CN100508607C - block matching method and device - Google Patents

Abstract

A block matching device comprises a plurality of operation modules, each operation module is used for calculating pixel differences between a plurality of target pixels in a target block and a plurality of reference pixels in a reference block, wherein each operation module comprises a plurality of operation units which are respectively used for calculating the pixel differences between one of the target pixels and one of the reference pixels; and a plurality of adding units respectively coupled to the operation modules, each adding unit for adding the calculation results of the operation units in the corresponding operation module.

Description

Method and device for block matching

technical field

The invention relates to a block matching method and device, in particular to a method and device for calculating pixel differences between blocks.

Background technique

Many video processing processes (such as motion estimation in MPEG2/MPEG4) need to use the operation result of image block matching. For example, when encoding a target block in a frame, encoding needs to be performed according to the difference between the target block and a reference macroblock that is most similar in the previous or subsequent frame on the time axis. Generally speaking, the block matching will match the target block with all similar-sized macroblocks to be matched in a search range of the previous frame or the next frame one by one, so as to find the most similar macroblock to the target block Reference block.

The size of the target block allowed by different video standard specifications is different, such as 8 X 8, 8 X 16, 16 X 8 or 16 X 16 and so on. However, in the prior art, target blocks of different sizes require different circuits for block matching, thus increasing the cost and complexity of circuit implementation. Since block matching requires a large number of calculations, how to implement block matching algorithms in a more efficient manner is one of the topics of concern in the industry.

Contents of the invention

Therefore, one of the objectives of the present invention is to provide a block pixel difference calculation device capable of processing target blocks of different sizes.

According to an embodiment of the present invention, a block matching device is disclosed, which includes a plurality of computing modules, and each computing module is used to calculate a plurality of target pixels in a target block and a plurality of reference pixels in a reference block The pixel difference between them, wherein each operation module includes a plurality of operation units, respectively used to calculate the pixel difference between one of these target pixels and one of these reference pixels; and a plurality of addition units, respectively coupled to these operations module, each adding unit is used to add up the calculation results of multiple calculation units in the corresponding calculation module, and each addition unit is used to accumulate the calculation results of the corresponding calculation module in multiple calculation cycles.

According to an embodiment of the present invention, a block pixel difference computing device is disclosed, which is used to calculate the difference between a target block of a target frame and a first reference block and a second reference block in a reference frame, The target block includes a first pixel and a second pixel, the first reference block includes a first reference pixel and a second reference pixel, the second reference block includes the second reference pixel and a first reference pixel Three reference pixels, the block pixel difference calculation device includes: a first calculation unit used to calculate the difference between the first pixel and the first reference pixel; a second calculation unit used to calculate the difference between the first pixel and the first reference pixel The difference between the second reference pixel; a third computing unit used to calculate the difference between the second pixel and the second reference pixel; a fourth computing unit used to calculate the difference between the second pixel and the third reference pixel difference; a first addition unit, coupled to the first and third calculation units, used to add up the calculation results of the first and third calculation units; and a second addition unit, coupled to the second and fourth calculation units The computing unit is used for summing up the computing results of the second and fourth computing units.

According to an embodiment of the present invention, a block pixel difference calculation method is disclosed, which is used to calculate the difference between a target block of a target frame and a first reference block and a second reference block in a reference frame, The target block includes a first pixel and a second pixel, the first reference block includes a first reference pixel and a second reference pixel, the second reference block includes the second reference pixel and a first reference pixel Three reference pixels, the block pixel difference calculation method includes: (a) calculating the difference between the first pixel and the first reference pixel; (b) calculating the difference between the first pixel and the second reference pixel; (c) ) calculating the difference between the second pixel and the second reference pixel; (d) calculating the difference between the second pixel and the third reference pixel; (e) summing the results of steps (a) and (b); and (f) summing up the operation results of steps (c) and (d).

Description of drawings

FIG. 1 is a schematic diagram of a target screen.

FIG. 2 is a schematic diagram of a reference frame.

FIG. 3 is a schematic diagram of a block pixel difference calculation device according to an embodiment of the present invention.

FIG. 4 is a schematic diagram of the data flow of an embodiment of matching an 8×8 target block by the block pixel difference computing device in FIG. 3 .

Fig. 5 is a schematic diagram of a target block with a size of 16 X 8.

FIG. 6 is a schematic diagram of the data flow of an embodiment of matching a 16×8 target block by the block pixel difference computing device in FIG. 3 .

Fig. 7 is a schematic diagram of a target block with a size of 16 X 16.

FIG. 8 is a schematic diagram of the data flow of an embodiment of matching a 16×16 target block by the block pixel difference computing device in FIG. 3 .

Description of reference symbols

100, 500, 700 target screen 110, 510, 710 target block 200 reference screen 210 search area 300 Block pixel difference operation device 302, 304, 306, 308, 310, 312, 314, 316 Operation module 322, 324, 326, 328, 330, 332, 334, 336 Addition unit 402, 404, 406, 408 arithmetic unit 512, 514, 712, 714, 716, 718 subblock

Detailed ways

Please refer to FIG. 1 , which is a schematic diagram of a target frame 100 . The target frame 100 includes a target block 110 with a size of 8×8. For convenience of description, each pixel included in the target block 110 is marked with a corresponding coordinate. In the following description, each pixel in the target block 110 is called C(x, y), where (x, y) is the coordinate of the pixel.

FIG. 2 is a schematic diagram of a reference frame 200 . As known to those skilled in the art, the reference frame 200 is usually a previous frame or a subsequent frame of the target frame 100, but not limited thereto. The reference frame 200 includes a search area 210 with a size of n×m. In the following description, each pixel in the search area 210 is called R(x, y), where (x, y) is the coordinate of the pixel.

FIG. 3 is a schematic diagram of a block pixel difference calculation device 300 according to an embodiment of the present invention. The block pixel difference computing device 300 includes 8 computing modules (computing module) 302-316; and 8 adding units (adding unit) 322-336. Each calculation module includes 8 processing elements (PEs), which are used to calculate the difference between a pixel of the target block 110 and a pixel of the search area 210 . In this embodiment, each computing unit (PE) is used to calculate an absolute difference (absolute difference, AD) between pixels. As shown in FIG. 3 , all computing units in the same computing module are coupled to corresponding adding units. In this embodiment, each adding unit is used to sum up the operation results of all the operation units in the corresponding operation module, that is, has the function of an adder. In this embodiment, each adding unit is also used to add up the operation results obtained by the corresponding operation module in a plurality of operation cycles, that is, it has the function of an accumulator.

Please refer to FIG. 4 , which shows a data flow diagram 400 of an embodiment in which the block pixel difference calculation device 300 matches multiple reference blocks (reference blocks) in the search area 210 of the target block 110 and the reference frame 200 . For the convenience of illustration, each reference block in the search area 210 is defined by the coordinates of its upper leftmost pixel. For example, a reference block in the upper left corner of the search area 210 is defined as a reference block RB _{8 x 8} (1, 1), and a reference block RB _{8 x 8} (1, 1) is shifted to the right by one pixel distance A block is defined as a reference block RB _{8 x 8} (2, 1) and so on. In addition, in order to avoid the illustration being too complicated, the block pixel difference calculation device 300 is simplified in FIG. 4 , and its internal connections are not shown.

In Fig. 4, eight horizontal dotted lines (horizontal dotted lines) passing through the block pixel difference operation device 300 respectively represent the data streams of eight pixels in the same column (row) in the target block 110; The oblique dotted lines of the difference operation device 300 respectively represent the data flow of fifteen pixels in the same row in the search area 210 . It should be noted that in this embodiment, each pixel data is synchronously (that is, also in the same operation cycle) input to all the operation units on the corresponding dotted line, so that the pixel data is loaded to the block pixels When the difference operation device 300 is used, no delay will be caused.

In the first operation cycle, the first column of pixel data in the target block 110, that is, pixels C(1,1), C(2,1), . . . , C(7,1) and C(8 , 1), will be synchronously input to all computing units corresponding to the horizontal dotted line. For example, in the first computing cycle, the pixel C(1,1) will be synchronously input to eight computing units such as the computing units 402 and 404 in the first row of computing units, and the pixel C(2,1) will be synchronously input to the There are eight arithmetic units such as the arithmetic units 406, 408 in the second column of arithmetic units, and so on. At the same time, the first 15 pixels in the first column of pixels in the search area 210, that is, pixels R(1,1), R(2,1), . . . , R(14,1) and R(15,1 ), will be synchronously input to all computing units on the corresponding oblique dotted lines. For example, in the first operation cycle, the pixel R(1,1) is synchronously input to the operation unit 402 , and the pixel R(2,1) is synchronously input to the operation units 404 and 406 , and so on.

In the second operation cycle, the second column of pixel data in the target block 110, that is, pixels C(1,2), C(2,2), . . . , C(7,2) and C(8 , 2), will be respectively synchronously input to all computing units corresponding to a horizontal dotted line. At the same time, the first 15 pixels in the second column of pixels in the search area 210, that is, pixels R(1,2), R(2,2), . . . , R(14,2) and R(15,2 ), will be synchronously input to all computing units corresponding to a diagonal dotted line. By analogy, in the eighth operation cycle, the eighth row of pixel data in the target block 110, that is, pixels C(1,8), C(2,8), . . . , C(7,8) and C(8, 8) will be synchronously input to all computing units corresponding to a horizontal dotted line. At the same time, the first 15 pixels in the eighth row of pixels in the search area 210, that is, the pixels R(1,8), R(2,8), . . . , R(14,8) and R(15,8 ), will be synchronously input to all computing units corresponding to a diagonal dotted line.

In each operation cycle, each operation unit will synchronously calculate the absolute difference (AD) of the two pixel values loaded by it. For example, in the first operation cycle, the operation unit 402 in the operation module 302 will calculate the absolute difference between the pixel C(1,1) and the pixel R(1,1), and the operation unit 406 will calculate the absolute difference between the pixel C(2,1) ) and the absolute difference of pixel R(2,1). Meanwhile, the computing unit 404 in the computing module 304 will calculate the absolute difference between the pixel C(1,1) and the pixel R(2,1), and the computing unit 408 will calculate the absolute difference between the pixel C(2,1) and the pixel R(3, 1) The absolute difference. In the second calculation cycle, the calculation unit 402 will calculate the absolute difference between the pixel C(1,2) and the pixel R(1,2), and the calculation unit 406 will calculate the absolute difference between the pixel C(2,2) and the pixel R(2,2). ), the operation unit 404 will calculate the absolute difference between pixel C(1,2) and pixel R(2,2), and the operation unit 408 will calculate the absolute difference between pixel C(2,2) and pixel R(3,2) Absolutely bad. By analogy, in the eighth operation cycle, the operation unit 402 will calculate the absolute difference between the pixel C (1, 8) and the pixel R (1, 8), and the operation unit 406 will calculate the absolute difference between the pixel C (2, 8) and the pixel R (2,8), the operation unit 404 will calculate the absolute difference between pixel C(1,8) and pixel R(2,8), and the operation unit 408 will calculate the absolute difference between pixel C(2,8) and pixel R( 3, 8) the absolute difference.

而运算模块302的八个运算单元在第二运算周期的运算结果总和可表示为

依此类推，运算模块302的八个运算单元在第八运算周期的运算结果总和可表示为

换言之，加法单元322加总并累加运算模块302自第一运算周期至第八运算周期的运算结果的数学式可表示为：As can be seen from the foregoing, the sum of the calculation results of the eight calculation units of the calculation module 302 in the first calculation cycle can be expressed as

The sum of the calculation results of the eight calculation units of the calculation module 302 in the second calculation cycle can be expressed as

By analogy, the sum of the calculation results of the eight calculation units of the calculation module 302 in the eighth calculation period can be expressed as

In other words, the mathematical expression of the addition unit 322 summing up and accumulating the operation results of the operation module 302 from the first operation cycle to the eighth operation cycle can be expressed as:

$\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

As known to those skilled in the art, the D value of the formula (1) is the sum of the absolute difference between the target block 110 and the reference block RB _{8 x 8} (1, 1) in the upper left corner of the search area 210 (sum of absolute difference, SAD).

而运算模块304的八个运算单元在第二运算周期的运算结果总和可表示为

依此类推，运算模块304的八个运算单元在第八运算周期的运算结果总和可表示为

换言之，加法单元324加总并累加运算模块304自第一运算周期至第八运算周期的运算结果的数学式可表示为：Similarly, the sum of the calculation results of the eight calculation units of the calculation module 304 in the first calculation cycle can be expressed as

The sum of the calculation results of the eight calculation units of the calculation module 304 in the second calculation period can be expressed as

By analogy, the sum of the calculation results of the eight calculation units of the calculation module 304 in the eighth calculation period can be expressed as

In other words, the mathematical expression of the addition unit 324 summing up and accumulating the operation results of the operation module 304 from the first operation cycle to the eighth operation cycle can be expressed as:

$\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

The value of formula (2) is the sum of absolute differences between the target block 110 and the reference block RB _{8 x 8} (2,1) in the search area 210 .

By analogy, the mathematical formula of the addition unit 336 summing up the calculation results of the operation module 316 from the first operation cycle to the eighth operation cycle can be expressed as:

$\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

The value of formula (3) is the sum of absolute differences between the target block 110 and the reference block RB _{8 x 8} (8, 1) in the search area 210 .

Therefore, after the operation of the first eight operation cycles, the values accumulated in the eight adding units are respectively the eight reference blocks corresponding to the target block 110 and the search area 210 (that is, the reference block RB _{8 x 8} (1, 1), RB _{8 x 8} (2, 1), ..., and RB _{8 x 8} (8, 1)).

Next, in the ninth operation cycle, the first column of pixel data in the target block 110, that is, pixels C(1,1), C(2,1), . . . , C(7,1) and C (8, 1), as mentioned above, are respectively synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, in the first column of pixels in the search area 210, 15 pixels starting from pixel R(9,1), that is, pixels R(9,1), R(10,1), ..., R(22 , 1) and R(23, 1) will be synchronously input to all computing units corresponding to the diagonal dotted lines. In the tenth operation cycle, the second row of pixel data in the target block 110, that is, pixels C(1,2), C(2,2), . . . , C(7,2) and C(8, 2), which will be synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, in the second column of pixels in the search area 210, the 15 pixels starting from the pixel R(9,2), that is, the pixels R(9,2), R(9,2), . . . , R(22 , 2) and R(23, 2) will be synchronously input to all computing units corresponding to the diagonal dotted line. By analogy, after the operations of the ninth to sixteenth computing cycles, the accumulated values in the eight adding units are respectively the following eight reference blocks in the target block 110 and the search area 210 (that is, the reference block RB _{8 x 8} (9, 1), RB _{8 x 8} (10, 1), ..., and RB _{8 x 8} (16, 1)).

It can be seen from the above that the block pixel difference calculation device 300 of this embodiment can calculate the sum of absolute differences (SAD) between the target block 110 and the eight reference blocks in the search area 210 every eight computing cycles. In other words, the time required to calculate the sum of absolute differences between the target block 110 and a single reference block is only one computing cycle on average. In addition, there is no delay during the block matching operation, so the best operation efficiency can be achieved.

The aforementioned block pixel difference calculation device 300 loads the pixel values of the same column (row) in the target block 110 in the same calculation cycle, and performs calculations with the pixel values of the same row in the search area 210, which is only the present invention An embodiment of . In practice, the block pixel difference calculation device 300 can also load the pixel values of the same row (column) in the target block 110 and the pixel values of the same row in the search area 210 to perform calculations in the same calculation cycle.

In addition, the block pixel difference computing device 300 of the present invention also has the advantage of supporting target blocks of different sizes. Assuming that the size of the target block is 16×8, as shown in Figure 5, the block pixel difference calculation device 300 only needs to divide the target block 510 in the target frame 500 in Figure 5 into two equal parts with a size of 8 The sub-blocks (sub-blocks) 512 and 514 of X 8 can use the same operation method as described above to perform block matching.

FIG. 6 is a schematic diagram of a data flow of an embodiment in which the block pixel difference computing device 300 matches the target block 510 with multiple reference blocks in the search area 210 of the reference frame 200 .

In the first operation cycle, the first column of pixel data in the sub-block 512, that is, pixels C(1,1), C(2,1), . . . , C(7,1) and C(8 , 1), will be synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, the first 15 pixels in the first column of pixels in the search area 210, that is, pixels R(1,1), R(2,1), . . . , R(14,1) and R(15,1 ), will be synchronously input to all computing units on the corresponding oblique dotted line. In the second operation cycle, the second column of pixel data in sub-block 512, that is, pixels C(1,2), C(2,2), . . . , C(7,2) and C(8 , 2), will be respectively synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, the first 15 pixels in the second column of pixels in the search area 210, that is, pixels R(1,2), R(2,2), . . . , R(14,2) and R(15,2 ), will be synchronously input to all computing units on the corresponding oblique dotted line.

By analogy, in the ninth operation cycle, the first row of pixel data in the sub-block 514, that is, pixels C(9,1), C(10,1), . . . , C(15,1) and C(16, 1) will be synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, in the first column of pixels in the search area 210, 15 pixels starting from pixel R(9,1), that is, pixels R(9,1), R(10,1), ..., R(22 , 1) and R(23, 1) will be synchronously input to all computing units corresponding to the diagonal dotted lines. Next, in the tenth operation cycle, the second row of pixel data in the sub-block 514, that is, the pixels C(9, 2), C(10, 2), . . . , C(15, 2) and C( 16, 2), which will be synchronously input to all computing units corresponding to the horizontal dotted line. At the same time, in the second column of pixels in the search area 210, the 15 pixels starting from the pixel R(9,2), that is, the pixels R(9,2), R(9,2), . . . , R(22 , 2) and R(23, 2) will be synchronously input to all computing units corresponding to the diagonal dotted line.

By analogy, after sixteen operation cycles, the accumulated values in the eight adding units are the target block 510 and the eight consecutive reference blocks in the search area 210 (that is, the reference block RB _{16 x 8} (1, 1), RB _{16 x 8} (2, 1), ..., and RB _{16 x 8} (8, 1)). The time required to calculate the sum of absolute differences between the target block 510 and a single reference block is only two computing cycles on average. In practice, the block pixel difference calculation device 300 can also load the pixel values of the same row (column) in the sub-block 512 or 514 and the pixel values of the same row in the search area 210 to perform the calculation in the same calculation cycle. .

If the size of the target block is 16 X 16, as shown in Figure 7. The block pixel difference calculation device 300 also only needs to divide the target block 710 in the target frame 700 of FIG. block match. FIG. 8 is a schematic diagram of a data flow of an embodiment in which the block pixel difference computing device 300 matches the target block 710 with multiple reference blocks in the search area 210 of the reference frame 200 . Since the operation of the block pixel difference calculation device 300 is substantially the same as that of the foregoing embodiments, it is not repeated here.

In this embodiment, after thirty-two calculation cycles, the values accumulated by the eight addition units of the block pixel difference calculation device 300 are the eight consecutive reference values in the target block 710 and the search area 210 respectively. The sum of absolute differences between the blocks (ie, the reference blocks RB _{16 x 16} (1, 1), RB _{16 x 16} (2, 1), . . . , and RB _{16 x 16} (8, 1)). In other words, the time required to calculate the sum of absolute differences between the target block 710 and a single reference block is only four computing cycles on average. Similarly, in practice, the block pixel difference calculation device 300 can also load the pixel values of the same row (column) in the sub-block 712, 714, 716 or 718, and the same row (column) in the search area 210 in the same calculation cycle. row pixel values to perform operations on.

As can be seen from the above, the block pixel difference computing device 300 of the present invention can process target blocks of different sizes such as 8 X 8, 16 X 8, 8 X 16, 16 X 16, etc. by using the same set of computing unit (PE) matrix , so that the utilization of the circuit can be greatly improved.

In addition, in the foregoing embodiments, the size of the matrix of the computing units of the block pixel difference computing device 300 is 8×8, which is only a preferred embodiment of the present invention, and does not limit the scope of application of the present invention. In practice, it is only necessary to use a 4 X 4 sized computing unit matrix, or even a 2 X 2 sized computing unit matrix, to realize the aforementioned block matching operations of different sizes.

The above descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope of the patent of the present invention.

Claims (9)

1. A block matching device, comprising: A plurality of computing modules, each computing module is used to calculate the pixel difference between multiple target pixels in a target block and multiple reference pixels in a reference block, wherein each computing module includes multiple computing units, The computing unit is used to calculate the pixel difference between one of the target pixels and one of the reference pixels; and A plurality of addition units are respectively coupled to these operation modules, each addition unit is used to sum up the calculation results of a plurality of operation units in the corresponding operation module, and each addition unit is used to accumulate the calculation results of the corresponding operation module Computational results over multiple computing cycles. 2. The block matching device as claimed in claim 1, wherein one of the target pixels is synchronously input to a plurality of first computing units among the computing units. 3. The block matching device according to claim 2, wherein the first computing units belong to different computing modules. 4. The block matching device as claimed in claim 1, wherein each operation unit is used to calculate an absolute difference between one of the target pixels and one of the reference pixels. 5. The block matching device as claimed in claim 1, wherein the target pixels are located in the same column or row in the target block. 6. The block matching device as claimed in claim 1, wherein the reference pixels are located in the same column or row in the reference block. 7. A block pixel difference calculation method, used to calculate the difference between a target block in a target frame and a first reference block and a second reference block in a reference frame, the target block includes There is a first pixel and a second pixel, the first reference block includes a first reference pixel and a second reference pixel, the second reference block includes the second reference pixel and a third reference pixel, the The block pixel difference operation method includes: calculating the first pixel and the first reference pixel to obtain a first difference value; calculating the first pixel and the second reference pixel to obtain a second difference value; calculating the second pixel and the second reference pixel to obtain a third difference value; calculating the second pixel and the third reference pixel to obtain a fourth difference value; summing the first difference value and the second difference value; and Summing up the third difference value and the fourth difference value. 8. The block pixel difference calculation method according to claim 7, wherein each calculation step is performed synchronously. 9. The block pixel difference calculation method according to claim 7, wherein each calculation step is to calculate the absolute difference between pixels.

Images