JP2005006351A - Motion vector detection apparatus and motion vector detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a motion vector detection device capable of speeding up motion vector detection processing and reducing power consumption.
A motion vector detection device and a motion vector detection method according to the present invention perform block matching between a processing target block and a reference block, and perform motion vector detection in the motion vector detection device. In addition to controlling the data transfer, the processing steps at the time of motion vector detection and the amount of data to be processed are reduced, thereby speeding up the motion vector detection process and reducing the power consumption.
[Selection] Figure 5

Description

  The present invention relates to image compression processing, and more particularly to motion vector detection processing applicable to H.261, H.263, MPEG4, and the like, which are standards for low bit rates intended for image communication. .

Currently, standards for low bit rates for image communication, such as H.261, H.263, and MPEG4, detect motion vectors for compression using mutual frames.
In general, the motion vector detection processing is performed on a block composed of a plurality of pixels in the current image, and the processing target block is located in the temporally previous image with respect to the current image. A motion vector indicating the direction and amount of motion of an image is detected by a matching method.

  FIG. 31 is an explanatory diagram for explaining a conventional general motion vector detection process. In the figure, the processing target block T is a block to be encoded in the current image composed of NX × NY pixels, and the reference area is in the vicinity of the same spatial position as the processing target block T and the processing target. This is a search area of a previous image made up of MX × MY pixels that is larger than a block. The reference block I is a block at an arbitrary position on the reference area composed of NX × NY pixels.

In the general motion vector detection by the block matching method, the processing target block T is compared with the reference block I in the reference area, and the position of the reference block I most matching with the processing target block T is obtained. . Thereafter, the distance and direction between the position of the processing target block T and the position of the reference block I having the best matching in the reference area are detected as a motion vector. Note that the matching between the processing target block T and the reference block I is an absolute difference error sum that is the sum of the absolute values of the pixel value differences of the pixels constituting the block, as shown in Equation (1) below. It is done by using.
Absolute difference error sum = Σ | Ref (Mx + x, My + y) −Src (x, y) | (1)

  Here, Ref (Mx + x, My + y) in the expression (1) indicates the pixel position of the reference block, and Src (x, y) indicates the pixel position of the processing target block.

  When it is desired to detect a more accurate motion vector, the pixel value at a position between the pixel data in the reference area may be obtained by performing an interpolation process, and the position may be matched with the processing target block.

  The calculation of the absolute difference error sum described above is the sum of the absolute values of the differences between the pixel values of the NX × NY pixels constituting the processing target block, and the same processing is performed on all the reference blocks in the reference region. A huge amount of calculation is required.

  Normally, when the absolute difference error sum at a certain search position is relatively large, the absolute difference error sum is already exceeded during calculation of the absolute difference error sum. For this reason, although the position of the search region can be excluded from the search position candidates for detecting the motion vector, a wasteful calculation has been executed, so that a loss has occurred in terms of calculation speed and power consumption.

  In addition, the motion vector generally shows little change in the vicinity of the processing target block for which the motion vector is searched. Especially when a high-resolution image is not required, However, the search for the motion vector is similarly performed, and a loss occurs in terms of calculation speed and power consumption.

  On the other hand, in order to eliminate this loss, before calculating using the dedicated calculation mechanism, the general-purpose calculation mechanism may check whether there is a change and then decide whether to execute the search. This puts a load on the general-purpose arithmetic unit.

  On the other hand, in order to eliminate this load, if the dedicated arithmetic unit has the function of calculating the absolute difference error sum only once, if there is no change and there is no need to search, the dedicated arithmetic mechanism In order to obtain the result, a loss occurred in terms of calculation speed and power consumption.

  In addition, for example, when the reference block search for the processing target block proceeds in a direction in which the absolute difference error sum is minimized with respect to the surrounding reference blocks of the processing target block, the reference block once searched is 2 There have been cases in which search is heavily performed, and a loss has occurred in terms of calculation speed and power consumption.

  In addition, since the absolute difference error sum is calculated by calculating the sum of absolute values of the differences between the pixel values for all the pixels constituting the block, even when a high-resolution video is not required. More computations than necessary are performed, and loss occurs in terms of computation speed and power consumption.

  In addition, in order to eliminate this loss, some sub-sampling is performed on every other pixel in the block to reduce the amount of calculation. In this case, it is used for motion vector detection. Since the pixel data to be processed becomes a part of the read pixel data, the other pixel data that has been read and is not used for motion vector detection is discarded, resulting in a loss in terms of power consumption. .

  In addition, when the position of the processing target block comes to the end position of the effective area range of the image data, since there are areas where motion vector search cannot be performed in the reference area, data is transferred to this area. However, there was a loss in terms of power consumption without being utilized for searching.

  In addition, since there is an overlapping area between the reference areas of neighboring processing target blocks, reading the pixel data of the entire reference area into the dedicated calculation unit for each search of each processing block is the calculation speed and consumption. There was a loss in terms of power.

  In addition, when the position of the processing block comes to the end of the effective area range of the image data, depending on the position of the motion vector, the image data of the reference area used for the calculation comes out of the effective area range, and the motion vector search is performed. It had the problem that it was not possible.

  In addition, when the pixel value at the position between the pixels constituting the reference block is obtained by performing the interpolation process, generally, the pixel data constituting the reference block is read in one reading direction. When the pixel data is read in the direction, it is necessary to wait several cycles until the pixels necessary for the vertical interpolation process are prepared in order to obtain the pixel value at the position between the vertical pixels. There was a loss.

  The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a motion vector detection device capable of speeding up motion vector detection processing and reducing power consumption.

  In order to solve the above-described problem, a motion vector detection device according to claim 1 of the present invention includes a processing target block composed of a plurality of pixels in a current image, and a temporally previous block with respect to the current image. In the motion vector detection apparatus for detecting a motion vector by performing block matching with a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image, an address of the processing target block and the reference area data A first address generation unit that generates a reference area data instructed by the first address generation unit, and a processing target block instructed by the first address generation unit. A second storage unit that holds data; a first address generation unit that generates an address of data output from the first storage unit and the second storage unit; A motion vector detection unit that detects a motion vector using the data output from the first storage unit and the data output from the second storage unit, and the motion vector detection unit includes: When the absolute value of the pixel value difference between the pixel data and the pixel data of the processing target block is sequentially calculated, the absolute value of the calculated difference is subtracted from the threshold value, and the subtraction result becomes negative, The calculation of the absolute difference error sum for the reference block being processed is stopped.

  According to a second aspect of the present invention, there is provided a motion vector detection device comprising: a processing target block comprising a plurality of pixels in a current image; and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit Output from the first storage unit, a second address generation unit that generates an address of data output from the first storage unit, and the second storage unit A motion vector detection unit that detects a motion vector using the received data and data output from the second storage unit, and a search for a reference block in the reference area data that has already been searched by the motion vector detection unit A search history holding unit for holding a history, wherein the motion vector detection unit skips search execution for a reference block having a search history held by the search history holding unit. To do.

  The motion vector detection device according to claim 3 of the present invention is the motion vector detection device according to claim 2, wherein the search history holding unit includes a reference block at a search target position and eight peripheral positions thereof. The presence / absence of a search for is stored as a search history.

  According to a fourth aspect of the present invention, there is provided a motion vector detection apparatus comprising: a processing target block including a plurality of pixels in a current image; and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit Output from the first storage unit, a second address generation unit that generates an address of data output from the first storage unit, and the second storage unit A motion vector detection unit that detects a motion vector using the output data and data output from the second storage unit, and reference area data that the first address generation unit outputs to the first storage unit Based on a transfer rule holding unit that holds a transfer rule that is a rule for generating an address, and an area determination code that indicates a valid area of image data, the transfer rule holding unit detects the transfer rule, and A transfer rule detection unit that outputs to the address generation unit, and the first address generation unit outputs to the first storage unit based on the transfer rule output from the transfer rule detection unit An address of reference area data is generated.

  According to a fifth aspect of the present invention, there is provided a motion vector detection apparatus comprising: a processing target block comprising a plurality of pixels in a current image; and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data And a memory having a double buffer configuration for simultaneously reading and reading data, the first storage unit holding the reference area data instructed by the first address generation unit, and the first A second storage unit that holds data of the block to be processed instructed by the address generation unit, the first storage unit, and the second storage unit A second address generation unit that generates an address of data to be input, a data output from the first storage unit, and a motion vector detection unit that detects a motion vector from the data output from the second storage unit; A storage data control unit that controls reading and reading of reference area data from and to the first storage unit, and the storage data control unit is one of the two buffer memories held by the first storage unit. A part of valid data is copied from one buffer memory that has been read to one buffer memory from which data is read, and the amount of data newly written to the first storage unit is reduced. It is characterized by this.

  The motion vector detection device according to claim 6 of the present invention is the motion vector detection device according to any one of claims 1 to 5, wherein the motion vector detection unit is stored in the first storage unit. It is an integer accuracy motion vector detection unit that detects an integer accuracy motion vector from the output data and the data output from the second storage unit.

  The motion vector detection device according to claim 7 of the present invention is the motion vector detection device according to any one of claims 1 to 5, wherein the motion vector detection unit is stored in the first storage unit. Input data, and an interpolation pixel generation unit that generates pixel data with a small number of precision, a data output from the interpolation pixel generation unit, and a motion vector with a smaller number of precision than the data output from the second storage unit It comprises a small-precision motion vector detection unit for detection.

  The motion vector detection device according to claim 8 of the present invention is the motion vector detection device according to any one of claims 1 to 5, wherein the motion vector detection unit is stored in the first storage unit. An integer precision motion vector detection unit that detects an integer precision motion vector from the output data and the data output from the second storage unit, and the output data from the first storage unit as inputs, From an interpolation pixel generation unit that generates pixel data, a data output from the interpolation pixel generation unit, and a minority motion vector detection unit that detects a motion vector with a lower precision than the data output from the second storage unit It is characterized by.

  According to a ninth aspect of the present invention, there is provided a motion vector detecting apparatus comprising: a processing target block comprising a plurality of pixels in a current image; and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit A second address generation unit that generates an address of data output from the first storage unit and the second storage unit, and an output from the first storage unit. Interpolation pixel generation unit that receives data as input and generates data of a minority precision pixel, and detects a motion vector with a minority precision from the data output from the interpolation pixel generation unit and the data output from the second storage unit The read direction of the reference area data held by the first storage unit is determined according to the position of the data of the minority precision pixel generated by the minority motion vector detection unit and the interpolation pixel generation unit, and the second address generation unit The second address generation unit generates an address of pixel data output from the first storage unit based on the read direction determined by the read direction determination unit. It is characterized by doing.

  The motion vector detection device according to claim 10 of the present invention is the motion vector detection device according to claim 9, wherein the reading direction determination unit applies to the image data stored in the first storage unit. It is characterized in that a decision is made to read data in the horizontal or vertical direction.

  The motion vector detection device according to claim 11 of the present invention is the motion vector detection device according to claim 9 or 10, wherein the interpolation pixel generation unit includes at least two shift registers. It has a shift register group and an interpolation processing unit for performing an interpolation process using pixel data held by the shift register group.

  The motion vector detection device according to claim 12 of the present invention is the motion vector detection device according to claim 9 or 10, wherein the interpolated pixel generation unit includes at least the number of pixels read in a specific direction +2. A shift register group including a plurality of shift registers, and an interpolation processing unit that performs an interpolation process using pixel data held by the shift register group.

  According to a thirteenth aspect of the present invention, there is provided a motion vector detection apparatus comprising: a processing target block including a plurality of pixels in a current image; and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit From the first storage unit, a second address generation unit that generates an address of data output from the first storage unit, and the second storage unit Force data is input, and an interpolation pixel generation unit that generates data of a minority precision pixel, and a motion vector with a minority precision are detected from the data output from the interpolation pixel generation unit and the data output from the second storage unit. A small-precision motion vector detection unit, and the minority-precision motion vector detection unit is composed of a plurality of pixel data from the interpolation pixel generation unit, upper and lower, left and right, right upper right and lower right diagonal, If a decimal precision reference block is output that is located at either upper left or lower left, upper right or lower left, or lower right or lower left, the decimal precision is output. It is characterized in that detection of a small number of precision motion vectors for the reference blocks is executed in parallel.

  The motion vector detection device according to claim 14 of the present invention is the motion vector detection device according to any one of claims 9 to 13, further comprising data output from the first storage unit, And an integer precision motion vector detection unit for detecting an integer precision motion vector from the data output from the second storage unit.

  The motion vector detection device according to claim 15 of the present invention is a processing target block composed of a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit A second address generation unit that generates an address of data output from the first storage unit and the second storage unit, and an output from the first storage unit. An integer-precision motion vector detection unit that detects an integer-precision motion vector from the received data and the data output from the second storage unit, and a search that is a condition for determining whether or not the position can be searched A search availability condition is detected from the search availability condition holding unit based on a search availability condition holding unit that holds availability conditions and an area determination code indicating an effective area of the reference area data, and the second address generation unit A search availability condition detection unit for outputting, and the second address generation unit determines the availability of the search using the search availability condition detected by the search availability condition detection unit when detecting the integer precision motion vector. It is characterized by this.

  According to a sixteenth aspect of the present invention, in the motion vector detecting device according to the fifteenth aspect, the output data from the first storage means is further inputted, and pixel data with a small number of precision is input. An interpolation pixel generation unit that generates a motion vector, and a minority precision motion vector detection unit that detects a minority motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit. The second address generation unit uses the search availability condition holding unit and the search availability condition detection unit in combination when detecting an integer precision motion vector and a decimal precision motion vector. To do.

  The motion vector detection device according to claim 17 of the present invention is the motion vector detection device according to claim 15, further comprising the output data from the first storage means as input, and pixel data with a small number of precision. An interpolation pixel generation unit that generates a motion vector, and a minority precision motion vector detection unit that detects a minority motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit. The second address generation unit uses the search availability condition holding unit and the search availability condition detection unit at the time of detecting an integer precision motion vector to determine the availability of the search and uses the search availability result Thus, it is characterized by determining whether or not search is possible when detecting a small-precision motion vector.

  The motion vector detection apparatus according to claim 18 of the present invention is a processing target block composed of a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. First address generation for generating an address of the processing target block and the reference area data in a motion vector detection device that detects a motion vector by performing block matching with a reference block including a plurality of pixels in the data , A first storage unit that holds reference area data instructed by the first address generation unit, and a second storage that holds data of the processing target block instructed by the first address generation unit A second address generation unit that generates an address of data output from the first storage unit and the second storage unit, and an output from the first storage unit. An integer precision motion vector detection unit for detecting an integer precision motion vector using the output data and the data output from the second storage unit, and holding the motion vector detected in the previous motion vector search, A search execution enable / disable determining unit that determines whether or not to execute a motion vector search by comparing an absolute difference error value calculated by the integer precision motion vector detection unit with a specific threshold; The address generation means uses the motion vector detected in the previous motion vector search held by the search execution feasibility determination unit to determine the address of the data output from the first storage unit and the second storage unit. It is characterized by generating.

  The motion vector detection device according to claim 19 of the present invention is the motion vector detection device according to claim 18, wherein the second address generation unit further outputs data output from the first storage means. Is output as a calculation result of the present apparatus.

  The motion vector detection device according to claim 20 of the present invention is the motion vector detection device according to claim 18 or claim 19, further comprising the output data from the first storage unit as input, Interpolation pixel generation unit that generates pixel data with minority precision, and minority motion vector detection that detects a motion vector with minority precision from data output from the interpolation pixel generation unit and data output from the second storage unit It comprises a part.

  A motion vector detection method according to a twenty-first aspect of the present invention includes a processing target block including a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step, and a motion vector detection step of detecting a motion vector using the reference area data and data of the processing target block, wherein the motion vector detection step includes pixel data of the reference block If the absolute value of the difference between the pixel value of the target block and the pixel data of the processing target block is sequentially calculated, and the absolute value of the calculated difference is subtracted from the threshold value, the subtraction result becomes negative. The calculation of the absolute difference error sum for the reference block is stopped.

  A motion vector detection method according to a twenty-second aspect of the present invention includes a processing target block including a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step, a motion vector detection step for detecting a motion vector using the reference region data and processing target block data, and reference region data that has already been searched by the motion vector detection step A search history holding step for holding a search history of the reference block in the search block, and the motion vector detecting step performs a search execution for the reference block having the search history held by the search history holding step. It is characterized by skipping.

  The motion vector detection method according to claim 23 of the present invention is the motion vector detection method according to claim 22, wherein the search history holding step includes a reference block of a search target position and eight peripheral positions thereof. The presence / absence of a search for is stored as a search history.

  A motion vector detection method according to a twenty-fourth aspect of the present invention includes a processing target block composed of a plurality of pixels in a current image and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step, a motion vector detection step for detecting a motion vector using the reference area data and the data of the processing target block, and the first address generation step generates an address of the reference area data A transfer rule holding step for holding a transfer rule that is a rule for performing the transfer, and a transfer rule holding step for detecting the transfer rule based on an area determination code indicating a valid area of the image data, and generating the first address A transfer rule detection step for outputting to the step, wherein the first address generation step generates an address of reference area data based on the transfer rule detected by the transfer rule detection step. To do.

  A motion vector detection method according to a twenty-fifth aspect of the present invention includes a processing target block composed of a plurality of pixels in a current image and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data And a first storage step for holding the reference area data instructed by the first address generation step using a memory having a double buffer configuration for simultaneously reading and reading data, and the first storage step A second storage step for holding the data of the processing target block instructed by the address generation step; and A second address generation step of generating an address of pixel data to be output from the reference area data and the data of the processing target block, and a motion vector for detecting a motion vector using the reference area data and the data of the processing target block And a storage data control step for controlling reading and reading of the reference area data in the first storage step, wherein the storage data control step is used by the first storage step. Of the one buffer memory, one piece of valid data is copied from the other buffer memory that has been read into one buffer memory from which data is read, and newly written to the first storage step. It is characterized by reducing the amount of data.

  The motion vector detection method according to claim 26 of the present invention is the motion vector detection method according to any one of claims 21 to 25, wherein the motion vector detection step includes the reference region data and a process. This is an integer precision motion vector detection step of detecting an integer precision motion vector using the data of the target block.

  The motion vector detection method according to claim 27 of the present invention is the motion vector detection method according to any of claims 21 to 25, wherein the motion vector detection step constitutes the reference region data. Interpolation pixel generation step that generates pixel data with pixel data as input, and minority accuracy that detects a motion vector with minority accuracy using the data output from the interpolation pixel generation step and the data of the processing target block It comprises a motion vector detection step.

  The motion vector detection method according to claim 28 of the present invention is the motion vector detection method according to any of claims 21 to 25, wherein the motion vector detection step includes the reference region data and a process. An integer-precision motion vector detection step for detecting an integer-precision motion vector using the data of the target block; and an interpolation pixel generation step for generating pixel data with a small number of precision using pixel data constituting the reference area data as an input; The method includes a minority precision motion vector detection step of detecting a minority precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block.

  A motion vector detection method according to a twenty-ninth aspect of the present invention includes a processing target block including a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step to be generated, an interpolation pixel generation step for generating pixel data with a small number of precision by using pixel data constituting the reference region data, data output from the interpolation pixel generation step, and processing The minority-precision motion vector detection step for detecting the minority-precision motion vector using the data of the target block, and the position of the minority-precision pixel data generated in the interpolation pixel generation step are held in the first storage step. A reading direction determining step for determining a reading direction of the reference area data and outputting to the second address generating step, wherein the second address generating step is based on the reading direction determined by the reading direction determining step. To generate an address of pixel data to be output from the reference area data. It is characterized in that.

  The motion vector detection method according to claim 30 of the present invention is the motion vector detection method according to claim 29, wherein the reading direction determination step is horizontal to the image data held in the first storage step. Alternatively, it is determined that data is read in the vertical direction.

  A motion vector detection method according to a thirty-first aspect of the present invention includes a processing target block composed of a plurality of pixels in a current image and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step to be generated, an interpolation pixel generation step for generating pixel data with a small number of precision by using pixel data constituting the reference region data, data output from the interpolation pixel generation step, and processing A minority-precision motion vector detection step for detecting a minority-precision motion vector using the data of the target block, and the minority-precision motion vector detection step is constituted by a plurality of pixel data from the interpolation pixel generation step. There are minority reference blocks located either on the top and bottom, left and right, top right and bottom right, top left and bottom left, top right and top left, or bottom right and bottom left. When output, the detection of the minority precision motion vector for each of the outputted minority reference blocks is performed in parallel. Than it is.

  A motion vector detection method according to a thirty-second aspect of the present invention is the motion vector detection method according to any one of the thirty-third to thirty-first aspects, wherein the reference region data and the processing target block data are further converted. An integer precision motion vector detection step is used to detect an integer precision motion vector.

  A motion vector detection method according to a thirty-third aspect of the present invention includes a processing target block including a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step, an integer accuracy motion vector detection step for detecting an integer accuracy motion vector using the reference area data and the data of the block to be processed, and whether or not the search can be performed Based on a search availability condition holding step that holds a search availability condition that is a condition for determining, and an area determination code that indicates an effective area of the reference area data, a search availability condition is detected from the search availability condition holding step, A search availability condition detection step for outputting to the second address generation step, wherein the second address generation step uses the search availability condition detected by the search availability condition detection step when detecting an integer precision motion vector. Thus, it is characterized by determining whether or not search is possible.

  A motion vector detection method according to a thirty-fourth aspect of the present invention is the motion vector detection method according to the thirty-third aspect, further comprising: inputting pixel data constituting the reference area data; An interpolation pixel generation step to generate; a minority precision motion vector detection step to detect a motion vector with a minority precision using the data output from the interpolation pixel generation step and the data of the processing target block; and The address generation step is characterized in that the search availability condition holding step and the search availability condition detection step are used in combination when detecting an integer precision motion vector and detecting a decimal precision motion vector.

  A motion vector detection method according to a thirty-fifth aspect of the present invention is the motion vector detection method according to the thirty-third aspect, further comprising: inputting pixel data constituting the reference area data; An interpolation pixel generation step to generate; a minority precision motion vector detection step to detect a motion vector with a minority precision using the data output from the interpolation pixel generation step and the data of the processing target block; and The address generation step uses the search availability condition holding step and the search availability condition detection step when detecting an integer precision motion vector, and determines whether or not a search is possible, and uses the search availability result to determine a decimal precision motion vector. It is characterized by determining whether or not search is possible at the time of detection.

  A motion vector detection method according to a thirty-sixth aspect of the present invention includes a processing target block including a plurality of pixels in a current image, and a predetermined reference area of a previous image temporally preceding the current image. In a motion vector detection method for detecting a motion vector by performing block matching with a reference block including a plurality of pixels in data, first address generation for generating an address of the processing target block and the reference area data A first storage step for holding the reference area data instructed by the first address generation step, and a second storage for holding the data of the processing target block instructed by the first address generation step Step, and the address of the pixel data to be output from the stored reference area data and the data of the block to be processed A second address generation step, an integer accuracy motion vector detection step for detecting an integer accuracy motion vector using the reference area data and the data of the processing target block, and a motion detected in the previous motion vector search A search execution enable / disable determining step of determining whether or not to execute a motion vector search by holding a vector and comparing the absolute difference error value calculated by the integer precision motion vector detection step with a specific threshold value. The second address generation step uses the motion vector detected in the previous motion vector search held by the search execution possibility determination step, and outputs the pixel from the reference area data and the data of the processing target block. A data address is generated.

  The motion vector detection method according to claim 37 of the present invention is the motion vector detection method according to claim 36, wherein the second address generation step is further held by the first storage step. Data output from the area data is output as a calculation result.

  The motion vector detection method according to claim 38 of the present invention is the motion vector detection method according to claim 36 or claim 37, further comprising pixel data constituting the reference region data as an input, and a small number. An interpolation pixel generation step for generating accurate pixel data; and a minority precision motion vector detection step for detecting a minority motion vector using the data output from the interpolation pixel generation step and the data of the processing target block. It is characterized by this.

  According to the motion vector detection apparatus and method according to claim 1 or 21 of the present invention, the integer precision motion vector detection unit 206 sequentially subtracts the absolute difference error value from the threshold value, and the threshold value becomes negative. At this point, the processing block can be improved and the power consumption can be reduced by excluding the reference block candidate that is the minimum absolute difference error sum for the processing target block and stopping the calculation in the middle.

  In addition, according to the motion vector detection device and method according to claim 2 or claim 22 of the present invention, by leaving a search history for a reference block that has already been searched, A double search can be prevented. Thereby, the processing speed can be improved and the power consumption can be reduced.

  In addition, according to the motion vector detection device and method according to claim 3 or claim 23 of the present invention, by leaving a search history for a reference block that has already been searched, the same reference block can be stored. A double search can be prevented. Thereby, the processing speed can be improved and the power consumption can be reduced.

  According to the motion vector detection device and method according to claim 4 or claim 24 of the present invention, the first address generation unit has the first address generator in accordance with the position of the reference area data on the image data. By changing the transfer rule of the reference area data to be output to the storage unit, the amount of data transferred to the first storage unit can be minimized, and the processing speed can be improved and the power consumption can be reduced. it can.

  According to the motion vector detecting apparatus and method of claim 5 or 25 of the present invention, data is read out of two buffer memories held by the first storage means. By copying a part of valid data from another buffer memory that has been read to the buffer memory, the amount of data newly read into the first storage unit can be reduced, and the processing speed can be reduced. Can be improved and power consumption can be reduced.

  According to the motion vector detection device and method according to claim 6 or 26 of the present invention, in the motion vector detection device for detecting an integer precision motion vector, the motion vector detection device according to claim 1, Alternatively, the same effect as that of the invention described in claims 21 to 25 can be obtained.

  According to the motion vector detection device and method according to claim 7 or claim 27 of the present invention, in the motion vector detection device for detecting a motion vector with a small number of precision, the motion vector detection device according to claims 1 to 5, Alternatively, the same effect as that of the invention described in claims 21 to 25 can be obtained.

  According to the motion vector detection device and method according to claim 8 or claim 28 of the present invention, in the motion vector detection device capable of detecting an integer precision motion vector and a decimal precision motion vector, The same effects as those of the inventions according to claims 1 to 5 or claims 21 to 25 can be obtained.

  Further, according to the motion vector detection device and method according to claim 9 or 29 of the present invention, by changing the data reading direction with respect to the image image according to the interpolation position of the pixel, the interpolation pixel generation unit It is possible to generate small-precision pixel data shifted by a specific distance in the vertical or horizontal direction at the same time without changing the configuration, and at the same time, it is possible to improve processing speed and reduce power consumption. .

  In addition, according to the motion vector detection device and method according to claim 10 or 30 of the present invention, the data reading direction with respect to the image image is changed to either horizontal or vertical depending on the interpolation position of the pixel. This makes it possible to generate small-precision pixel data with a specific distance shifted in the vertical or horizontal direction at high speed and at the same timing without changing the configuration of the interpolated pixel generation unit. Electric power can be reduced.

  According to the motion vector detection device of the present invention, at least two shift register groups are shifted when generating pixel data with small precision shifted by a specific distance in the vertical or horizontal direction. By having a register, it becomes possible to generate pixel data with a small number of precision at high speed and at the same timing, and it is possible to reduce the amount of data transfer, improve the processing speed, and reduce power consumption.

  According to the motion vector detection device of the twelfth aspect of the present invention, at least the number of pixels read out in the specific direction by the shift register group +2 when generating pixel data with a small number of precisions shifted by a specific distance in the oblique direction. By having one shift register, it is possible to generate pixel data with a small number of precision at high speed and at the same timing, and it is possible to reduce the amount of data transfer and the processing speed and reduce power consumption.

  According to the motion vector detection apparatus and method of claim 13 or claim 31 of the present invention, the first memory is used to generate the decimal interpolation pixels used for the detection of the minority motion vectors. Even when all pixel data constituting the reference block stored in the unit is read out, up and down, left and right, right up and down right, left up and down left, right up and down left Reduces the amount of processing of the interpolation pixel generation unit by executing in parallel the calculation processing of the absolute difference error of the two half-pixel accuracy processing target blocks shifted by half a pixel in the upper, lower right, lower left, etc. In addition, the amount of data read from the first storage unit can be reduced. Therefore, the processing speed of the motion vector detection device can be improved and the power consumption can be reduced.

  According to the motion vector detection device and method according to claim 14 or claim 32 of the present invention, further comprising an integer precision motion vector detection unit that detects an integer precision motion vector. In the motion vector detection device according to any one of claims 9 to 11, an integer precision motion vector search can be performed.

  According to the motion vector detection device and method according to claim 15 or 33 of the present invention, the search availability condition detection unit detects a search availability condition from the search availability condition holding unit, and the search availability condition By generating the address of the pixel data output from the first storage unit by the second address generation unit based on the above, it is possible to accurately generate the pixel address of an area where an effective search can be performed, which is useless Processing can be performed at high speed without performing search processing.

  Further, according to the motion vector detection device and method according to claim 16 or claim 34 of the present invention, the search availability condition holding unit and the search availability condition detection unit used for integer precision motion vector detection are provided with decimal precision. By using it together with motion vector detection, the circuit scale can be reduced and the manufacturing cost can be reduced.

  According to the motion vector detection apparatus and method according to claim 17 or claim 35 of the present invention, the second address generation unit detects the search availability condition holding unit when detecting an integer precision motion vector, and A search availability condition detection unit is used to determine whether or not a search is possible, and by using the search availability result to determine whether or not a search is possible at the time of detection of a small number of precision motion vectors, the detection of a minority precision motion vector is performed again. Whether or not search is possible can be determined without detecting the search availability condition, and the processing speed can be improved. Further, since the search result at the time of detecting an integer precision motion vector is also used at the time of detecting a decimal precision motion vector, the circuit scale can be reduced and the manufacturing cost can be reduced.

  In addition, according to the motion vector detection device and method according to claim 18 or 36 of the present invention, a position where a motion vector is highly likely to be detected (for example, the motion vector of the block searched immediately before) It is possible to improve the processing speed and reduce power consumption by performing block matching on the reference block at the same position and determining whether or not the subsequent search execution is possible based on the matching result (absolute difference error sum). it can.

  Further, according to the motion vector detection device and method according to claim 19 or 37 of the present invention, even if the motion vector detected in the previous motion vector search is used, the second address is used. The generation unit outputs the pixel data indicated by the motion vector position detected in the previous search to the external device by outputting the data output from the first storage unit as the calculation result of the device. be able to.

  According to the motion vector detection apparatus and method according to claim 20 or 38 of the present invention, the effect of the invention according to claim 18, 19, 36, or 37. In addition, a motion vector with a small number of precision can be detected.

(Embodiment 1)
Hereinafter, a motion vector detection apparatus according to Embodiment 1 of the present invention will be described with reference to FIGS.
FIG. 1 is a block diagram showing a configuration of a motion vector detection device according to Embodiment 1 of the present invention. In FIG. 1, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, And an integer precision motion vector detection unit 106.

  The data holding unit 101 is a large-capacity memory that holds image data. In accordance with an instruction from the first address generation unit, the data in the reference area is stored in the first storage unit and the data in the processing target block is stored in the second storage unit. Output to.

  The first address generation unit 102 searches for data of a processing target block that is data to be encoded, and a reference block that performs block matching with the processing target block, from the image data stored in the data storage unit 101. The address of the reference area data that is the area to be performed is generated.

  In the motion vector detection device of the present invention described below, the processing target block and the reference block are image units composed of a plurality of pixel groups. Here, for example, one block is 16 pixels long and 16 pixels wide. A description will be given by taking an example composed of a total of 256 pixels.

  The first storage unit 103 is a storage region that holds the reference region data instructed by the first address generation unit 102, and performs block matching with the processing target block according to the instruction of the second address generation unit 105. The pixel data of the reference block is output.

  In the motion vector detection device of the present invention described below, the motion vector range to be detected by the motion vector detection device is -16 to +15.5, and the first search region corresponding to the motion vector range is the first. The pixel data of the reference area held by the storage unit 103 will be described by taking, as an example, data composed of 48 vertical pixels and 48 horizontal pixels centering on the processing target block position.

  The second storage unit 104 is a storage area that holds the data of the processing target block instructed by the first address generation unit 102, and stores the pixel data of the processing target block in accordance with the instruction of the second address generation unit 105. Output.

  The second address generation unit 105 generates the addresses of the pixel data output from the first storage unit 103 and the second storage unit 104, and notifies the integer precision motion vector detection unit 106 that the search process has been completed. Output.

  The search process is a process for searching a reference block having a minimum absolute difference error with respect to the processing target block from the reference region. This search process ends comparison with a specific number of reference blocks. Or the comparison with all the reference blocks in the search area range ends, or the absolute difference error of the reference block at the specific position with respect to the processing target block The process ends when it is detected that the error is smaller than the difference error.

  The integer precision motion vector detection unit 106 uses the pixel data of the reference block output from the first storage unit 103 and the pixel data of the processing target block output from the second storage unit 104, to the processing target block. The absolute difference error sum of the reference block is calculated, the reference block that minimizes the absolute difference error sum is searched, and the motion vector is detected.

  Next, the second address generation unit 105, which is a characteristic part of the motion vector detection device according to the first embodiment of the present invention, will be described in detail.

  Hereinafter, pixel data address generation processing by the second address generation unit 105 will be described with reference to FIGS. Note that the second address generation unit 105 according to the first embodiment performs sampling in a checkered pattern with respect to the arrangement of pixels in the image.

  FIG. 2 is a block diagram showing a configuration of second address generation section 105 according to Embodiment 1 of the present invention. FIG. 3 shows subsampling performed by second address generation section 105 according to Embodiment 1 of the present invention. It is a figure which shows an example of the macroblock which performs. As shown in FIG. 3, the address value of each pixel is set to an address value from 0 to 255 with an initial value of 0.

  In FIG. 2, the second address generation unit 105 includes a column counter 11, a row counter 12, an address holding unit 13, and an adder 14.

  The row counter 11 and the column counter 12 are both down counters, and a row counter value and a column counter value are set as shown in FIG. Further, the row counter 11 is decremented each time an address of pixel data is output from the second address generation unit 105. On the other hand, the column counter 12 is decremented when the row counter 11 becomes 0, and every time the row counter 11 is decremented, the LSB (the least significant bit) of the column counter value is added to the adder 14. ) Is inverted and output.

  The address holding unit 13 holds an address value (hereinafter, an address value held in advance is referred to as an initial address value), and when the row counter 11 is decremented, the initial address value held by the address holding unit 13 is increased by two. The address value that has been corrected is output to the adder 14, and the initial address value held by the address holding unit 13 is updated to the address value output to the adder 14. Note that the initial value of the initial address value held by the address holding unit 13 is zero.

  The adder 14 adds the inverted value of the LSB (least significant bit) of the column counter value output from the column counter 12 to the address value output from the address holding unit 13 and outputs the result as a pixel data address value. That is, when the LSB (least significant bit) of the column counter 12 output from the column counter 12 is “0”, that is, when the column counter value is an even number, “0” is set to the address value output from the address holding unit 13. The inverted “1” is added and output as a pixel data address value. On the other hand, when the LSB (least significant bit) of the column counter 12 output from the column counter 12 is “1”, that is, when the column counter value is an odd number, the address value output from the address holding unit 13 is set to “1”. “0” obtained by inverting is added and output as a pixel data address value.

Next, the operation of the second address generation unit 105 will be described with reference to FIGS.
FIG. 4 is a flowchart for explaining an address generation process of pixel data of the second address generation unit 105 according to the first embodiment of the present invention.

  When the row counter 11 is decremented (step S101), the column counter 12 determines whether or not the row counter value is “0” (step S102).

  At this time, if the row counter value is “0”, the column counter 12 is decremented (step S103), and the inverted value of the LSB (least significant bit) of the decremented column counter value is output to the adder 14 ( Step S104).

  If the row counter value is not “0”, the column counter value of the column counter 12 is not changed, and the inverted value of the LSB (least significant bit) of the column counter value is output to the adder 14 (step S104). ).

  On the other hand, when the row counter 11 is decremented (step S101), the address holding unit 13 outputs a value obtained by adding 2 to the held initial address value to the adder 14 and the initial value held by the address holding unit 13 The address value is updated to the address value output from the address holding unit 13, that is, the initial address value held by the address holding unit 13 plus 2 (step 105).

  Next, the adder 14 adds the inverted value of the LSB (least significant bit) of the column counter value output from the column counter 12 in step S104 to the address value output from the address holding unit 13 (step S106). ) And output as a pixel data address value (step S107).

  Thus, by providing the row counter 11, the column counter 12, the address holding unit 13, and the adder 14, the position of the pixel data can be subsampled in a checkered pattern with respect to the pixel arrangement in the image image. Compared to the case of obtaining the absolute difference error sum for all the constituent pixels, it is possible to obtain the same level of accuracy in the case of relatively high compression, and is necessary for obtaining the absolute difference error sum. The calculation amount can be halved, leading to an improvement in processing speed and a reduction in power consumption.

  In addition, compared to the case where only the horizontal component or the vertical component is sub-sampled to reduce the data processing amount, it is less affected by the pixel distribution. In addition, the absolute difference error sum can be obtained more accurately and the motion vector can be obtained more accurately than in the case where sub-sampling is performed at the same ratio in the horizontal direction and the vertical direction to reduce the data processing amount.

  In the first embodiment, the second address generation unit 105 samples the pixel data positions in a checkered pattern with respect to the pixel arrangement in the image image. However, the present embodiment is not limited to this. According to the second address generation unit 105 by 1, the sub-sampling can be freely performed on the pixel arrangement in the image image by arbitrarily changing the setting values of the row counter 11, the column counter 12, and the address holding unit 13. Can do.

(Embodiment 2)
Hereinafter, a motion vector detection apparatus according to Embodiment 2 of the present invention will be described with reference to FIGS.
FIG. 5 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 2 of the present invention. In FIG. 5, the motion vector detection apparatus includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, An integer precision motion vector detection unit 206 is included.

  The motion vector detection device according to the second embodiment is different from the motion vector detection device according to the first embodiment described above only in that the integer precision motion vector detection unit 206 has a feature. Therefore, the same components as those in the motion vector detection device according to the first embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

  The integer precision motion vector detection unit 206, which is a characteristic part of the motion vector detection device according to the second embodiment of the present invention, will be described below with reference to FIGS.

  FIG. 6 is a block diagram showing the configuration of the integer precision motion vector detection unit 206 according to Embodiment 2 of the present invention. In FIG. 6, the integer precision motion vector detection unit 206 includes an absolute difference error value calculation unit 21, a subtracter 22, a counter 23, an absolute difference error sum calculation unit 24, and a motion vector generation unit 25.

  The absolute difference error value calculation unit 21 calculates an absolute value of a difference between pixel values of the input pixel data of the processing target block and the pixel data of the reference block corresponding to the pixel data of the processing target block. Is output to the subtractor 22 as an absolute difference error value.

  The subtracter 22 sequentially subtracts the absolute difference error value from the threshold value held by the absolute difference error sum calculation unit 24 every time the absolute difference error value calculation unit 21 outputs an absolute difference error value. In the following description, a threshold before the subtraction by the subtracter 22 is started is referred to as an initial threshold.

  The counter 23 is an up counter having a specific count number, and increments the counter 23 every time the absolute difference error value from the threshold is subtracted by the subtracter 22. The counter 23 according to the second embodiment has a count from 0 to 127. This is because the integer address motion vector detection unit 206 uses the pixel data sampled in a checkered pattern for the 16 pixel × 16 pixel macroblock to be processed by the second address generation unit 105, and the absolute difference This is for calculating the error sum.

  The absolute difference error sum calculation unit 24 holds an initial threshold value, and holds a threshold value obtained by subtracting the absolute difference error value by the subtractor 22.

  Further, the absolute difference error sum calculation unit 24 sets the counter 23 to 0 when the threshold value becomes negative as a result of the subtraction by the subtracter 22, and the absolute difference error value calculation unit 21 and the subtractor 22. Is output to the second address generation unit 105 to exclude the position of the reference block from the candidates for the minimum position of the absolute difference error sum.

  On the other hand, if the value of the threshold is positive as a result of the subtraction by the subtracter 22, the subtraction by the subtractor 21 is repeated until the value of the counter 23 becomes "127". If the threshold value remains positive as a result of the subtraction by the subtractor 22 and the value of the counter 23 becomes “127”, the counter 23 is set to 0, and the initial threshold value held by the absolute difference error sum calculation unit 24 Then, the threshold value of the positive value is subtracted, and the subtracted value is set as a new initial threshold value held by the absolute difference error sum calculation unit 24.

  In addition, since the reference block is a candidate for a reference block with the smallest absolute difference error sum, data representing the position of the reference block and the position of the processing target block output from the second address generation unit 105 is used as a motion vector. The data is output to the generation unit 25. Whether or not the threshold value of the absolute difference error sum calculation unit 24 is negative is determined by whether the MSB (most significant bit) of the absolute difference error sum calculation unit 24 is “1” (negative). It can be determined by determining only whether or not the bit is.

  Further, when the absolute difference error sum calculation unit 24 receives a signal from the second address generation unit 105 indicating that the search process has been completed, the absolute threshold difference sum calculation unit 24 uses the initial threshold value held by the absolute difference error sum calculation unit 24 as an evaluation function. Output.

  When the motion vector generation unit 25 receives data representing the position of the reference block and the position of the processing target block from the absolute difference error sum calculation unit 24, the motion vector generation unit 25 generates a motion vector from the amount of change. The motion vector generation unit 25 generates a motion vector every time the output from the absolute difference error sum calculation unit 24 is received. If a previously generated motion vector exists, the motion vector is newly determined. Update the generated motion vector.

  In addition, when the motion vector generation unit 25 receives a signal indicating that the search processing is completed from the second address generation unit 105, the motion vector generation unit 25 uses the motion vector held by the motion vector generation unit 24 as the motion vector of the processing target block. Output as.

Next, the operation of the integer precision motion vector detection unit 206 will be described with reference to FIG.
FIG. 7 is a flowchart for explaining the operation of the integer precision motion vector detection unit 206 according to the second embodiment of the present invention. The operation of the integer precision motion vector detection unit 206 in the flowchart shown in FIG. 7 is for calculating the absolute difference error sum of a specific reference block with respect to the processing target block. This process is started by inputting corresponding pixel data.

  When the pixel data of the processing target block and the pixel data of the reference block corresponding to the pixel data of the processing target block are input, the absolute difference error value calculator 21 calculates the pixel data of the processing target block and the pixel data of the reference block. Is calculated and output to the subtractor 22 as an absolute difference error value (step S201).

  When the absolute difference error calculation unit 21 outputs an absolute difference error value, the subtracter 22 subtracts the absolute difference error value from the threshold value held by the absolute difference error sum calculation unit 24 (step S202).

  When the absolute difference error value is subtracted from the threshold value by the subtracter 22, the counter 23 increments the counter value (step S203).

  Next, the absolute difference error sum calculation unit 24 determines whether or not the threshold value is a negative value as a result of subtracting the absolute difference error value from the threshold value in step S202, and determines whether the threshold value MSB (most significant Whether or not (bit) is “1” is determined (step S204). Note that “1” for the MSB (most significant bit) indicates that the threshold is negative, and “0” for the MSB (most significant bit) indicates that the threshold is positive.

  At this time, if the MSB (most significant bit) of the threshold is “1”, the absolute difference error sum of the reference block cannot be the minimum absolute difference error sum, so the value of the counter 23 is set to “0”. (Step S209), the process for the reference block is not performed until the end, and the process ends.

  On the other hand, if the MSB (most significant bit) of the threshold is not “1”, whether or not the value of the counter 23 is “127”, that is, the absolute difference error values of all the pixels to be processed are from the threshold. It is determined whether or not the value has been subtracted (step S205).

  At this time, if the value of the counter 23 is not “127”, the process goes to step S201 to calculate absolute difference error values for the remaining pixels (step S201), and thereafter the same processing is performed (steps S202 to S205). ).

  On the other hand, when the value of the counter 23 is “127”, the absolute difference error sum calculation unit 24 determines that the reference block with respect to the processing target block is a reference block candidate with the smallest absolute difference error sum, and thus the second The data indicating the position of the reference block and the position of the processing target block output from the address generation unit 105 is output to the motion vector generation unit 25 (step S207).

  When receiving the data representing the position of the reference block and the position of the processing target block, the motion vector generation unit 25 generates a motion vector from the amount of change (step S207). At this time, if a previously generated motion vector exists, the previously generated motion vector is updated to a newly generated motion vector.

  The absolute difference error sum calculation unit 24 subtracts the value of the positive threshold when the value of the counter 23 becomes “127” from the initial threshold held by the absolute difference error sum calculation unit 24, and the value obtained by the subtraction. Is updated as a new initial threshold value held by the absolute difference error sum calculation unit 24 (step S208).

  Thereafter, the absolute difference error sum calculation unit 24 sets the value of the counter 23 to “0” (step S209), and ends the processing for the reference block.

  As described above, the integer precision motion vector detection unit 206 performs the reference block search process that minimizes the sum of absolute difference errors of the reference block with respect to the processing target block described with reference to FIG. The same process is performed for the reference blocks.

  When the search for the reference block in the reference area is completed for the processing target block, the second address generation unit 105 outputs a signal to the effect that the search process is completed to the integer precision motion vector detection unit 206. The difference error sum calculation unit 24 outputs the initial threshold value held by the absolute difference error sum calculation unit 24 as an evaluation function, and the motion vector generation unit 26 uses the motion vector held by the motion vector generation unit as the motion of the processing target block. Output as a vector.

  In this way, the integer precision motion vector detection unit 206 sequentially subtracts the absolute difference error value from the threshold value, and when the threshold value becomes negative, the reference block candidate that becomes the minimum absolute difference error sum with respect to the processing target block. By excluding from the above and canceling the calculation in the middle, the processing speed can be improved and the power consumption can be reduced.

  Further, whether or not the reference block being processed is a candidate for a reference block having the smallest absolute difference error sum with respect to the processing target block is determined based on whether or not the result of subtraction from the threshold value is negative. Therefore, the determination can be performed only by the value of the MSB (most significant bit) of the threshold value of the absolute difference error sum calculation unit 24, and the processing speed can be further improved and the power consumption can be reduced.

  In the second embodiment of the present invention, the motion vector detection apparatus detects the motion vector using the pixel data of the reference block and the pixel data of the processing target block sub-sampled in a checkered pattern. Without being limited thereto, the same effect can be obtained as long as the motion vector detection device detects the motion vector using the pixel data of the reference block and the pixel data of the processing target block.

(Embodiment 3)
Hereinafter, a motion vector detection apparatus according to Embodiment 3 of the present invention will be described with reference to FIGS.
FIG. 8 is a block diagram showing an example of a configuration of a motion vector detection device according to Embodiment 3 of the present invention. In FIG. 8, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, It consists of an integer precision motion vector detection unit 206 and a search history holding unit 301.

  It should be noted that the motion vector detection apparatus according to the third embodiment holds the history of reference blocks for which search has been completed, and the motion vector detection according to the second embodiment described above only in that double search of the same reference block is prevented. Different from the device. Therefore, the same components as those in the motion vector detection device according to the second embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

  The search history holding unit 301 holds a history of reference blocks searched by the integer precision motion vector detection unit 206. In the third embodiment, a description will be given of a case where there are a total of nine reference block histories for a position to be searched and eight peripheral positions.

  Hereinafter, an example of reference block search processing that minimizes the sum of absolute difference errors using the search history by the motion vector detection device according to Embodiment 3 of the present invention will be described with reference to FIGS. 9 and 10.

  FIG. 9 is an explanatory diagram for explaining an example of a search process performed by the second address generation unit 105 of the motion vector detection device according to the third embodiment. In FIG. 9, a solid line indicates a processing target block, and a dotted line indicates a reference block. This search process is a process of searching for a reference block having a minimum absolute difference error from the processing target block in the reference area.

  As shown in FIG. 9, as an example of the search process by the second address generation unit 105 of the motion vector detection device according to the third embodiment, first, the second address generation unit 105 first displays the target pixel in the processing target block. A pixel address is generated so as to search for five reference blocks that are shifted by one pixel vertically, right and left with respect to x, and the integer precision motion vector detection unit 206 performs a search for the reference block. Is called.

  The result is output to the second address generation unit 105, and the address generation unit 105 next minimizes the absolute difference error sum when any of the upper, lower, left, and right absolute difference error sums is minimized. The same processing is performed on the reference block shifted by one pixel vertically and horizontally with the reference block as the center. This process is repeated until the absolute difference error sum of the reference block located in the middle is minimized, and when the reference block located in the middle where the absolute difference error sum is minimized is detected, the reference block is processed. The block is detected as a reference block having the smallest absolute difference error sum for the block, and the search is terminated.

  FIG. 10 is a diagram illustrating a data structure in the search history holding unit 301, where a to l are codes indicating reference blocks. A range surrounded by a thick line 1 is a history range held by the search history holding unit 301, a solid line 2 indicates a reference block for which search has been completed, and a dotted line 3 indicates an unsearched reference block.

  For example, if the processing target block position is at the position h, first, as described above, a search is performed on five reference blocks for the processing target block. When the search is completed, the search history holding unit 301 sets a flag for the reference block e, the reference block g, the reference block h, the reference block i, and the reference block k that have been searched (FIG. 10A).

  Next, as a result of the search by the integer precision motion vector detection unit 206, for example, reference is made among the absolute difference error sums of the reference block e, the reference block g, the reference block h, the reference block i, and the reference block k with respect to the processing target block. When the absolute difference error of the block e is minimized, the search history holding unit 301 shifts all the history held by the search history holding unit 301 downward (FIG. 10B).

  As a result, the history held by the search history holding unit 301 is rewritten to be centered on the reference block e that is the central reference block in the next search (FIG. 10C).

  The second address generation unit 105 holds a search history when searching for the reference block b, the reference block d, the reference block e, the reference block f, and the reference block h with the reference block e as the center. The presence or absence of a flag in the unit 301 is determined, and pixel addresses of only the pixel data of the reference block b, the reference block d, and the reference block f that are not flagged are generated. Therefore, the integer precision motion vector detection unit 206 does not perform the search process of the reference block e and the reference block h that have already been searched twice.

  As described above, by providing the search history holding unit 301 and leaving the search history for the reference block that has already been searched, it is possible to prevent the double search for the same reference block. Thereby, the processing speed can be improved and the power consumption can be reduced.

  In the third embodiment, the search processing by the second address generation unit 105 has been described in which the search is advanced in the direction in which the absolute difference error sum is minimized among the reference blocks located in the upper, lower, left, and right directions. The present invention is not limited to this, and any search history holding unit 301 may be used as long as the second address generation unit performs the search process based on the search history held by the search history holding unit 301.

(Embodiment 4)
Hereinafter, a motion vector detection apparatus according to Embodiment 4 of the present invention will be described with reference to FIGS.

  FIG. 11 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 4 of the present invention. In FIG. 11, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, It consists of an integer precision motion vector detection unit 106, a transfer rule holding unit 401, and a transfer rule detection unit 402.

  Note that the motion vector detection device according to the fourth embodiment depends on where the reference area data output to the first storage unit 103 exists on the image data held by the data holding unit 101. The only difference from the motion vector detection apparatus according to the first embodiment is that the transfer rule of the reference area data transferred from the data holding unit 101 to the first storage unit 103 is changed. Therefore, the same components as those in the motion vector detection device according to the first embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

The transfer rule holding unit 401 holds the area determination code and the transfer rule as a pair.
The transfer rule detection unit 402 detects a transfer rule based on the area determination code output from the first address generation unit 102 and outputs the transfer rule to the first address generation unit 102 and the second address generation unit 105.

Here, the region determination code will be described with reference to FIG.
FIG. 12 is a diagram showing image data, and the region determination code according to the fourth embodiment of the present invention indicates to which position on the image data the reference region where the current motion vector search is performed corresponds. It is a code consisting of 4 bits. Each bit indicates the upper end, the lower end, the left end, and the right end from the upper order. For example, when the bit indicating the upper end is “1”, it indicates that the reference area is at the upper end of the image data. When the bit is “0”, it indicates that the reference area is not the upper end of the image data. The area determination code is held by the first address generation unit 102 that controls the address of the image data.

  Hereinafter, reference area data transfer processing by the first address generation unit 102 of the motion vector detection device according to the fourth embodiment of the present invention will be described with reference to FIGS. 12, 13, and 14. FIG.

  FIG. 13 is a diagram showing an example of an area determination code and a transfer rule stored in the transfer rule holding unit 401, and FIG. 14 shows an address of 48 × 48 pixels constituting reference area data. is there.

  When the area determination code is output from the first address generation unit 102 to the transfer rule detection unit 402, the transfer rule detection unit 402 displays the transfer rule corresponding to the area determination code held in the transfer rule holding unit 401. To detect.

  The transfer rule detected by the transfer rule detection unit 402 is output to the first address generation unit 102. The first address generation unit 102 generates a transfer address of reference area data to be output to the first storage unit 103 based on the detected transfer rule.

  For example, when the area determination code is “0000”, the first address generation unit has a horizontal length of 48 pixels, one pixel in the horizontal direction from the transfer start address position illustrated in FIG. That is, data for 48 pixels with pixel addresses from 0 to 47 is read. That is, as shown in FIG. 14, when the reading of the data for 48 pixels from 0 to 47 is completed, the data is shifted from the transfer start address position by 1 pixel in the vertical direction with respect to the image image, and from the left end. Data for 48 pixels in the horizontal direction with pixel addresses from 48 to 95 are similarly read out. This shift in the vertical direction is continued until the length of the rectangle in the vertical direction is 48, that is, the vertical length is moved by 48 pixels.

  For example, when the area determination code is “1010”, reading of pixel data is started from the position where the pixel address is 784, as shown in FIG. Data of 32 pixels having a length of 32, that is, pixel addresses of 784 to 815 are read. When the reading of data for 32 pixels with pixel addresses 784 to 815 is completed, the pixel address is shifted by 1 pixel in the vertical direction from the position of the pixel address 784 and the horizontal direction with pixel addresses 832 to 863 is shifted. Similarly, data for 32 pixels are read out. This shift in the vertical direction is continued until the length of the rectangle in the vertical direction is 32, that is, moves 32 pixels in the vertical direction.

  Note that the transfer process of the reference area data when the area determination code is “0001”, “0010”, “0100”, “0101”, “0110”, “1000”, “1001” is the area determination code described above. Since it is clear from the description of “0000” and “1010” and FIG. 13, the description is omitted.

  As described above, the first storage unit 102 changes the transfer rule of the reference area data output from the first address generation unit 102 to the first storage unit 103 according to the position of the reference area data on the image data. The amount of data transferred to the unit 103 can be minimized, the processing speed can be improved, and the power consumption can be reduced.

  In the motion vector detection device according to the fourth embodiment, the region determination code held by the first address generation unit 102 has been described. However, the present invention is not limited to this. For example, the user can specify the region determination code. It may be one that is directly input to the transfer rule detection unit 402.

  Moreover, although the motion vector detection apparatus according to Embodiment 4 of the present invention has shown the transfer rule as shown in FIG. 13 as an example of the transfer rule corresponding to the area determination code, the present invention is not limited to this, and the area determination code As long as the first address generation unit 102 generates the transfer address of the data based on the transfer rule corresponding to.

(Embodiment 5)
Hereinafter, a motion vector detection apparatus according to Embodiment 5 of the present invention will be described with reference to FIGS.

  FIG. 15 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 5 of the present invention. In FIG. 15, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, The integer precision motion vector detection unit 106, the transfer rule detection unit 402, the transfer rule holding unit 401, and the stored data control unit 501 are included.

  The motion vector detection device according to the fifth embodiment is a motion vector detection device in which the first storage unit 103 has a double buffer structure, and the stored data control unit 501 stores the data in the first storage unit 103. Only the control point is different from the motion vector detection apparatus according to the fourth embodiment described above. Therefore, the same components as those in the motion vector detection device according to the first embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

First, a double buffer structure, which is a memory structure in the first storage unit 103, will be described with reference to FIG.
FIG. 16 is an explanatory diagram illustrating the memory structure of the first storage unit 103 according to the fifth embodiment of the present invention. In FIG. 16, the first storage unit 103 includes two buffer memories, a memory X and a memory Y. Holding. In these two memories, for example, when data is read from and read from the memory Y, data is read from the other memory X. When both processes are completed, the memory X and the memory Y are exchanged, and the data written into the memory Y is read out at the same time as the data is read into the memory X. As a result, data can be input / output efficiently.

Next, the operation of the stored data control unit 501 according to the fifth embodiment of the present invention will be described.
In general, when obtaining motion vectors for a plurality of blocks constituting image data, the image data is sequentially shifted one block to the right from the block located at the upper left corner, and each block is processed. On the other hand, a motion vector is detected. Therefore, when the first storage unit 103 having the double buffer structure as described above is used to input / output reference region data that is a region to be searched, pixel data is transferred between the memory X and the memory Y. Therefore, every time the processing target block changes, duplicate data transfer is performed.

  Therefore, the stored data control unit 501 of the motion vector detection device according to the fifth embodiment of the present invention, among the data stored in the first storage unit 103 having a double buffer structure, as shown in FIG. The data stored in the memory X is transferred from the memory X to the memory Y. Thereafter, data necessary for the memory Y is read from the data holding unit 101. The same processing is performed in the same way in the transfer from the memory Y to the memory X.

  The storage data control unit 501 determines the range of data in the first storage unit 103 to be copied based on the area determination code output from the first address generation unit 102. The region determination code is the same as that described in the fourth embodiment, and a description thereof will be omitted.

A specific example of the operation of the storage data control unit 501 according to the fifth embodiment will be described below with reference to FIG.
FIG. 18 is a diagram illustrating a correspondence between the area determination code, data newly read by the first storage unit 103, and data copied in the first storage unit 103.

  For example, as shown in FIG. 18, when the area determination code for reading data into the memory Y is “0000”, the storage data control unit 501 stores the data in the memory X in the first storage unit 103. The right six blocks of data are copied to the left of the memory Y. As a result, the data storage unit 103 only needs to newly read only the remaining three blocks of data from the data holding unit 101.

  When the area determination code for reading data into the memory Y is “0001”, the storage data control unit 501 stores the data in the memory X in the first storage unit 103 as shown in FIG. The right six blocks of data are copied to the left of the memory Y. In addition, when the area determination code is “0001”, that is, when the rightmost bit is “1”, the data of the block located on the right side does not exist, and thus the data is not read.

  If the area determination code for reading data into the memory Y is “0010”, the stored data control unit 501 does not copy data as shown in FIG. As described above, when motion vectors are obtained for a plurality of blocks, motion vector detection is performed while sequentially shifting the processing target block to be processed to the right, so that the area determination code is “0010”. That is, when the second bit from the right is “1”, since the position of the processing target block for detecting the motion vector is located at the left end of the image data, the data used in the previous search can be used. This is because it cannot be done.

  The operation of the storage data control unit 501 in the case of other area determination codes is obvious from the area determination codes “0000”, “0001”, “0010” and FIG.

  As described above, the storage data control unit 501 controls the data stored in the first storage unit 103 by using the area determination code, thereby reducing the amount of data read from the data holding unit 101. As a result, the processing speed can be improved and the power consumption can be reduced.

  In the motion vector detection device according to the fifth embodiment, the area determination code is held by the first address generation unit 102. However, the present invention is not limited to this. For example, the user can set the area determination code. It may be one that is directly input to the stored data control unit 501.

  In addition, the motion vector detection device according to the fifth embodiment is sufficient if at least the memory structure of the first storage unit 103 has a double buffer structure, and the data holding unit 101 and the second storage unit. The memory structure 104 can be anything.

  In Embodiments 1 to 5 of the present invention, the motion vector detection unit that detects a motion vector is the integer accuracy motion vector detection unit 106 that detects an integer accuracy motion vector. However, the present invention is not limited to this. For example, the motion vector detection unit may determine a motion vector with a small number of precision, or may determine a motion vector with an integer precision and a motion vector with a small precision. Note that the small-precision motion vector detection will be described in Embodiment 6 described later.

(Embodiment 6)
Hereinafter, a motion vector detection device according to a sixth embodiment of the present invention will be described with reference to FIGS.
FIG. 19 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 6 of the present invention. In FIG. 19, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, An interpolation pixel generation unit 601, a minority precision motion vector detection unit 602, and a readout direction determination unit 603 are included.

  The interpolation pixel generation unit 601 is obtained by performing interpolation processing of pixel values at arbitrary positions between the pixels of the reference area data held in the first storage unit. In the sixth embodiment of the present invention, a description will be given by taking as an example a case where a half pixel that is a pixel value at the middle position between pixels is generated in order to simplify the description.

  FIG. 20 is an explanatory diagram for explaining the interpolation pixel generation unit 601 of the motion vector detection device according to the sixth embodiment of the present invention. The black circle is the pixel position, and the white circle is the half obtained by the interpolation pixel generation unit 601. Indicates the position of the pixel. For example, in order to generate the half pixel m, data of pixel values for four pixels around the pixel A, the pixel B, the pixel C, and the pixel D is necessary, and the interpolation pixel generation unit 601 generates a half pixel m = The pixel value of the half pixel m is obtained by calculating (pixel A + pixel B + pixel C + pixel D) / 4. In addition, in order to generate the half pixel n, pixel value data corresponding to two pixels of the pixel B and the pixel D existing in the vertical direction is necessary, and the interpolation pixel generation unit 601 has the half pixel n = (pixel B + The pixel value of half pixel n is determined by calculating pixel D) / 2. Furthermore, in order to generate the half pixel o, data of pixel values for two pixels of the pixel C and the pixel D existing in the horizontal direction is necessary, and the interpolation pixel generation unit 601 has the half pixel o = (pixel C + The pixel value of the half pixel o is obtained by calculating the pixel D) / 2.

  The small-precision motion vector detection unit 602 outputs half-pixel data generated by the interpolation pixel generation unit 601 using the pixel data of the reference block output from the first storage unit 103, and is output from the second storage unit 104. Using the pixel data of the processing target block, the absolute difference error sum between the processing target block and the reference block with half pixel accuracy shifted by 1/2 pixel from the reference block is calculated, and the absolute difference error sum is minimized. A motion vector is detected by searching for a reference block with half-pixel accuracy. The search for the half-pixel precision reference block by the small-precision motion vector detection apparatus 602 according to the sixth embodiment proceeds in the direction in which the minimum half-pixel precision reference block exists around the processing target block position. The search direction is output to the read direction determination unit 603 as a search position vector.

  The readout direction determination unit 603 determines the readout direction of the pixel data of the reference area output from the first storage unit 103 according to the position of the half pixel data that is generated by the interpolation pixel generation unit 601. Note that the position of the half-pixel data generated by the interpolation pixel generation unit 601 is determined based on the search position vector output from the above-described minority precision motion vector detection unit 602.

Hereinafter, half-pel generation processing of the motion vector detection device according to the sixth embodiment of the present invention will be described with reference to FIGS. 21 and 22.
21 and 22 are explanatory diagrams for explaining generation of half-pixel data by the interpolation pixel generation unit 601 according to the sixth embodiment of the present invention. 21 and 22, the interpolation pixel generation unit 601 includes an interpolation processing unit 81 and a shift register group 82 as illustrated.

  The interpolation processing unit 81 performs an interpolation process on the pixel data held by the shift register 82, and calculates a pixel value at a position between the pixels. The shift register group 82 includes 21 shift registers, and shifts read pixel data every time pixel data is input. In the sixth embodiment, a pixel value with half-pixel accuracy, which is the center position between pixels, is obtained.

  When the interpolated pixel generation unit 601 creates a reference block with half-pixel accuracy shifted by 1/2 pixel in any of the left, right, upper right, lower right, upper left, and lower left directions with respect to the reference block, The reading direction determination unit 603 instructs the second address generation unit 105 to read the reference area data stored in the first storage unit 103 in the horizontal direction. As shown in FIG. 21, the second address generation unit 105 performs 18 times of 18 pixels of data located in the horizontal direction from the pixel data of the reference area stored in the first storage unit 103, 18 times in the vertical direction. read out.

The read pixel data enters the shift register group 82 in the interpolation pixel generation unit 601, and interpolation processing is performed by the interpolation processing unit 81.
In addition, the shift register group 82 of the interpolation pixel generation unit 601 generates pixel data that is shifted by half a pixel in the diagonal direction by holding at least (the number of data to be read in the horizontal direction + 2) shift registers. However, it is possible to perform interpolation processing of pixel data shifted by half a pixel in an oblique direction without reading data again, thereby improving processing speed and reducing power consumption.

On the other hand, when the interpolated pixel generation unit 601 creates a reference block with half-pixel accuracy shifted by 1/2 pixel in any of the upper, lower, upper right, lower right, upper left, and lower left directions with respect to the reference block. The read direction determination unit 603 instructs the second address generation unit 105 to read the reference area data stored in the first storage unit 103 in the vertical direction. As shown in FIG. 22, the second address generation unit 105 generates 18 pixels of data located in the vertical direction 18 times in the horizontal direction from the pixel data in the reference area stored in the first storage unit 103. read out.
The read pixel data enters the shift register group 82 in the interpolation pixel generation unit 601, and interpolation processing is performed by the interpolation processing unit 81.

  Thus, by generating pixel data that is shifted by half a pixel in the vertical direction using pixel data that has been read in the vertical direction, the amount of calculation for performing interpolation processing of the pixel data that is shifted by a half pixel in the vertical direction is reduced. Therefore, the processing speed can be improved and the power consumption can be reduced.

  Further, the shift register group 82 of the interpolation pixel generation unit 601 generates pixel data that is shifted by half a pixel in the diagonal direction by holding at least (the number of data to be read in the vertical direction + 2) shift registers. However, it is possible to perform interpolation processing of pixel data shifted by half a pixel in an oblique direction without reading data again, thereby improving processing speed and reducing power consumption.

  That is, by changing the data reading direction for the image image according to the interpolation position of the pixel, the half pixels n and o that are shifted by half a pixel in the vertical direction or the horizontal direction without changing the configuration of the interpolation pixel generation unit 601 can be processed at high speed. Moreover, it can be generated at the same timing, and the processing speed can be improved and the power consumption can be reduced. Further, when generating a half pixel m that is shifted by half a pixel in an oblique direction, at least the shift register group 82 includes 20 shift registers, so that half pixel data can be generated at high speed and at the same timing. Thus, it is possible to reduce the amount of data transfer and the processing speed and reduce the power consumption.

  21 and 22, the interpolation pixel generation unit 601 shows the shift register group 82 having 21 shift registers. However, the present invention is not limited to this. For example, pixel data shifted by half a pixel in the vertical and horizontal directions is used. When generating, it is sufficient that it holds at least two shift registers, and in order to generate pixel data shifted by half a pixel in an oblique direction, it holds at least 20 shift registers. If it is.

(Embodiment 7)
Hereinafter, a motion vector detection apparatus according to Embodiment 7 of the present invention will be described with reference to FIGS. 23 and 24. FIG.
FIG. 23 is a block diagram showing an example of a configuration of a motion vector detection device according to Embodiment 7 of the present invention. In FIG. 23, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, An interpolation pixel generation unit 601, a reading direction determination unit 603, and a minority precision motion vector detection unit 701 are included.

  Note that the motion vector detection device according to the seventh embodiment includes two absolute difference error sum calculation units in which the decimal precision motion vector detection unit 702 calculates the absolute difference error sum, and the two small numbers corresponding to the processing target block. The only difference from the motion vector detection apparatus according to the sixth embodiment is that block matching of the accuracy reference block is performed simultaneously. Therefore, the same components as those in the motion vector detection device according to the sixth embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

Hereinafter, the small-precision motion vector detection processing by the motion vector detection device according to the seventh embodiment of the present invention will be described.
As described in the first embodiment, the second address generation unit 105 generates pixel data addresses that are subsampled in a checkered pattern with respect to the processing target block and the reference block.

  However, in the interpolated pixel generation unit 601 according to the seventh embodiment, a half-pixel precision reference block corresponding to the pixel data of the processing target block subsampled into a checkered pattern output from the second storage unit 104 is generated. In this case, as described in the sixth embodiment, it is necessary to read all pixel data constituting the reference block and pixel data of one pixel around the pixel data.

  Therefore, the interpolation pixel generation unit 601 uses the entire pixel data constituting the reference block read from the first storage unit 103 and the data for one pixel around the pixel block, and the processing target block sampled in a checkerboard pattern. 2 corresponding half-pixel precision processing blocks, for example, upper and lower, left and right, upper right and lower right, upper left and lower left, upper left and lower left, upper right and upper left, Data of the minority precision pixels of the two half-pixel precision reference blocks that are shifted by half a pixel respectively to the lower right and the lower left are generated and output to the minority motion vector detection unit 701.

  At this time, for example, the pixel data of the minority reference block shifted from the integer precision reference block by half a pixel up and down, for example, the pixel data of the upper end column and the lower end column of the pixel column constituting the minority precision reference block. They are different from each other, and the same values can be used for other overlapping portions. Therefore, the data processing amount of the interpolation pixel generation unit 601 can be reduced. This is two half-pixel accuracy, shifted left and right, diagonally right and diagonally up and down, diagonally diagonally up and down diagonally left, diagonally right and diagonally up and diagonally down diagonally. The same applies when generating data of the minority precision pixels of the reference block.

Hereinafter, the configuration of the minority precision motion vector detection unit 701 will be described with reference to FIG.
FIG. 24 is a block diagram illustrating the configuration of the small-precision motion vector detection unit 701 of the motion vector detection device according to the seventh embodiment of the present invention. In the figure, the minority precision motion vector detection unit 701 includes a first absolute difference error sum calculation unit 71, a second absolute difference error sum calculation unit 72, and a motion vector generation unit 73.

  Both the first absolute difference error sum calculation unit 71 and the second absolute difference error sum calculation unit 72 correspond to the pixel data of the processing target block subsampled in a checkered pattern and the pixel data of the processing target block. The calculation unit calculates an absolute difference error sum that is a sum of differences between absolute values of pixel data constituting the processing target block and the half-pixel accuracy reference block, using the pixel data of the half-pixel accuracy reference block.

  The motion vector generation unit 73 is calculated by the first absolute difference error sum calculation unit 71 and the second absolute difference error sum calculation unit 72 and has a small precision with which the absolute difference error sum is reduced with respect to the processing target block. A motion vector is generated from the position of the reference block.

  For example, when block matching is performed by the minority precision motion vector detection unit 701 on two half pixel precision reference blocks shifted by half a pixel to the left and right with respect to the processing target block subsampled in a checkered pattern, interpolation is performed. The pixel data of the processing target block with half-pixel accuracy shifted by a half pixel to the left from the pixel generation unit 601 and the pixel data of the processing target block are input to the first absolute difference error calculating unit 71 and shifted by a half pixel to the right The pixel data of the processing target block with half pixel accuracy and the pixel data of the processing target block are input to the second absolute difference error sum calculation unit 71.

  In the first absolute difference error sum calculation unit 71 and the second absolute difference error sum calculation unit 72, a small-precision accuracy shifted by half a pixel on the right side and the left side with respect to the processing target block subsampled in a checkered pattern. The absolute difference error sum of the reference block is calculated and output to the motion vector generation unit 73.

  The motion vector generation unit 73 compares the absolute difference error sums output from the first absolute difference error sum calculation unit 71 and the second absolute difference error sum calculation unit 72, and the one with the smaller absolute difference error sum value. A motion vector for a reference block with a small number of precision is detected.

  In this way, a reference block stored in the first storage unit 103 is configured to generate a small-precision interpolation pixel used for detection of a small-precision motion vector with respect to the processing target block subjected to sub-sampling. Even when all the pixel data to be read are read out, the small-precision motion vector detection unit 701 is equipped with two absolute difference error sum calculation units, and the interpolation pixel generation unit 601 generates two half-pixel accuracy reference blocks simultaneously. 2 and a half shifted by half a pixel, such as up and down, left and right, right up and down right, left up and down left, right up and left up, right down and left down The calculation processing of the absolute difference error sum of the pixel accuracy processing target block can be performed simultaneously, the processing amount of the interpolation pixel generation unit 601 can be reduced, and the amount of data read from the first storage unit 103 can be reduced Rukoto can, it is possible to improve the processing speed, leading to reduction in power consumption.

  In the motion vector detection device according to the seventh embodiment of the present invention, the second address generation unit 105 generates the pixel address so that the processing target block has a checkerboard pattern. For example, the second address generation unit 105 is not limited to the pixel address method generated by the second address generation unit 105, such as one in which the second address generation unit 105 generates addresses of all the pixels constituting the processing target block.

(Embodiment 8)
Hereinafter, a motion vector detection apparatus according to an eighth embodiment of the present invention will be described with reference to FIGS.
FIG. 25 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 8 of the present invention. In FIG. 25, the motion vector detection apparatus includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, It includes an integer precision motion vector detection unit 106, a transfer rule holding unit 401, a transfer rule detection unit 402, a search availability condition holding unit 801, and a search availability condition detection unit 802.

  Note that the motion vector detection apparatus according to the eighth embodiment uses the reference block output from the first storage unit 103 based on the search availability condition detected by the second address generation unit 105 by the search availability condition detection unit 802. Only in the point that the address of the pixel data is generated is different from the motion vector detection apparatus according to the fourth embodiment described above. Therefore, the same components as those in the motion vector detection device according to the fourth embodiment described above are denoted by the same reference numerals, and description thereof is omitted.

The search availability condition holding unit 801 holds a pair of an area determination code and a search availability condition, which is a condition for determining whether or not the position can be searched.
The search enable / disable condition detection unit 802 detects the search enable / disable condition from the search enable / disable condition holding unit 801 based on the area determination code output by the first address generation unit 102, and outputs the search enable / disable condition to the second address generation unit 105.
The region determination code is the same as the region determination code described with reference to FIG. 12 in the fourth embodiment of the present invention, and a description thereof will be omitted.

  Hereinafter, reference address pixel pixel address generation processing by the second address generation unit 105 of the motion vector detection device according to the eighth embodiment of the present invention will be described with reference to FIGS. 25 and 26. FIG.

  FIG. 26A is a diagram illustrating an example of the area determination code and the search enable / disable condition stored in the search enable / disable condition holding unit 801. As shown in FIG. 26 (b), whether the search is possible when the horizontal axis with the right direction as positive is the x axis, the vertical axis with the down direction is positive is the y axis, and the position of the pixel p is the origin. The search availability condition held by the condition holding unit 801 is shown. Note that the area availability condition shown in FIG. 26A indicates an area where a search can be performed.

  When the area determination code is output from the first address generation unit 102 to the search availability condition detection unit 802, the search availability condition detection unit 802 searches the availability condition search unit 802 for the search availability condition corresponding to the area determination code. Is output to the second address generation unit 105.

  The second address generation unit 105 generates the address of the pixel data output from the first storage unit 103 based on the search availability condition output from the search availability condition detection unit 802.

Next, a specific example of the motion vector detection device according to the eighth embodiment of the present invention will be described.
For example, when the region determination code output from the first address generation unit 102 is “0000”, the search availability condition detection unit 802 is a search availability condition from the search availability condition holding unit 801 (−16 ≦ x ≦ 15, −16 ≦ y ≦ 15) is detected and output to the second address generator 105.

  The second address generation unit 105 uses the search enable / disable condition detected by the search enable / disable condition detection unit 802, and sets the pixel address so that it falls within the range of −16 ≦ x ≦ 15 and −16 ≦ y ≦ 15. Generate.

  The case where the area determination code is “0001” “0010” “0100” “0101” “0110” “1000” “1001” “1010” is the same as the case where the area determination code is “0000”. Therefore, explanation is omitted.

  As described above, the search availability condition detection unit 802 detects the search availability condition using the region determination code from the search availability condition holding unit 801, and the second address generation unit 105 performs the first storage based on the search availability condition. By generating the address of the pixel data output from the unit 103, it is possible to accurately generate the pixel address of an area where effective search can be performed, and to perform processing at high speed without performing useless search processing Can do.

(Embodiment 9)
Hereinafter, a motion vector detection apparatus according to Embodiment 9 of the present invention will be described with reference to FIG.
FIG. 27 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 9 of the present invention. In FIG. 27, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 105, Integer accuracy motion vector detection unit 106, transfer rule holding unit 401, transfer rule detection unit 402, interpolation pixel generation unit 601, minority accuracy motion vector detection unit 602, readout direction determination unit 603, and search availability condition holding Section 801 and search enable / disable condition detection section 802.

  In the ninth embodiment of the present invention, the search availability condition holding unit 801 and the search availability condition detection unit 802 used when detecting the integer precision motion vector of the motion vector detection apparatus according to the eighth embodiment are provided with a decimal precision motion vector. When the integer precision motion vector is detected and the decimal precision motion vector is detected, the search availability condition detection unit 802 detects the search availability condition, and the second address generation unit 105 performs the search. The pixel address is generated using the search availability condition detected by the availability condition detection unit 802. Each component is the same as that of the motion vector detection device described in the sixth and eighth embodiments, and a description thereof will be omitted.

  As described above, the search availability condition holding unit 801 and the search availability condition detection unit 802 used for integer precision motion vector detection are also used for the decimal precision motion vector detection, thereby reducing the circuit scale and reducing the manufacturing cost. Leads to.

(Embodiment 10)
Hereinafter, a motion vector detection apparatus according to Embodiment 10 of the present invention will be described with reference to FIG.
FIG. 28 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 10 of the present invention. In FIG. 28, the motion vector detection device includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 905, Integer accuracy motion vector detection unit 106, transfer rule holding unit 401, transfer rule detection unit 402, interpolation pixel generation unit 601, minority accuracy motion vector detection unit 602, readout direction determination unit 603, and search availability condition holding Section 801 and search enable / disable condition detection section 802.

  Note that the motion vector detection apparatus according to the tenth embodiment of the present invention performs an integer precision motion vector search using the search availability condition detected by the search availability condition detection unit 802 when detecting an integer precision motion vector. Detect motion vectors with accuracy. Thereafter, when detecting a small-precision motion vector, a small-precision motion vector is detected using the result of the search availability at the time of detecting the integer-precision motion vector. Each component is the same as that of the motion vector detection device described in the sixth and eighth embodiments, and a description thereof will be omitted.

  The second address generation unit 905 generates an address of pixel data used for integer precision motion vector detection based on the search availability condition detected by the search availability condition detection unit 802, and performs a search when detecting the integer precision motion vector. Search enable / disable data, which is information indicating whether or not it has been performed, is held for each reference block position. Further, in the case of detecting a small-precision motion vector after detecting an integer-precision motion vector, the search availability at the time of detecting a small-precision motion vector is determined using the search availability data.

The operation of the motion vector detection device according to the tenth embodiment of the present invention will be described below with reference to FIG.
FIG. 29 shows an example of a motion vector search at the time of detecting an integer precision motion vector, and a reference block at a position shifted by one pixel vertically and horizontally from a reference block at a specific position (a block located at the center) (FIG. It is a figure which shows the position of (it represents with a dotted line). It should be noted that the integer precision motion vector detection of the motion vector detection device according to the tenth embodiment of the present invention has the smallest absolute difference error sum of the reference block located at the center among the reference blocks in the positional relationship as shown in FIG. In this case, the reference block position is set as a target position for integer vector motion vector detection.

  When detecting an integer precision motion vector, the second address generation unit 905 holds whether or not a reference block located at the top, bottom, left, and right of this motion vector detection target position can be searched as 4-bit search enable / disable data. Specifically, for example, as shown in FIG. 29, it is assumed that the position of the top, bottom, left, and right from the left end bit is represented, and the bit of the reference block position where the search could not be performed is “1”, and the search could be performed. The bit at the reference block position is held as “0”.

  In this way, after performing an integer-precision motion vector search, in order to obtain a more accurate motion vector, even when performing a decimal-precision motion vector search within the range of the area searched when detecting an integer-precision motion vector, The second address generation unit 905 can generate a pixel address output from the first storage unit 103 based on the search enable / disable data consisting of the 4 bits. Thereby, it is possible to determine whether or not the search is possible without newly detecting the search availability condition when detecting the motion vector with a small number of precision, and the processing speed can be improved. Further, since the search result at the time of detecting an integer precision motion vector is also used at the time of detecting a decimal precision motion vector, the circuit scale can be reduced and the manufacturing cost can be reduced.

(Embodiment 11)
Hereinafter, a motion vector detection apparatus according to Embodiment 11 of the present invention will be described with reference to FIG.
FIG. 30 is a block diagram showing an example of the configuration of a motion vector detection device according to Embodiment 11 of the present invention and its periphery. In FIG. 30, the motion vector detection apparatus includes a data holding unit 101, a first address generation unit 102, a first storage unit 103, a second storage unit 104, a second address generation unit 1005, It consists of an integer precision motion vector detection unit 106 and a search execution possibility determination unit 1001. It should be noted that among the constituent elements constituting the motion vector detection device described in the first embodiment, the same reference numerals are given and the description thereof is omitted.

  The search execution feasibility determining unit 1001 holds the motion vector detected by the previous motion vector detection and compares the absolute difference error calculated by the integer precision motion vector detection unit 106 with a specific threshold value to execute the search. Determine whether or not.

  The second address generation unit 1005 generates addresses of data output from the first storage unit 103 and the second storage unit 104, and also outputs pixel data at the motion vector detection position from the first storage unit 103. A through path for passing through the integer precision motion vector detection unit 106 and outputting it to the outside is provided.

Next, the operation of the motion vector detection device according to the eleventh embodiment of the present invention will be described.
The second address generation unit 1005 uses the motion vector detected in the previous search held by the search execution availability determination unit 1001 and the pixel data output from the first storage unit 103 and the second storage unit 104. An address of pixel data to be output is generated.

  That is, since the motion vector that is generally detected hardly changes in the vicinity of the processing target block that has been searched, the second address generation unit 1005 first determines the position ( For example, the address of the pixel data is generated so as to perform block matching with respect to a reference block at the same position as the motion vector of the block searched immediately before.

  The integer precision motion vector detection unit 106 includes the pixel data of the reference block read from the first storage unit 103 by the second address generation unit 105 and the processing target block read from the second storage unit 104. Using the pixel data as an input, the absolute difference error sum between the processing target block and the reference block is calculated and output to the search execution feasibility determining unit 1001.

In this search execution feasibility determination unit 1001, it is determined whether or not the absolute difference error calculated by the integer precision motion vector detection unit 106 is equal to or less than a specific threshold value. The position of the reference block determined by the unit 1005 using the motion vector detected in the previous search is determined as the motion vector detection position, and the motion vector search process ends.
On the other hand, if the absolute difference error is greater than or equal to a specific threshold value, it is output to the second address generation unit 1005 so as to execute a motion vector search process for another reference block.

  In this way, block matching is performed on a reference block at a position where a motion vector is likely to be detected (for example, the same position as the motion vector of the block searched immediately before), and the matching result (absolute difference error sum) Thus, it is possible to improve processing speed and reduce power consumption by determining whether or not subsequent search execution is possible.

  At this time, if the motion vector search is not executed as a result of the above determination, the pixel data used in the motion vector detection process and thereafter is used as the previous reference block stored in the first storage unit 103. The pixel data indicated by the motion vector position detected in the search is required.

  Therefore, in the motion vector detection device according to the eleventh embodiment, the search execution availability determination unit 1001 causes the second address generation unit 1005 to obtain pixel data at the motion vector position detected in the previous search as an integer precision motion. An instruction to give a through path, which is a path for passing through the vector detection unit 106 and outputting to the outside, is output, and the second address generation unit 1005 gives a through path to the first storage unit and is stored. The pixel data of the reference block is output through the integer precision motion vector detection unit 106 to the external device.

  As described above, even when the motion vector detected in the previous motion vector search is used, the second address generation unit 1005 provides a through path to detect the previous motion vector used after the motion vector detection process. Pixel data indicated by the determined motion vector position can be output to an external device.

  In the motion vector detection device according to the eleventh embodiment of the present invention, the motion vector detecting means has been described as having only the integer precision motion vector detection unit 106 for detecting which motion vector is an integer. Without limitation, an interpolation pixel generation unit that generates pixel data with minority precision for detecting a motion vector with minority precision, data output from the interpolation pixel generation unit, and output from the second storage unit It may further comprise a minority precision motion vector detection unit that detects a minority precision motion vector from the data.

(Embodiment 12)
In addition to the first to eleventh embodiments, the motion vector detection device of the present invention may be a combination of at least two of the first to eleventh embodiments. The effects similar to those of the first to eleventh embodiments can be obtained.

  The motion vector detection apparatus according to the present invention is very useful because the motion vector detection process can be speeded up and the power consumption can be reduced.

It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 1 of this invention. It is a block diagram which shows the structure of the 2nd address generation part by Embodiment 1 of this invention. It is a figure which shows an example of the macroblock which the 2nd address generation part by Embodiment 1 of this invention performs a subsampling. It is a flowchart for demonstrating the address generation process of the pixel data of the 2nd address generation part by Embodiment 1 of this invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 2 of this invention. It is a block diagram which shows the structure of the motion vector detection part by Embodiment 2 of this invention. It is a flowchart for demonstrating operation | movement of the integer precision motion vector detection part by Embodiment 2 of this invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 3 of this invention. It is explanatory drawing for demonstrating an example of the motion vector search process by the motion vector detection apparatus in Embodiment 3 of this invention. It is a figure which shows the data structure in the search log | history holding part by Embodiment 3 of this invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 4 of this invention. It is explanatory drawing for demonstrating an area | region determination code. It is a figure which shows an example of the area | region determination code and the transfer rule stored in the transfer rule holding | maintenance part by Embodiment 4 of this invention. It is a figure which shows an example of the pixel address of reference area data. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 5 of this invention. It is explanatory drawing for demonstrating the memory structure of the 1st memory | storage part by Embodiment 5 of this invention. It is explanatory drawing for demonstrating the data processing in the 1st memory | storage part by Embodiment 5 of this invention. It is a figure which shows a response | compatibility with the area | region determination code, the data newly read by the 1st memory | storage part, and the data copied in a 1st memory | storage part. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 6 of this invention. It is explanatory drawing for demonstrating the interpolation pixel production | generation part by Embodiment 6 of this invention. It is explanatory drawing for demonstrating the production | generation of the half pixel data by the interpolation pixel production | generation part by Embodiment 6 of this invention. It is explanatory drawing for demonstrating the production | generation of the half pixel data by the interpolation pixel production | generation part by Embodiment 6 of this invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 7 of this invention. FIG. 10 is a block diagram illustrating the configuration of a minority precision motion vector detection unit according to Embodiment 7 of the present invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 8 of this invention. (A) It is a figure which shows an example of the area | region determination code stored in the search availability condition holding | maintenance part by Embodiment 8 of this invention, and search availability conditions. (B) It is explanatory drawing for demonstrating the search availability condition shown to Fig.26 (a). It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 9 of this invention. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 10 of this invention. It is a figure which shows an example of the motion vector search at the time of integer precision motion vector detection. It is a block diagram which shows the structure of the motion vector detection apparatus by Embodiment 11 of this invention. It is explanatory drawing for demonstrating the conventional general motion vector detection process.

Explanation of symbols

101 Data storage unit 102 First address generation unit 103 First storage unit 104 Second storage unit 105, 905, 1005 Second address generation unit 106, 206 Integer precision motion vector detection unit 301 Search history storage unit 401 Transfer Rule holding unit 402 Transfer rule detection unit 501 Storage data control unit 601 Interpolated pixel generation unit 602, 701 Minority precision motion vector detection unit 603 Reading direction determination unit 1001 Search execution enable / disable determination unit 11 Column counter 12 Row counter 13 Address storage unit 14 Addition 21 Absolute difference error value calculation unit 22 Subtractor 23 Counter 24 Absolute difference error sum calculation units 25 and 73 Motion vector detection unit 71 First absolute difference error sum calculation unit 72 Second absolute difference error sum calculation unit 81 Interpolation processing Part 82 Shift register group

Claims (38)

Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
A motion vector detection unit that detects a motion vector using the data output from the first storage unit and the data output from the second storage unit;
The motion vector detection unit sequentially calculates the absolute value of the difference between the pixel data of the reference block pixel data and the pixel data of the processing target block, and subtracts the calculated absolute value of the difference from the threshold value. A motion vector detection apparatus, wherein, when the result becomes negative, calculation of an absolute difference error sum for a reference block being processed is stopped. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
A motion vector detection unit for detecting a motion vector using the data output from the first storage unit and the data output from the second storage unit;
A search history holding unit that holds a search history of reference blocks in reference area data that has already been searched by the motion vector detection unit;
The motion vector detection device skips execution of a search for a reference block in which a search history held by the search history holding unit exists. The motion vector detection device according to claim 2,
The search history holding unit holds, as a search history, the search target position and the presence / absence of a search for the reference blocks at eight peripheral positions. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
A motion vector detection unit for detecting a motion vector using the data output from the first storage unit and the data output from the second storage unit;
A transfer rule holding unit that holds a transfer rule that is a rule for generating an address of reference area data that the first address generation unit outputs to the first storage unit;
A transfer rule detection unit that detects the transfer rule from the transfer rule holding unit and outputs the transfer rule to the first address generation unit based on an area determination code indicating an effective area of image data;
The first address generation unit generates an address of reference area data to be output to the first storage unit based on the transfer rule output from the transfer rule detection unit. apparatus. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A memory having a double buffer configuration for simultaneously reading and reading data, wherein the first storage unit holds reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
A motion vector detection unit for detecting a motion vector from the data output from the first storage unit and the data output from the second storage unit;
A storage data control unit that controls reading and reading of reference area data from and to the first storage unit;
The storage data control unit is effective for one buffer memory from which data is read out of the two buffer memories held by the first storage unit from another buffer memory that has been read. A motion vector detecting apparatus, wherein a part of data is copied and a data amount newly written to the first storage unit is reduced. In the motion vector detection device according to any one of claims 1 to 5,
The motion vector detection unit is an integer accuracy motion vector detection unit that detects an integer accuracy motion vector from the data output from the first storage unit and the data output from the second storage unit. A motion vector detection device. In the motion vector detection device according to any one of claims 1 to 5,
The motion vector detection unit receives output data from the first storage unit as an input, an interpolation pixel generation unit that generates pixel data with a small number of precision,
A motion vector detection apparatus comprising: a minority precision motion vector detection unit for detecting a minority motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit. In the motion vector detection device according to any one of claims 1 to 5,
The motion vector detection unit detects an integer precision motion vector from the data output from the first storage unit and the data output from the second storage unit;
An interpolation pixel generation unit that receives the output data from the first storage unit and generates pixel data with a small number of precision;
A motion vector detection apparatus comprising: a minority precision motion vector detection unit for detecting a minority motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
An interpolation pixel generation unit that receives the output data from the first storage unit and generates data of a small number of precision pixels;
A minority-precision motion vector detection unit that detects a minority-precision motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit;
A reading direction determining unit that determines a reading direction of reference area data held by the first storage unit according to a position of data of the minority precision pixels generated by the interpolation pixel generating unit and outputs the reference direction data to the second address generating unit; Comprising
The motion vector detection device, wherein the second address generation unit generates an address of pixel data output from the first storage unit based on the readout direction determined by the readout direction determination unit. The motion vector detection device according to claim 9,
The motion vector detection device, wherein the read direction determination unit determines to read data horizontally or vertically with respect to the image data stored in the first storage unit. In the motion vector detection device according to claim 9 or 10,
The interpolation pixel generation unit includes at least a shift register group including two shift registers,
A motion vector detection device comprising: an interpolation processing unit that performs interpolation processing using pixel data held by the shift register group. In the motion vector detection device according to claim 9 or 10,
The interpolation pixel generation unit includes at least a shift register group including a number of pixels to be read in a specific direction + 2 shift registers,
A motion vector detection device comprising: an interpolation processing unit that performs interpolation processing using pixel data held by the shift register group. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
An interpolation pixel generation unit that receives the output data from the first storage unit and generates data of a small number of precision pixels;
A minority precision motion vector detection unit for detecting a minority precision motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit;
The minority-precision motion vector detection unit includes, from the interpolation pixel generation unit, upper and lower, left and right, upper right and lower right, upper left and lower left, upper left and lower right, and upper right. When a minority precision reference block located at either the upper left corner or the lower right corner or the lower left diagonal position is output, detection of the minority precision motion vector for each of the output minority reference blocks is performed. A motion vector detection apparatus which is executed in parallel. The motion vector detection device according to any one of claims 9 to 13,
The motion vector further includes an integer precision motion vector detection unit that detects an integer precision motion vector from the data output from the first storage unit and the data output from the second storage unit. Detection device. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
An integer precision motion vector detection unit for detecting an integer precision motion vector from the data output from the first storage unit and the data output from the second storage unit;
A search enable / disable condition holding unit that holds a search enable / disable condition that is a condition for determining whether or not the position can be searched;
A search enable / disable condition detection unit that detects a search enable / disable condition from the search enable / disable condition holding unit based on an area determination code indicating an effective area of the reference area data, and outputs to the second address generation unit;
The motion vector detection device, wherein the second address generation unit determines whether or not a search is possible by using the search availability condition detected by the search availability condition detection unit when detecting an integer precision motion vector. The motion vector detection device according to claim 15, wherein
Furthermore, an interpolation pixel generation unit that receives the output data from the first storage means and generates pixel data with a small number of precision,
A minority-precision motion vector detection unit for detecting a minority-precision motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit;
The second address generation unit uses the search availability condition holding unit and the search availability condition detection unit in combination when detecting an integer precision motion vector and a decimal precision motion vector. Vector detection device. The motion vector detection device according to claim 16, wherein
Furthermore, an interpolation pixel generation unit that receives the output data from the first storage means and generates pixel data with a small number of precision,
A minority-precision motion vector detection unit for detecting a minority-precision motion vector from the data output from the interpolation pixel generation unit and the data output from the second storage unit;
The second address generation unit uses the search availability condition holding unit and the search availability condition detection unit when detecting an integer precision motion vector to determine the availability of the search, and uses the search availability result to determine a small number. A motion vector detection apparatus for determining whether or not a search is possible when detecting an accurate motion vector. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection device that detects a motion vector by performing
A first address generation unit that generates an address of the processing target block and the reference area data;
A first storage unit for holding reference area data instructed by the first address generation unit;
A second storage unit that holds data of a processing target block instructed by the first address generation unit;
A second address generation unit that generates an address of data output from the first storage unit and the second storage unit;
An integer precision motion vector detection unit for detecting an integer precision motion vector using the data output from the first storage unit and the data output from the second storage unit;
The motion vector detected by the previous motion vector search is retained, and the absolute difference error value calculated by the integer precision motion vector detection unit is compared with a specific threshold value to determine whether the motion vector search can be executed. A search execution feasibility determination unit for determining,
The second address generation means outputs from the first storage unit and the second storage unit using the motion vector detected in the previous motion vector search held by the search execution possibility determination unit. A motion vector detecting device for generating an address of data. The motion vector detection device according to claim 18, wherein
The second address generation unit further outputs the data output from the first storage means as a calculation result of the present apparatus. In the motion vector detection device according to claim 18 or 19,
Furthermore, an interpolation pixel generation unit that receives the output data from the first storage unit and generates pixel data with a small number of precision,
A motion vector detection device comprising: a minority motion vector detection unit for detecting a motion vector with a minority precision from the data output from the interpolation pixel generation unit and the data output from the second storage unit . Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
A motion vector detection step of detecting a motion vector using the reference area data and the data of the processing target block,
The motion vector detection step sequentially calculates the absolute value of the difference between the pixel values of the pixel data of the reference block and the pixel data of the processing target block, and subtracts the calculated absolute value of the difference from the threshold value. A motion vector detection method, characterized in that, when a result becomes negative, calculation of an absolute difference error sum for a reference block being processed is stopped. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
A motion vector detection step of detecting a motion vector using the reference area data and the data of the processing target block;
A search history holding step for holding a search history of reference blocks in the reference area data that has already been searched by the motion vector detection step;
In the motion vector detection method, the search execution is skipped for the reference block in which the search history held in the search history holding step exists. The motion vector detection method according to claim 22,
The search history holding step holds, as a search history, the search target position and the presence / absence of search for the reference blocks at eight peripheral positions. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
A motion vector detection step of detecting a motion vector using the reference area data and the data of the processing target block;
A transfer rule holding step in which the first address generation step holds a transfer rule that is a rule for generating an address of reference area data;
A transfer rule detection step of detecting the transfer rule from the transfer rule holding step based on an area determination code indicating an effective area of the image data and outputting to the first address generation step;
The first address generation step generates a reference area data address based on the transfer rule detected by the transfer rule detection step. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step using a memory having a double buffer configuration for simultaneously reading and reading data;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
A motion vector detection step of detecting a motion vector using the reference area data and the data of the processing target block,
A storage data control step for controlling reading and reading of reference area data in the first storage step,
In the stored data control step, the effective data from the other buffer memory that has been read out of one buffer memory from which data is read out of the two buffer memories used in the first storage step. A motion vector detection method characterized in that a part of the data is copied and the amount of data newly written in the first storage step is reduced. The motion vector detection method according to any one of claims 21 to 25,
The motion vector detection step is an integer accuracy motion vector detection step of detecting an integer accuracy motion vector using the reference area data and data of a processing target block. The motion vector detection method according to any one of claims 21 to 25,
The motion vector detection step receives pixel data constituting the reference area data as an input, and generates an interpolation pixel generation step that generates pixel data with a small number of precision.
A motion vector detection method comprising: a minority precision motion vector detection step of detecting a minority precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block. The motion vector detection method according to any one of claims 21 to 25,
The motion vector detection step includes an integer accuracy motion vector detection step of detecting an integer accuracy motion vector using the reference region data and data of a processing target block;
Interpolation pixel generation step for generating pixel data with a small number of precision by using pixel data constituting the reference area data as input, and
A motion vector detection method comprising: a minority precision motion vector detection step of detecting a minority precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
Interpolation pixel generation step for generating pixel data with a small number of precision by using pixel data constituting the reference area data as input, and
A minority precision motion vector detection step of detecting a minority precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block;
A reading direction determination step for determining the reading direction of the reference area data held in the first storage step according to the position of the data of the minority precision pixels generated in the interpolation pixel generation step and outputting to the second address generation step And
The second address generation step generates an address of pixel data to be output from the reference area data based on the reading direction determined by the reading direction determination step. The motion vector detection method according to claim 29,
The motion vector detection method characterized in that the reading direction determination step determines to read data horizontally or vertically with respect to the image data held in the first storage step. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
Interpolation pixel generation step for generating pixel data with a small number of precision by using pixel data constituting the reference area data as input, and
A minority-precision motion vector detection step for detecting a minority-precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block;
The small-precision motion vector detection step includes, from the interpolation pixel generation step, upper and lower, left and right, upper right and lower right, upper left and lower left, upper left and lower right, and upper right. When a minority precision reference block located at either the upper left corner or the lower right corner or the lower left diagonal position is output, detection of the minority precision motion vector for each of the output minority reference blocks is performed. A motion vector detection method characterized by being executed in parallel. The motion vector detection method according to any one of claims 29 to 31,
The motion vector detection method further includes an integer precision motion vector detection step of detecting an integer precision motion vector using the reference area data and the data of the processing target block. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
An integer precision motion vector detection step of detecting an integer precision motion vector using the reference region data and the data of the processing target block;
A search enable / disable condition holding step for holding a search enable / disable condition that is a condition for determining whether or not the position can be searched; and
A search availability condition detecting step for detecting a search availability condition from the search availability condition holding step based on an area determination code indicating an effective area of the reference area data, and outputting to the second address generation step;
The second address generation step determines whether or not the search is possible using the search availability condition detected by the search availability condition detection step when detecting the integer precision motion vector. The motion vector detection method according to claim 33,
Furthermore, the pixel data constituting the reference area data as an input, an interpolation pixel generation step for generating pixel data with a small number of precision,
A minority-precision motion vector detection step for detecting a minority-precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block;
The second address generation step uses the search availability condition holding step and the search availability condition detection step in combination when detecting an integer precision motion vector and a decimal precision motion vector. Vector detection method. The motion vector detection method according to claim 33,
Furthermore, the pixel data constituting the reference area data as an input, an interpolation pixel generation step for generating pixel data with a small number of precision,
A minority-precision motion vector detection step for detecting a minority-precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block;
The second address generation step uses the search availability condition holding step and the search availability condition detection step at the time of detecting an integer precision motion vector to determine whether search is possible and uses the search availability result to determine a small number. A motion vector detection method characterized by determining whether or not a search is possible when detecting an accurate motion vector. Block matching between a processing target block consisting of a plurality of pixels in the current image and a reference block consisting of a plurality of pixels in a predetermined reference area data of the previous image temporally preceding the current image In a motion vector detection method for detecting a motion vector by performing
A first address generation step for generating an address of the processing target block and the reference area data;
A first storage step for holding the reference area data instructed by the first address generation step;
A second storage step for holding the data of the processing target block instructed by the first address generation step;
A second address generation step of generating an address of pixel data to be output from the held reference area data and data of the processing target block;
An integer precision motion vector detection step of detecting an integer precision motion vector using the reference region data and the data of the processing target block;
The motion vector detected by the previous motion vector search is held, and the absolute difference error value calculated by the integer precision motion vector detection step is compared with a specific threshold value to determine whether the motion vector search can be executed. A search execution feasibility determination step for determining,
The second address generation step uses the motion vector detected in the previous motion vector search held in the search execution possibility determination step, and outputs pixel data from the reference area data and the data of the processing target block A motion vector detection method comprising generating an address of The motion vector detection method according to claim 36, wherein
The second address generation step further outputs data output from the reference area data held in the first storage step as a calculation result. The motion vector detection method according to claim 36 or claim 37,
Furthermore, the pixel data constituting the reference area data as an input, an interpolation pixel generation step for generating pixel data with a small number of precision,
A motion vector detection method, comprising: a minority precision motion vector detection step for detecting a minority precision motion vector using the data output from the interpolation pixel generation step and the data of the processing target block.

Images