range S is represented by a relative position with respect to the large search range W. Here, an offset (Hoff, Voff) is defined as a vector starting from the upper left corner which is the search start point of the search range W and ending at the upper left corner which is the search start point of the search range S. , The position of the search range S. At this time, for example, when the image is a still image and the search range S is set at the position of the coding target block CB, as shown in FIG. 2B, the offset value is Hoff = 8 Voff = 8 become. Also, the horizontal pixel width of the search range is defined as a search horizontal (H) pixel width as a first search width, and the vertical pixel width of the search range is defined as a search vertical (V) as a second search width.
Pixel width.

FIG. 3 is a diagram showing a configuration of the motion vector detecting device according to the present embodiment. In FIG. 3, reference numeral 10 denotes an address generation unit as an address generation unit for generating a memory address MA based on the search H pixel width, the search V pixel width, and the offset, and 20 denotes an original search range for storing image data of the original search range. Search range storage memory as storage means,
30 calculates an evaluation value indicating the degree of correlation between the current block and the candidate block in the search range, and outputs a motion vector corresponding to the maximum degree of correlation and an evaluation value corresponding to the motion vector. The motion vector detecting unit 40 determines a search H pixel width, a search V pixel width, and an offset from the motion vector and the evaluation value output by the motion vector detection unit 30, and outputs these to the address generation unit 10. Is an encoding target block storage memory for storing image data of the encoding target block. Here, a large search range W of 46 × 46 pixels is set as the original search range.

FIG. 4 is a flowchart showing the operation of the control unit 40. The operation of the control unit 40 will be described with reference to the flowchart of FIG. 4 and FIG.

Now, it is assumed that a search range for detecting a motion vector is set for a coding target block CB in a target frame. The control unit 40 sets the motion vector BMV for the search frame range setting reference block AB.
Based on the correlation evaluation value BAE at that time, a search H pixel width, a search V pixel width, and an offset representing the search range are supplied to the address generation unit 10.

First, in step S1, the correlation evaluation value BAE of the range setting reference block AB is compared with a threshold value AE0. As a result of this comparison, when the evaluation value BAE is larger than the threshold value AE0, since the degree of correlation corresponding to the motion vector BMV is low, a portion similar to the coding target block CB in a predetermined area in the search frame If it is determined that there is no search range, the original search range, that is, a large search range W of 46 × 46 pixels is set as the search range for motion vector detection without narrowing the search range (step S).
2). That is, in this case, the search H pixel width and the search V pixel width are both “46”, and the offset is (0,
0).

When the correlation degree evaluation value BAE of the range setting reference block AB is not larger than the threshold value AE0, the correlation degree corresponding to the motion vector BMV is high, so that the coding target is included in a predetermined area in the search frame. Block C
It is determined that there is a portion similar to B, and a small search range S of 30 × 30 pixels is set as a search range for motion vector detection (step S3). And step S
4, the magnitude | BMV (x, y) | of the motion vector BMV of the range setting reference block AB and the threshold value MV0
Compare with

As a result of the comparison, when the magnitude of the motion vector BMV is larger than the threshold value MV0, it is determined that the motion of the encoding target block CB is large, and an offset is set using the component of the motion vector BMV. (Step S5). Specifically, the offset (Hoff, Vo
The value of each component of ff) is set as follows. Hoff = x + 8 Voff = y + 8 That is, in this case, both the search H pixel width and the search V pixel width are set to “30”, and the offset is (x +
8, y + 8).

On the other hand, when the magnitude of the motion vector BMV is not larger than the threshold value MV0, it is determined that the motion of the encoding target block CB is minute and a still image, and the offset is set to (8, 8). I do. That is, in this case, both the search H pixel width and the search V pixel width are “30”.
And the offset is (8, 8).

The address generator 10 generates a memory address MA based on the search H pixel width, the search V pixel width, and the offset output from the controller 40. The motion vector detection unit 30 performs a motion vector detection on the encoding target block CB using the pixel data in the search range read from the search range storage memory 20 (step S10).

FIG. 5 is a block diagram showing the configuration of the address generator 10. As shown in FIG.
0 compares the H counter 11 as a first counter for counting clocks, the search H pixel width with the count value of the H counter 11, and when they match, resets the count value of the H counter 11 and counts. H comparator 12 as a comparator that outputs up signal CU, and V counter 1 as a second counter that counts up signal CU
3, a V-comparator 14 that compares the search V pixel width with the count value of the V counter 13 and generates an operation signal for terminating the operation of the motion vector detector 30 when they match, an offset value and the H counter 11 And a memory address conversion unit 15 that generates a memory address MA of the search range storage memory 20 based on the count value of the V counter 13 and the count value of the V counter 13. The H counter 11 and the V counter 13 are reset by the start signal supplied from the control unit 40, and the address generation unit 10 starts operating.

The operation of the address generator 10 shown in FIG. 5 will be described with reference to FIGS.

FIG. 6 is a diagram showing the relationship between the original search range stored in the search range storage memory 20 and the search range for detecting a motion vector. Here, as shown in FIG. 6, the search range storage memory 20 stores pixel data of horizontal 12 pixels × vertical 8 pixels as an original search range, and uses a search H pixel width as a search range for detecting a motion vector. It is assumed that the search range is set to 8, the search V pixel width is 6, and the offset is (2, 1).

FIG. 7 is a diagram showing a result of mapping the two-dimensional pixel data shown in FIG. 6 into a one-dimensional address space of the search range storage memory 20. The address generation unit 10
A memory address MA for reading pixel data in a search range for detecting a motion vector from the search range storage memory 20 is generated.

FIG. 8 is a timing chart showing the operation of the address generator 10. As shown in FIG. 8, when a start signal is input, the H counter 11 starts counting clocks. When the count value of the H counter 11 reaches a value corresponding to the search H pixel width, the H comparator 12 outputs a count-up signal CU, whereby the count value of the H counter 11 is reset. The V counter 13 outputs a counter up signal C
Count U. When the count value of the V counter 13 reaches a value corresponding to the search V pixel width, the V comparator 14 outputs an operation signal.

When the offset (Hoff, Voff) is input, the memory address converter 15 calculates the horizontal pixel width Hmax and the vertical pixel width Vmax of the original search range stored in the search range storage memory 20 from Address SA serving as starting point of search range for motion vector detection
Find 0. This starting address SA0 is obtained by the following equation. SA0 = Hmax × Voff + Hoff In the example of FIG. 6, SA0 = 12 × 1 + 2 = 14.

Then, the memory address conversion unit 15 obtains the memory address MA based on the following equation. MA = SA0 + Hmax × (count value of V counter 13)
As a result, the memory address MA changes as shown in FIG. For example, as shown in FIG.
, The memory address MA is obtained by adding the count value of the H counter 11 to the starting address SA0. By the next clock input, H
The count value of the counter 11 becomes 0, and the count value of the V counter 13 becomes 1. At this time, the memory address MA becomes: MA = 14 + 12 × 1 + 0 = 26.

As described above, according to the present embodiment, it is possible to accurately and efficiently detect a motion vector even for a panned image.

In the present embodiment, the block located at the same position as the block to be coded in the search frame is used as the range setting reference block. The block may be used as a range setting reference block. Further, a block located at the same position as or near the block to be coded in a frame further before the search frame may be used as the range setting reference block.

In this embodiment, the search range has two sizes, but more sizes may be set. It is also possible to make the search H pixel width and the search V pixel width variable using the magnitude of the motion vector component of the range setting reference block as an evaluation index.

(Modification of First Embodiment) In the first embodiment, it is also possible to apply a subsample together. In this case, in addition to the search H pixel width, the search V pixel width, and the offset value, the control unit 40 determines the number of horizontal subsamples DH and the number of second subsamples as the first number of subsamples in the search range. Is input to the address generation unit 10.

FIG. 9 is a flowchart showing the operation of the control unit 40 according to this modification. The difference from FIG. 4 is that, in addition to the setting of the size and position of the search range, the presence or absence of a subsample is specified. That is, in step S2A for setting the large search range W and in step S5A for setting the small search range S to a position other than the position of the block CB to be coded, the presence of sub-samples is specified, and In step S6A for setting to the position of CB, no sub-sample is specified.

FIG. 10 is a schematic diagram showing a search range for detecting a motion vector according to this modification. Here, the number DH of sub-samples in the horizontal direction and the number DV of sub-samples in the vertical direction are both 2. In the figure, (a) shows an example of a large search range W in which subsamples are specified, and (b)
Shows an example of a small search range S in which subsamples are specified, and (c) shows an example of a small search range S in which no subsamples are specified. The reason why the small search range S is set at the position of the encoding target block CB in step S6A is that it is determined that the motion of the encoding target block CB is minute. In this case, since it is not preferable to lower the search accuracy, as shown in FIG. 10C, no sub-sample is designated.

The operation of the address generator 10 according to the present modification will be described with reference to FIGS.

FIG. 11 is a diagram showing the relationship between the original search range stored in the search range storage memory 20 and the search range for motion vector detection. Here, as shown in FIG. 11, search H is used as a search range for motion vector detection.
Pixel width is 6, search V pixel width is 4, offset is (0,
0), a search range of 2 is set for both subsamples DH and DV.

FIG. 12 is a diagram showing a result of mapping the two-dimensional pixel data shown in FIG. 11 into a one-dimensional address space of the search range storage memory 20.

FIG. 13 is a timing chart showing the operation of the address generator 10. In the present modification, the memory address conversion unit 15 calculates the memory address MA based on the following equation.
Ask for. MA = SA0 + Hmax × (count value of V counter 13)
× DV + (count value of H counter 11) × DH As a result, the memory address MA changes as shown in FIG.

The present invention can be easily applied to detection of a motion vector corresponding to rotation and enlargement / reduction of an image by setting the subsamples DH and DV.

(Second Embodiment) In a second embodiment of the present invention, a search frame is provided with a plurality of blocks having the same size as the current block to be coded near the same position as the current block to be coded. Performs predetermined pixel operations on the current block and a plurality of blocks set in the search frame, and sets the size and position of the search range for motion vector detection based on the calculation results. Is what you do. Here, it is assumed that as a predetermined pixel operation, an accumulation operation of pixel data in a block is performed.

FIG. 14 is a diagram showing an example of block setting in a search frame according to the present embodiment. In FIG. 14, for a coding target block CB of 16 × 16 pixels as a unit area, 1
Nine blocks SB0 to SB8 of 6 × 16 pixels are set.

The motion vector detecting device according to the present embodiment is configured as shown in FIG. 3, similarly to the first embodiment. However, the control unit 40 does not receive the motion vector and the evaluation value output from the motion vector detection unit 30 as inputs.
Instead, the image data of each frame is input.

FIG. 15 is a flowchart showing the operation of the control unit 40 in the motion vector detecting device according to the present embodiment. First, in step S21, the coding target block CB and each block S set in the search frame are set.
For B0 to SB8, an accumulation operation of the pixel data in the block is performed. In step S22, the accumulation operation result AC (CB) of the encoding target block and the accumulation operation result AC (SBn) of each block of the search frame (n = 0 to 8)
Minimum value SAmin of the absolute value of the difference between
Compare with As a result of this comparison, when the minimum value SAmin is larger than the threshold value SA0, it is determined that there is no portion similar to the coding target block CB in a predetermined area of the search frame, and the search range is not narrowed. 46 × 4
A large search range W of 6 pixels is set as a search range for detecting a motion vector (step S23). That is, in this case, the search H pixel width and the search V pixel width are both set to "46", and the offset is set to (0, 0).

If the minimum value SAmin is not larger than the threshold value SA0, it is determined that there is a portion similar to the coding target block CB in a predetermined area of the search frame, and a small search range of 30 × 30 pixels is determined. S is set as a search range for motion vector detection (step S2
4). Then, in step S25, the block SB
When the absolute value of the difference between the accumulation operation result of k and the accumulation operation result of the encoding target block CB is minimized, the block SB
An offset is set so that the search range S is set at the position of k.

For example, a block SB0 is set at the same position as the block CB to be coded, and the block SB0 is set therearound.
When 1 to SB8 are set, (8, 8) is set as the offset when the absolute value of the difference between the accumulation operation result of the block SB0 and the accumulation operation result of the encoding target block CB is minimized. I do. Also, when the absolute value of the difference between the accumulation operation result of the block SB8 located at the lower right of the block SB0 and the accumulation operation result of the encoding target block CB is minimized, (16, 16) is set as the offset. I do.

As described above, according to the present embodiment, since the offset is set without using the motion vector detected in the previous frame, the offset can be set from the initial frame in which the motion vector detection is performed. Motion vector detection can be performed more efficiently than in the form.

Note that a part of the block to be encoded may be set as a unit area, a plurality of comparison areas having the same size as the unit area may be set, and predetermined pixel operations may be performed on these. In addition, the unit area may include an area near the encoding target block. Further, the unit area may have any shape.

As the predetermined pixel operation, in addition to the accumulation operation, for example, an accumulation operation of a difference between adjacent pixels, a dispersion operation of pixels, a frequency distribution operation of pixels, or the like may be used. Further, the calculation result may be weighted according to the position of the region.

[0072]

As described above, according to the present invention, the size and position of the search range can be freely set, so that the motion vector can be detected efficiently and accurately. In addition, since the search range is set based on the already detected motion vector and the correlation degree evaluation value corresponding to the motion vector, the search range is set particularly for a pan image in which the image moves in a certain direction. Becomes possible.

In the search frame, an area including a part similar to the block to be coded is set as a search range for detecting a motion vector. The search range can be set.

Furthermore, the size and position of the search range are set by the motion vector and the correlation evaluation value corresponding to the range setting reference block at or near the same position as the current block to be coded in the previous frame of the current frame. , An appropriate search range can be set.

[Brief description of the drawings]

FIG. 1 is a diagram for describing a first embodiment of the present invention, and is a diagram illustrating three temporally continuous frame images representing a pan image.

FIGS. 2A and 2B are schematic diagrams illustrating a search range in motion vector detection according to the first embodiment of the present invention. FIG. 2A illustrates an example of a large search range, and FIG. 2B illustrates an example of a small search range. is there.

FIG. 3 is a diagram illustrating a configuration of a motion vector detection device according to an embodiment of the present invention.

FIG. 4 is a flowchart illustrating an operation of a control unit according to the first embodiment.

FIG. 5 is a block diagram illustrating a configuration of an address generation unit.

FIG. 6 is a diagram illustrating a relationship between a maximum search range stored in a search range storage memory and a search range for detecting a motion vector.

7 is a diagram showing a result of mapping the two-dimensional pixel data shown in FIG. 6 into a one-dimensional address space of a search range storage memory.

FIG. 8 is a timing chart showing the operation of the address generation unit according to the first embodiment.

FIG. 9 is a flowchart illustrating an operation of a control unit according to a modification of the first embodiment.

FIG. 10 is a schematic diagram showing a search range in motion vector detection according to a modification of the first embodiment.

FIG. 11 is a diagram showing a relationship between an original search range stored in a search range storage memory and a search range for detecting a motion vector.

12 is a diagram showing a result of mapping the two-dimensional pixel data shown in FIG. 11 into a one-dimensional address space of a search range storage memory.

FIG. 13 is a timing chart illustrating an operation of an address generation unit according to a modification of the first embodiment.

FIG. 14 is a diagram illustrating an example of setting a comparison area in a search frame according to the second embodiment of the present invention.

FIG. 15 is a flowchart illustrating an operation of a control unit according to the second embodiment of the present invention.

FIG. 16 is a diagram showing block matching.

[Explanation of symbols]

CB Encoding target block AB Range setting reference block BMV Motion vector of range setting reference block BAE Correlation evaluation value Hoff, Voff offset value of range setting reference block MA Memory address CU Count-up signal DH Number of subsamples in horizontal direction (first number) DV The number of subsamples in the vertical direction (the number of second subsamples) SB0 to SB8 Blocks (comparison area) 10 Address generation unit (address generation means) 11 H counter (first counter) 12 H comparator (Comparator) 13 V counter (second counter) 15 Memory address conversion unit 20 Search range storage memory (original search range storage unit) 30 Motion vector detection unit (motion vector detection unit) 40 Control unit (control unit)

Claims (13)

[Claims] 1. An encoding target block of a target frame,
Motion vector detecting means for calculating an evaluation value indicating a degree of correlation with a candidate block within a search range of a search frame, and detecting a motion vector for the encoding target block based on the degree of correlation evaluation value; An original search range storage unit that stores pixel data in a search range, reads pixel data in accordance with a given memory address, and supplies the pixel data to the motion vector detection unit; A control unit that generates a first search width and a second search width that is a pixel width in the vertical direction, and an offset value that indicates a relative position of the search range with respect to an original search range, and a second search width that is generated by the control unit. Address generation means for generating a memory address to be given to the original search range storage means based on the first and second search widths and the offset value. Vector detection device. 2. The motion vector detecting device according to claim 1, wherein the motion vector detecting means supplies the detected motion vector and a correlation evaluation value corresponding to the motion vector to the control means. Wherein the control means generates first and second search widths and offset values based on the motion vector and the correlation evaluation value supplied from the motion vector detection means. Vector detection device. 3. The motion vector detection device according to claim 2, wherein the control unit sets a block at the same position as or adjacent to the coding target block in a previous frame of the target frame as a range setting reference block, and sets the range setting reference block. A motion vector detecting device for setting a size of a search range based on a correlation evaluation value corresponding to a reference block. 4. The motion vector detecting device according to claim 2, wherein the control unit sets a block at the same position as or adjacent to the coding target block in a frame preceding the target frame as a range setting reference block, and sets the range setting reference block. A motion vector detecting device for setting a position of a search range based on a motion vector corresponding to a reference block. 5. The motion vector detecting device according to claim 2, wherein the control unit sets a block located at the same position as or near the coding target block in a previous frame of the target frame as a range setting reference block, and sets the range setting reference block. When the correlation degree indicated by the correlation evaluation value corresponding to the reference block is relatively low, the original search range is set as the search range, while when relatively high, the size of the search range is set to be larger than the original search range. A motion vector detecting device which is set small and sets the position of the search range based on a motion vector corresponding to the range setting reference block. 6. The motion vector detecting device according to claim 1, wherein the offset value is a component of a vector having a search start point of the original search range as a start point and a search start point of the search range as an end point. 7. The motion vector detecting device according to claim 1, wherein the address generating means compares a first counter for counting clocks, the first search width, and a count value of the first counter. And when they match, resets the count value of the first counter and outputs a count-up signal; a second counter that counts the count-up signal; the offset value; And a memory address conversion unit for generating a memory address to be supplied to the original search range storage unit in accordance with a predetermined calculation formula based on the count value of the second counter and the count value of the second counter. apparatus. 8. The motion vector detecting device according to claim 1, wherein the control means includes: a first search width, a second search width, and an offset value.
A first sub-sample number representing a horizontal sub-sample degree and a second sub-sample number representing a vertical sub-sample degree in the search range; and the address generating means comprises: The memory address given to the range storage means is generated based on the first and second search widths, the first and second subsample numbers, and the offset value generated by the control means. Motion vector detection device. 9. A method for detecting a motion vector for an encoding target block, the method comprising: determining whether a portion similar to the encoding target block exists in a predetermined area of a search frame. And when it is determined in the first step that there is a similar part, the search frame includes a first area that includes the similar part and is smaller than the predetermined area. A second step of setting a search range for the motion vector, and, when it is determined that there is no similar part in the first step, setting an area larger than the first area as a search range for detecting a motion vector. A motion vector detecting method comprising the third step of: 10. The motion vector detecting method according to claim 9, wherein the first step sets a block located at the same position as or adjacent to the current block to be coded in a previous frame of the current frame as a range setting reference block, The determination is performed based on the correlation degree evaluation value corresponding to the motion vector detected for the range setting reference block, and the second step includes setting a position of a search range to the range setting reference block. A motion vector detection method characterized by setting based on a motion vector detected for the motion vector. 11. The motion vector detecting method according to claim 9, wherein the first step includes: a unit area including at least a part of the current block and a vicinity of the same position as the current block in the search frame. Performing a predetermined pixel operation on each of a plurality of comparison regions having the same size as the unit region provided in the above, and performing the determination based on the results of these operations. The second step includes: From the calculation result of the predetermined pixel calculation, the one that is most similar to the unit area among the plurality of comparison areas is determined, and the position of the search range is set based on the position of the comparison area. Characteristic motion vector detection method. 12. A step of setting a search range for detecting a motion vector, and a step of detecting a motion vector based on a degree of correlation between a coding target block and a candidate block in the search range. The search range setting step comprises: setting a block at the same position as or adjacent to the current block to be coded in the previous frame of the target frame as a range setting reference block, and detecting a size of the search range with respect to the range setting reference block. A motion vector detecting method for setting based on a correlation evaluation value corresponding to the calculated motion vector. 13. A step of setting a search range for detecting a motion vector, and a step of detecting a motion vector based on a degree of correlation between a coding target block and a candidate block in the search range. The search range setting step includes: setting a block at the same position as or adjacent to the current block to be coded in the previous frame of the target frame as a range setting reference block, and detecting a position of the search range with respect to the range setting reference block. A motion vector detection method characterized in that the motion vector is set on the basis of the obtained motion vector.