JP2004134912A - Data processing apparatus and method, recording medium, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technology capable of matching images only with a very small computation complexity and detecting a motion vector or the like with high accuracy. <P>SOLUTION: In the case that pixel values P1 to P9 of 9 pixels included in a class code tap is in a state shown in the figure, the pixel values P6 and P8 closest to a threshold value Th are excluded from the class code tap, and the remaining pixel values P1, P2, P3, P4, P5, P7, P9 are respectively compared with the threshold value Th and quantized into 0 or 1 resulting in that a 7-bit class code 1010011 is produced. The technology above is applicable to encoders for applying compression encoding to moving picture signals. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a data processing device and method, a recording medium, and a program, for example, a data processing device suitable for use when matching pixels on continuous images and detecting a motion vector based on a result of the matching. And a method, a recording medium, and a program.
[0002]
[Prior art]
For example, in a process of compressing and encoding an image signal according to the MPEG2 (Moving Picture Experts Group) method or the like, an encoding process based on a correlation between adjacent frames, that is, a so-called motion compensated inter-frame prediction is used. In motion-compensated inter-frame prediction, a process of detecting a motion vector for each pixel or for a pixel block of a predetermined size between adjacent frames (one is described as a current frame and the other is described as a reference frame) is required.
[0003]
Conventionally, a method called a block matching algorithm has been used as a method for detecting a motion vector (for example, see Patent Document 1).
[0004]
FIG. 1 shows a configuration example of a motion vector detection device that detects a motion vector according to a block matching algorithm. The motion vector detecting device 1 is configured to hold a frame of an input image signal for one frame and output the frame signal to a subsequent stage, and a frame signal based on two frame image signals input from the frame memories 2 and 3. It comprises a detection unit 4 for detecting a motion vector in an image of an image signal input from the memory 2.
[0005]
The frame memory 2 holds the input image signal for one frame, and outputs the held image signal to the frame memory 3 and the detection unit 4 when the image signal of the next frame is input. The frame memory 3 holds the image signal input from the frame memory 2 for one frame, and outputs the held image signal to the detection unit 4 when the image signal of the next frame is input.
[0006]
Therefore, the image signals of two frames before and after are input to the detection unit 4. Hereinafter, an image of one frame of the image signal input from the frame memory 2 to the detection unit 4 is referred to as a target frame Fc. Further, the image of the image signal one frame before the target frame Fc input from the frame memory 3 to the detection unit 4 is described as a reference frame Fr.
[0007]
The detection unit 4 calculates a motion vector in the target frame Fc according to a block matching algorithm. The block matching algorithm will be described with reference to the correspondence diagram between the target frame Fc and the reference frame Fr shown in FIG. 2 and the flowchart shown in FIG.
[0008]
In the block matching algorithm, all pixels in the target frame Fc are sequentially set as a target pixel, and a reference block of a predetermined size (L × L pixels) centering on the target pixel and a search block SR in the reference frame Fr Is calculated in accordance with the following equation (1).
Σ _i Σ _j = | Fc (i, j) -Frn (i, j) | (1)
[0009]
Here, Fc (i, j) is the pixel value of the pixel of the reference block, Frn (i, j) is the pixel value of the pixel of the reference block with the identification number n, and Σ _i is i from 1 to L The sum operation when incremented by one, Σ _j means the sum operation when _j is incremented by one from 1 to L.
[0010]
Then, a difference vector between the center coordinates of the reference block and the coordinates of the pixel of interest when the sum of absolute differences of the pixel values of the corresponding pixel pairs of the reference block and the reference block is minimized is calculated as a motion vector.
[0011]
Specifically, the following processing is performed on the target pixel of the target frame Fc. In step S1, the detection unit 4 sets a search area SR in the reference frame Fc, which is larger in size than the reference block, around the same coordinates as the coordinates (x, y) of the target pixel of the target frame Fc.
[0012]
In step S2, the detection unit 4 initializes a variable min that stores the minimum value of the sum of absolute differences to the maximum value. For example, if one pixel value is 8-bit pixel, the size of the reference block is 4 × 4 pixels, variable min is initialized to ^{4096 (= 2 8 × 16)} .
[0013]
In step S3, the detection unit 4 initializes the identification number n of the reference block to be moved in the search area SR to 1. In step S4, the detection unit 4 initializes a variable sum that stores the calculation result of the sum of absolute difference to 0.
[0014]
In step S5, the detection unit 4 determines the sum of absolute differences of pixels of a pixel pair located at the same position as the reference block of the target frame Fc and the reference block of the identification number n in the search area SR set in the reference frame Fr. Is calculated and assigned to the variable sum. In step S6, the detection unit 4 compares the variable sum, which is the operation result of step S5, with the variable min, and determines whether the variable sum is smaller than the variable min. If it is determined that the variable sum is smaller than the variable min, the process proceeds to step S7.
[0015]
In step S7, the detection unit 4 replaces the variable min with the variable sum. Further, the detecting unit 4 stores the identification number n of the current reference block as a motion vector number.
[0016]
In step S8, the detection unit 4 determines whether or not the identification number n of the reference block is the maximum value, that is, whether or not the reference block has been moved to the entire search area SR. If it is determined that the identification number n of the reference block is not the maximum value, the process proceeds to step S9. In step S9, the detection unit 4 increments the identification number n of the reference block by 1, returns to the process in step S4, and repeats the subsequent processes.
[0017]
If it is determined in step S6 that the variable sum is not smaller than the variable min, the process skips step S7.
[0018]
Thereafter, if it is determined in step S8 that the identification number n of the reference block is not the maximum value, that is, if the reference block has been moved to the entire search area SR, the process proceeds to step S10.
[0019]
In step S10, the detection unit 4 determines a difference vector between the center coordinates of the reference block corresponding to the identification number n stored as the motion vector number and the coordinates (x, y) of the target pixel of the target frame Fc. It is calculated as the motion vector of the target pixel of the frame Fc. This is the end of the description of the block matching algorithm.
[0020]
[Patent Document 1]
Japanese Patent No. 3277417
[Problems to be solved by the invention]
The block matching algorithm described above has a problem that the amount of calculation of the sum of absolute differences of the pixels in the pixel pair in step S5 is extremely large, and most of the time of the image compression processing is spent on this calculation. there were.
[0022]
The present invention has been made in view of such a situation, and an object of the present invention is to enable matching between images with only a small amount of calculation, and to accurately detect a motion vector and the like.
[0023]
[Means for Solving the Problems]
The data processing device according to the present invention includes: a setting unit that sets a block of a predetermined size including target data based on input data; and a threshold based on a plurality of data included in the block set by the setting unit. Calculating means for calculating, among a plurality of data included in the block set by the setting means, determining means for determining exclusion data based on the threshold value calculated by the calculating means, and a block set by the setting means Generating means for encoding data other than the exclusion data determined by the determining means among the plurality of included data to 0 or 1 based on the threshold value calculated by the calculating means, and generating a class code for the data of interest And characterized in that:
[0024]
The calculating means calculates a threshold by adding a half of a dynamic range of a plurality of data included in the block to a minimum value of a plurality of data included in the block set by the setting means. Can be.
[0025]
The determination unit may determine a predetermined number of data having a value close to the threshold value calculated by the calculation unit as the exclusion data among the plurality of data included in the block set by the setting unit. .
[0026]
The deciding means decides, as exclusion data, data having a value belonging to a predetermined range centered on the threshold value calculated by the calculating means, among a plurality of data included in the block set by the setting means. be able to.
[0027]
The generating unit sets data other than the exclusion data determined by the determining unit to 0 or 1 based on the threshold calculated by the calculating unit, among the plurality of data included in the block set by the setting unit. The data that has been coded and determined as the exclusion data by the determination unit can be coded into 0 and 1, respectively, to generate a plurality of class codes for the data of interest.
[0028]
The data may be a pixel value of a pixel forming an image.
[0029]
A data processing method according to the present invention includes a setting step of setting a block of a predetermined size including target data based on input data, and a plurality of data included in the block set in the processing of the setting step. A calculating step of calculating a threshold, a determining step of determining exclusion data based on the threshold calculated in the processing of the calculating step from a plurality of data included in the block set in the processing of the setting step, and a setting step The data other than the exclusion data determined in the processing of the determining step among the plurality of data included in the block set in the processing of the above is respectively assigned 0 or 1 based on the threshold value calculated in the processing of the calculating step. Generating a class code for the data of interest.
[0030]
The recording medium program according to the present invention includes a setting step of setting a block of a predetermined size including target data based on input data, and a plurality of data included in the block set in the processing of the setting step. A calculating step of calculating a threshold by using a setting step of determining exclusion data based on the threshold value calculated in the processing of the calculating step from among a plurality of data included in the block set in the processing of the setting step; Of the plurality of data included in the block set in the processing of the step, the data other than the exclusion data determined in the processing of the determining step are respectively set to 0 or 1 based on the threshold value calculated in the processing of the calculating step. Encoding and generating a class code for the data of interest.
[0031]
The program according to the present invention includes: a setting step of setting a block of a predetermined size including target data based on input data; and a threshold value based on a plurality of data included in the block set in the processing of the setting step. A calculating step of calculating, a determining step of determining exclusion data based on a threshold calculated in the processing of the calculating step from among a plurality of data included in the block set in the processing of the setting step, and a processing of the setting step Among the plurality of data included in the block set in the above, the data other than the exclusion data determined in the processing of the determination step, respectively, is encoded to 0 or 1 based on the threshold calculated in the processing of the calculation step, Generating a class code for the data of interest. To.
[0032]
In the data processing device and method and the program according to the present invention, a block of a predetermined size including data of interest is set, and a threshold is calculated based on a plurality of data included in the set block. Excluded data is determined from a plurality of included data based on the calculated threshold. Further, among the plurality of data included in the set block, data other than the determined exclusion data are respectively coded to 0 or 1 based on the calculated threshold, and a class code for the data of interest is generated. You.
[0033]
BEST MODE FOR CARRYING OUT THE INVENTION
An example of the configuration of a motion vector detection device to which the present invention is applied will be described with reference to FIG. The motion vector detecting device 11 includes frame memories 12 and 13, a class code generator 14, an ME memory 15, and a motion vector calculator 16.
[0034]
The frame memory 12 holds the input image signal for one frame, and outputs the held image signal to the frame memory 13 and the class code generation unit 14 when the image signal of the next frame is input. The frame memory 13 holds the input image signal for one frame, and outputs the held image signal to the class code generation unit 14 when the image signal of the next frame is input from the frame memory 12.
[0035]
Accordingly, the image signal input from the frame memory 13 to the class code generator 14 is one frame before the image signal input from the frame memory 12 to the class code generator 14. Hereinafter, the image of one frame of the image signal input from the frame memory 12 to the class code generation unit 14 is referred to as a target frame Fc. The image of the image signal one frame before the target frame Fc, which is input from the frame memory 13 to the class code generation unit 14, is described as a reference frame Fr.
[0036]
The class code generator 14 generates a class code indicating a spatial feature for each pixel of the target frame Fc, and outputs the generated class code to the motion vector calculator 16. Specifically, all the pixels of the target frame Fc are sequentially set as a target pixel, a class code tap of a predetermined size around the target pixel is determined, and a plurality of pixels included in the class code tap are determined. Is quantized to 0 or 1 by 1-bit ADRC (Adaptive Dynamic Range Coding) to generate a class code having a predetermined number of bits. The process of generating the class code will be described in detail with reference to FIG.
[0037]
Further, the class code generation unit 14 generates a class code indicating a spatial feature for each pixel of the reference frame Fr, and outputs the generated class code to the ME memory 15. However, since the reference frame Fr is the target frame Fc at the previous timing, its class code has already been generated one time before. Therefore, the class code generated for each pixel of the target frame Fc at the previous timing may be held, and may be used as the class code for each pixel of the current reference frame Fr.
[0038]
The ME memory 15 stores the class code for each pixel of the reference frame Fr input from the class code generation unit 14 in association with the class code and the coordinates of each pixel.
[0039]
FIG. 5 shows the structure of the ME memory 15. The ME memory 15 is composed of a × b cells indicated by feature amount addresses 0 to a and flag addresses 0 to b. Hereinafter, for example, the cell of the feature address 1 and the flag address 2 is described as a cell (1, 2).
[0040]
The feature address corresponds to the class code generated by the class code generation unit 14. For example, if the class code is 7 bits, when the maximum value a = 2 ⁷ next feature quantity addresses, class code is 9 bits, the maximum value a = 2 ⁹ feature quantity address.
[0041]
In the cells after the flag address 1 of the feature amount address 0, the coordinates of the pixels of the reference frame Fr in which the class code 0 has been generated by the class code generation unit 14 are stored in raster order. In the cell of the flag address 0 of the feature address 0, the number of cells after the flag address 1 of the feature address 0 in which the coordinates of the pixel of the class code 0 are stored. For example, when the class code 0 is generated for three pixels among the pixels of the reference frame Fr, the cells (0, 1), (0, 2), and (0, 3) are assigned to each of the three pixels. The coordinates are stored, and 3 is stored in cell (0, 0).
[0042]
In the cells after the flag address 1 of the feature amount address 1, the coordinates of the pixels of the reference frame Fr in which the class code 1 is generated by the class code generation unit 14 are stored in the raster order. The cell of flag address 0 of feature amount address 1 stores the number of cells after flag address 1 of feature amount address 0 in which the coordinates of the pixel of class code 1 are stored. For example, when the class code 1 is generated for 10 pixels among the pixels of the reference frame Fr, the coordinates of each of the 10 pixels are stored in cells (1, 1),..., Cell (1, 10). Then, 10 is stored in the cell (1, 0). The same applies to cells subsequent to the feature address 2.
[0043]
Referring back to FIG. The motion vector calculation unit 16 searches the ME memory 15 for the coordinates of the pixel of the reference frame Fr having the same class code for each pixel of the target frame Fc, and finds the closest distance to the coordinates of the target pixel among the searched pixels. Is determined as a pixel corresponding to the target pixel, and a motion vector of the target pixel is calculated.
[0044]
The structure of the ME memory 15 and the details of the processing of the motion vector calculation unit 16 have already been proposed by the present applicant as Japanese Patent Application No. 2002-222044.
[0045]
Next, the first class code generation processing by the class code generation unit 4 which is the main feature of the present invention will be described with reference to the flowchart in FIG. This process is executed for each pixel of interest, with each pixel of the target frame Fc being sequentially set as the pixel of interest.
[0046]
In step S11, the class code generation unit 14 sets a class code tap of a predetermined size centering on the target pixel, and acquires pixel values of a plurality of pixels included in the class code tap. In the following, as shown in FIG. 7, the description will be continued with the size of the class code tap set to 3 × 3 pixels and the pixel values from the upper left pixel to the lower right pixel being P1 to P9, respectively.
[0047]
In step S12, the class code generation unit 14 determines the maximum value P _MAX and the minimum value P _MIN of the pixel values P1 to P9. In step S13, the class code generation unit 14 calculates a dynamic range DR (= | maximum value P _MAX -minimum value P _MIN |) of the pixel values P1 to P9. In step S14, the class code generation unit 14 determines the threshold Th by adding the dynamic range DR / 2 to the minimum value P _MIN of the pixel values P1 to P9 as in the following equation (2).
Th = P _MIN + DR / 2 (2)
[0048]
In step S15, the class code generation unit 14 excludes a predetermined number (for example, two) of pixel values closest to the threshold Th from the pixel values P1 to P9. In step S16, the class code generation unit 14 compares each of the remaining seven pixels among the pixel values P1 to P9 with a threshold Th, and quantizes the remaining seven pixels to 1 when the pixel value is larger than the threshold Th, and less than the threshold Th. In this case, it is quantized to 0, and 7 bits arranged in numerical order are generated as the class code of the pixel of interest.
[0049]
For example, when the pixel values P1 to P9 of the nine pixels included in the class code tap are in the state shown in FIG. 8, the pixel values P6 and P8 closest to the threshold Th are excluded, and the 7-bit class code 1010011 is generated.
[0050]
Further, for example, when the pixel values P1 to P9 of the nine pixels included in the class code tap are in the state shown in FIG. 9, the pixel values P5 and P6 closest to the threshold Th are excluded, and the 7-bit class Code 1010101 is generated.
[0051]
Note that instead of excluding a predetermined number of pixel values close to the threshold Th, a predetermined number of pixel values close to the threshold Th are quantized to 0 and a 9-bit class code, respectively. In this case, a 9-bit class code may be generated.
[0052]
That is, for example, when the pixel values P1 to P9 of the nine pixels included in the class code tap are in the state shown in FIG. 10, the two pixel values P6 and P8 closest to the threshold Th are respectively Four types of 9-bit class codes 101000101, 101000111, 101001101, and 101001111 when quantized to 0 and when quantized to 1 may be generated as class codes corresponding to the pixel of interest.
[0053]
Also, instead of performing the above-described processing on a predetermined number of pixel values close to the threshold Th, all pixel values included in a predetermined range (± Δ) centered on the threshold Th are calculated. Then, the above-described processing may be performed.
[0054]
For example, if the pixel values P1 to P9 of the nine pixels included in the class code tap are in the state shown in FIG. 11, the pixel value P6 and the pixel value P6 included in a predetermined range (± Δ) centered on the threshold value Th Four types of 9-bit class codes 101000111, 101001111, 111000111, and 111001111 may be generated when the value P8 is quantized to 0 and when it is quantized to 1, respectively.
[0055]
Alternatively, a 7-bit class code may be generated by excluding the pixel values P6 and P8 included in a predetermined range (± Δ) centered on the threshold Th.
[0056]
As described above, of all the pixels included in the class code tap, pixel values near the threshold value Th are excluded from the quantization target, or the class code when quantized to 0 and 1 By generating a class code when quantization is performed, even if a pixel value near the threshold value Th fluctuates due to noise or the like, the occurrence of bit inversion in the class code is suppressed. be able to. Therefore, the robustness of the class code can be improved.
[0057]
The number of pixels constituting the class code tap and the number of bits of the class code are not limited to the examples described above, but are arbitrary. This is the end of the description of the first class code generation process.
[0058]
Next, a second class code generation process that can be substituted for the above-described first class code generation process will be described with reference to the flowchart in FIG. This process is executed for each pixel of interest, with each pixel of the target frame Fc being sequentially set as the pixel of interest.
[0059]
In step S21, the class code generation unit 14 sets a class code block of a predetermined size centering on the target pixel, and acquires pixel values of a plurality of pixels included in the class code block. In the following, as shown in FIG. 13, the description will be continued on the assumption that the size of the class code block is 3 × 3 pixels, and the pixel values from the upper left pixel to the lower right pixel are P1 to P9, respectively.
[0060]
In step S22, the class code generation unit 14 generates a plurality of class code taps using a predetermined number (hereinafter, for example, seven) of pixel values among the pixel values P1 to P9 included in the class code block. Generate candidates. The pattern of the class code tap candidate generated here is set in advance. FIG. 14 shows three examples of a plurality of class code tap candidates to be generated. The arrows on the class code tap candidates in FIG. 14 indicate the order in which pixel values are quantized and arranged to generate a 7-bit class code.
[0061]
In step S23, the class code generation unit 14 calculates an evaluation value for each class code tap candidate in order to determine which class code tap candidate is most suitable for the class code tap.
[0062]
The process of calculating the evaluation value of each class code tap candidate will be described with reference to the flowchart in FIG. In step S31, the class code generation unit 14 detects the maximum value P _MAX and the minimum value P _MIN among the seven pixel values included in the class code tap candidate. In step S32, the class code generation unit 14 calculates a dynamic range DR (= | maximum value P _MAX -minimum value P _MIN |) of the pixel values of the seven pixels included in the class code tap candidate.
[0063]
In step S33, the class code generation unit 14 determines the threshold Th using Expression (2).
[0064]
In step S34, the class code generation unit 14 calculates the sum of the differences between the pixel values of the seven pixels included in the class code tap candidates and the threshold value Th as the evaluation values of the class code tap candidates.
[0065]
For example, the evaluation value of the class code tap candidate 1 (FIG. 14A) having a pixel value as shown in FIG. 16 is calculated as in the following equation (3).
Evaluation value = | Th1-P3 | + | Th1-P4 | + | Th1-P5 | + | Th1-P6 | + | Th1-P7 | + | Th1-P8 | + | Th1-P9 | (3)
However, Th1 is a threshold calculated based on the pixel values of seven pixels included in the class code tap candidate 1.
[0066]
Further, for example, the evaluation value of the class code tap candidate 2 (FIG. 14B) having a pixel value as shown in FIG. 17 is calculated as in the following equation (4).
Evaluation value = | Th2-P3 | + | Th2-P4 | + | Th2-P5 | + | Th2-P6 | + | Th2-P7 | + | Th2-P8 | + | Th2-P9 | (4)
Note that Th2 is a threshold calculated based on the pixel values of seven pixels included in the class code tap candidate 2.
[0067]
This concludes the description of the process of calculating the evaluation value of each class code tap candidate. The process returns to step S24 in FIG.
[0068]
In step S24, the class code generation unit 14 has the largest evaluation value calculated in the processing in step S23 among the plurality of class code tap candidates (that is, the variance of each pixel value around the threshold Th is large). ) Are determined as class code taps.
[0069]
In step S25, the class code generation unit 14 quantizes the pixel values of the seven pixels included in the class code tap determined in step S24 to 0 or 1 by 1-bit ADRC, and generates a 7-bit class. Generate code.
[0070]
For example, when the class code tap candidate 1 including seven pixels having the pixel values shown in FIG. 16 is determined as the class code tap in the process of step S24, a 7-bit class code 1001100 is generated. For example, when the class code tap candidate 2 including seven pixels having the pixel values shown in FIG. 17 is determined as the class code tap in the process of step S24, a 7-bit class code 1001100 is generated. Is done.
[0071]
As described above, among the plurality of class code tap candidates, the one having the largest evaluation value (that is, the one having a large variance of each pixel value around the threshold value Th) is determined as the class code tap, and the class code tap is determined. Is generated, the occurrence of bit inversion in the class code can be suppressed. Therefore, the robustness of the class code can be improved.
[0072]
The number of pixels constituting the class code block and the number of pixels constituting the class code tap candidate are arbitrary. In addition to the above-described example, for example, the class code block is composed of 5 × 5 pixels. Alternatively, the class code tap candidate may be configured with nine pixels. This is the end of the description of the second class code generation processing.
[0073]
As described above, according to the present embodiment, the class code generation unit 14 provides a class code having high robustness to each pixel of the target frame Fc and the reference frame Fr by an easy operation of 1-bit ADRC. Can be generated. Therefore, the pixels of the target frame Fc and the pixels of the reference frame Fr can be matched with high accuracy. Therefore, it is possible to accurately detect a motion vector.
[0074]
Further, the present invention can be applied to a case where a class code is generated for arbitrary data such as audio data, for example, in addition to the pixel values of the pixels constituting the image.
[0075]
Incidentally, the series of processes described above can be executed by hardware, but can also be executed by software. When a series of processing is executed by software, a program constituting the software may execute various functions by installing a computer built into dedicated hardware or installing various programs. It is installed from a recording medium into a possible general-purpose personal computer configured as shown in FIG. 18, for example.
[0076]
This personal computer includes a CPU (Central Processing Unit) 31. An input / output interface 35 is connected to the CPU 31 via a bus 34. The bus 34 is connected to a ROM (Read Only Memory) 32 and a RAM (Random Access Memory) 33.
[0077]
The input / output interface 35 includes an input unit 36 including an input device such as a keyboard and a mouse for inputting operation commands by a user, a CRT (Cathode Ray Tube) or an LCD (Liquid Crystal Display) for displaying an image of a processing result. An output unit 37, a storage unit 38 such as a hard disk drive that stores programs and various data, and a communication unit that includes a modem, a LAN (Local Area Network) adapter, and executes communication processing via a network represented by the Internet. 39 are connected. A drive 40 for reading and writing data from and to a recording medium such as a magnetic disk 41, an optical disk 42, a magneto-optical disk 43, and a semiconductor memory 44 is connected.
[0078]
Programs that cause the CPU 31 to execute the above-described series of processes include a magnetic disk 41 (including a flexible disk), an optical disk 42 (including a CD-ROM (Compact Disc-Read Only Memory), a DVD (including a Digital Versatile Disc)), and a magneto-optical disk. 43 (including an MD (Mini Disc)) or stored in a semiconductor memory 44 and supplied to a personal computer, read by a drive 40 and installed in a hard disk drive built in the storage unit 38. The program installed in the storage unit 38 is loaded from the storage unit 38 to the RAM 33 and executed according to a command from the CPU 31 corresponding to a command from the user input to the input unit 36.
[0079]
In this specification, the steps of describing a program recorded on a recording medium include, in addition to processing performed in chronological order according to the described order, not only chronological processing but also parallel or individual processing. This includes the processing to be executed.
[0080]
【The invention's effect】
As described above, according to the present invention, matching between images can be performed with only a small amount of calculation. Further, according to the present invention, it is possible to accurately detect a motion vector and the like.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration example of a conventional motion vector detection device.
FIG. 2 is a diagram showing a correspondence relationship between a target frame Fc and a reference frame Fr.
FIG. 3 is a flowchart illustrating a block matching algorithm.
FIG. 4 is a block diagram illustrating a configuration example of a motion vector detection device according to an embodiment of the present invention.
FIG. 5 is a diagram showing a structure of the ME memory of FIG. 4;
FIG. 6 is a flowchart illustrating a first class code generation process by a class code generation unit in FIG. 4;
FIG. 7 is a diagram illustrating an example of a class code tap;
FIG. 8 is a diagram for explaining a first class code generation process.
FIG. 9 is a diagram illustrating a first class code generation process.
FIG. 10 is a diagram for describing a modification of the first class code generation process.
FIG. 11 is a diagram for describing a modification of the first class code generation process.
12 is a flowchart illustrating a second class code generation process performed by the class code generation unit in FIG.
FIG. 13 is a diagram illustrating an example of a class code block.
FIG. 14 is a diagram illustrating an example of class code tap candidates.
FIG. 15 is a flowchart illustrating an evaluation value calculation process in step S23 of FIG.
FIG. 16 is a diagram illustrating a second class code generation process.
FIG. 17 is a diagram illustrating a second class code generation process.
FIG. 18 is a block diagram illustrating a configuration example of a general-purpose personal computer.
[Explanation of symbols]
11 motion vector detecting device, 12, 13 frame memory, 14 class code generation unit, 15 ME memory, 16 motion vector calculation unit, 31 CPU, 41 magnetic disk, 42 optical disk, 43 magneto-optical disk,
44 Semiconductor Memory

Claims (9)

In a data processing device that generates a class code indicating a feature amount and processes data based on the generated class code,
Setting means for setting a block of a predetermined size including the data of interest based on the input data;
Calculation means for calculating a threshold based on a plurality of the data included in the block set by the setting means,
From among the plurality of data included in the block set by the setting unit, a determination unit that determines exclusion data based on the threshold value calculated by the calculation unit,
Of the plurality of data included in the block set by the setting unit, data other than the exclusion data determined by the determining unit, based on the threshold calculated by the calculating unit, 0 or And a generating means for generating the class code for the target data. The calculation means adds １ ／ of the dynamic range of the data included in the block to the minimum value of the data included in the block set by the setting means, and sets the threshold value. The data processing device according to claim 1, wherein the calculation is performed. The determining means determines a predetermined number of data having a value close to the threshold calculated by the calculating means as the exclusion data among the plurality of data included in the block set by the setting means. The data processing apparatus according to claim 1, wherein the data processing is performed. The determining unit excludes the data having a value belonging to a predetermined range centered on the threshold calculated by the calculating unit, among the plurality of data included in the block set by the setting unit. The data processing device according to claim 1, wherein the data is determined as data. The generating means, among the plurality of data included in the block set by the setting means, data other than the exclusion data determined by the determining means, each of the threshold values calculated by the calculating means And encoding the data determined as the exclusion data by the determination means into 0 and 1, respectively, to generate a plurality of the class codes for the data of interest. The data processing device according to claim 1. The data processing device according to claim 1, wherein the data is a pixel value of a pixel forming an image. In the data processing method of a data processing device that processes data based on the generated class code that generates a class code indicating a feature amount,
A setting step of setting a block of a predetermined size including the data of interest based on the input data;
A calculating step of calculating a threshold based on a plurality of the data included in the block set in the processing of the setting step,
From among the plurality of data included in the block set in the processing of the setting step, a determining step of determining exclusion data based on the threshold calculated in the processing of the calculating step,
Among the plurality of data included in the block set in the processing of the setting step, data other than the exclusion data determined in the processing of the determining step, respectively, the data calculated in the processing of the calculating step Generating a class code for the data of interest by encoding to 0 or 1 based on a threshold value. A program for generating a class code indicating a feature amount and processing data based on the generated class code,
A setting step of setting a block of a predetermined size including the data of interest based on the input data;
A calculating step of calculating a threshold based on a plurality of the data included in the block set in the processing of the setting step,
From among the plurality of data included in the block set in the processing of the setting step, a determining step of determining exclusion data based on the threshold calculated in the processing of the calculating step,
Among the plurality of data included in the block set in the processing of the setting step, data other than the exclusion data determined in the processing of the determining step, respectively, the data calculated in the processing of the calculating step Generating a class code for the data of interest by encoding the data into 0 or 1 based on a threshold value. Generate a class code indicating the feature amount, based on the generated class code, a computer that processes data,
A setting step of setting a block of a predetermined size including the data of interest based on the input data;
A calculating step of calculating a threshold based on a plurality of the data included in the block set in the processing of the setting step,
From among the plurality of data included in the block set in the processing of the setting step, a determining step of determining exclusion data based on the threshold calculated in the processing of the calculating step,
Among the plurality of data included in the block set in the processing of the setting step, data other than the exclusion data determined in the processing of the determining step, respectively, the data calculated in the processing of the calculating step A code that is coded to 0 or 1 based on a threshold value and generates the class code for the data of interest.

Images