CN101815219B - Mobile evaluation method for real-time embedded multimedia design - Google Patents

Abstract

The invention provides a method for executing mobile evaluation. The method comprises the following steps: selecting a current block in a current image; acquiring motion vectors and residual value data of a plurality of adjacent blocks around the current block; setting a preset critical value according to the residual value data of a plurality of adjacent blocks; comparing the current block with an initial reference block in a reference image to acquire an initial comparison result, and comparing the preset critical value with the initial comparison result; determining a predicted motion vector of the current block according to the motion vectors of a plurality of adjacent blocks if the initial comparison result is more than the preset critical value; and comparing blocks in a searching window range corresponding to the predicted motion vector so as to find a corresponding reference block matched with the current block.

Description

The mobile evaluation method of real-time embedded multimedia design

Technical field

The invention relates to a kind of method of assessing in order to carry out movement, refer to especially a kind of mobile assessment manner of execution that reduces memory span and frequency range.

Background technology

Along with the application of multimedia technology is more and more welcome, the requirement of video compression technology is also more and more important.Many video compression technology standards are suggested one after another, and at present the main flow specification has MPEG-4 and H.264/AVC.The basic principle of these standards is mainly the data of removing unnecessary (redundancy) in the view data, with the storage area of reduction image or the transmission quantity of image.Mobile assessment (MotionEstimation) is a considerable part in the Video coding, and it utilizes the similitude between continuous pictures to remove data repeatability (temporal redundancy) in time, and reaches the purpose of data compression.

Fig. 1 is for moving the schematic diagram of the block alignment algorithm of often taking in the assessment.Be picture size that the present picture (current frame) 100 of W * H is divided into the block size and is a plurality of blocks of N * N at first.Then, setting size in reference picture (reference frame) 110 (for example last picture or next picture) is (N+SR _H-1) * (N+SR _V-1) search window (search window) 112, and in search window 112, find the block 114 the most similar to a present block (current block) 104 in the present picture 100.Then, calculate two blocks 104 and 114 s' difference and motion-vector 120, remove the data of repetition by only transmitting difference and motion-vector 120, this step is exactly mobile assessment.In other words, the purpose of mobile assessment is that motion-vector and the error of finding out each block in the present picture represents present picture.Yet because mobile assessment needs the many candidate block of comparison, this high operand will cause memory band width significantly to increase.

The hardware structure of Fig. 2 display video coded system 200, wherein reference picture and present picture are to be stored in external memory storage 220, and the required data of mobile assessment are then passed through external bus 230 loading internal memories 212 for computing engines (such as flush bonding processor) 214.Therefore, when carrying out mobile assessment, in order to carry out the comparing computing, required candidate block data will be by externally memory 220 and 212 transfers of internal storage of external bus 230 in the search window of reference picture, and significantly increase memory band width.Generally speaking, the size of search window 112 is to decide according to standards such as screen resolution and/or compliant compressions.Search window 112 is larger, needs the data volume of loading internal memory also more, and required memory band width is also larger.Therefore, need to provide a kind of too high mobile assessment manner of execution of memory band width demand that solves.

Summary of the invention

In view of the existing problem of prior art, the present invention is directed to square search (square search) algorithm a kind of MPEG-4 of being applicable to and low-power H.264/AVC and dynamical method for video coding are provided, can significantly reduce memory span and frequency range, and then reduce hardware cost, reduce the electric weight use and speed execution speed.

According to an aspect of the present invention, provide a kind of in order to carry out the method for mobile assessment, comprised following steps: selected present block in present picture; Obtain the motion-vector and the residual value data that are positioned at present block a plurality of adjacent block on every side; Residual value data setting predetermined critical according to a plurality of adjacent block; Compare present block and the initial reference block in reference picture and obtain initial comparison result, and relatively predetermined critical and initially comparison result; If initial comparison result then determines the prediction motion-vector of present block greater than predetermined critical according to the motion-vector of a plurality of adjacent block; And in the search window scope of correspondence prediction motion-vector, carry out the block comparison, to seek the corresponding reference block that is complementary with present block.

According to a further aspect in the invention, provide a kind of computer fetch medium, in order to stored program instructions, wherein when program command is executed on the calculation element, will make calculation element carry out above-mentioned method.

Other side of the present invention, part will be stated in follow-up explanation, and part can be learnt in illustrating easily, or can be learnt by embodiments of the invention.Each aspect of the present invention can be utilized specifically noted element and combination in the appended claim and understand and reach.Need to understand, aforesaid summary of the invention and following detailed description be usefulness for example only all, is not to limit the present invention.

Description of drawings

Graphic is to be combined with this specification and to consist of its part, in order to embodiments of the invention to be described, and together with specification in order to explain principle of the present invention.Embodiment described herein is preferred embodiment of the present invention, yet, must understand configuration and the element of the present invention shown in being not limited to, wherein:

Fig. 1 is for using the block alignment algorithm to move the schematic diagram of assessment;

Fig. 2 shows the hardware structure of a video coding system;

Fig. 3 is that carrying out with square search algorithm of one embodiment of the invention moved the schematic diagram of assessing;

Fig. 4 uses square search to move the schematic diagram of assessment for describing;

Fig. 5 shows an example that uses raster scan;

Fig. 6 show needle is utilized framework to the grade A of reference picture to four kinds of Data duplications of D; And

Fig. 7 shows that the execution of one embodiment of the invention moves the method flow diagram of assessment.

[main element label declaration]

100 present pictures

104 present blocks

110 reference pictures

112 search windows

114 blocks

212 internal storages

214 computing engines

220 external memory storages

230 buses

300 present pictures

302 present blocks

310 reference pictures

312,314 blocks

320 search windows

402,404,412,414,416 blocks

422,424,426,428,430 blocks

405 square figures

500 pictures

510,511,512,513,514 blocks

610,620 search windows

612,614 blocks

622,624 blocks row

630,640 reference pictures

632,634 search windows

642,644 search windows row

Embodiment

The dynamic assessment method that the present invention is directed to square search (square search) algorithm and cooperate Data duplication to utilize framework to propose effectively to reduce memory band width and reduce internal storage (on-chip memory) demand, its according to and adjacent block between spatially dependence, dynamically adjust the size of search window, replace the known dynamic assessment method that needs to load the monoblock search window.In order to make narration of the present invention more detailed and complete, can and cooperate with reference to following description that Fig. 3's to Fig. 7 is graphic.Device, element and method step described in right following examples in order to the present invention to be described, is not to limit the scope of the invention only.

Fig. 3 is that carrying out with square search algorithm of one embodiment of the invention moved the schematic diagram of assessing, it is the present block 302 of N * N for size in the present picture 300, frame goes out a search window 320 around the position of corresponding at present block 302 in reference picture 310, to find out the block the most similar to present block 302 in search window 320.The comparison method that adopts in this embodiment is for calculating the sad value of each candidate block in present block 302 and the search window 320, and its account form is as follows:

$\mathrm{SAD}={\mathrm{\Σ}}_{i=0}^{N-1}{\mathrm{\Σ}}_{j=0}^{N-1}\left|(C_{\mathrm{ij}}-R_{\mathrm{ij}})\right|$

Cij represents present block, and Rij represents a candidate block.In other words, the intensity of each pixel in the intensity of each pixel in the present block and the candidate block is subtracted each other, just can obtain sad value to the absolute value addition of resulting N * N difference again.Sad value is less, and it is more similar to represent two blocks.But it should be noted that, in this embodiment, though be as the similarity degree of judgement with present block 302 with sad value, only mode is not limited to this, other comparison mode also is applicable to the present invention such as mean square error (mean square error) or mean absolute error (mean absolute error) etc.

Embodiment shown in Figure 3 uses the mode of square search to find out the block the most similar to present block 302 from reference picture 310, and wherein square search algorithm is in mobile evaluation process, in the search window scope, uses square figure to compare.Fig. 4 uses square search to move the schematic diagram of assessment for describing.At first, in step 1, in reference picture take the block 402 of the present block of correspondence position as starting point, centered by block 402 by 9 square figures 405 that block was consisted of in find the block (being the block of sad value minimum) that conforms to most, the hypothesis block that conforms to most is block 404 in this example.Then, in step 2, point centered by block 404 uses square pattern to continue to look for the block that conforms to most again, need be written into newly-increased block 412,414 this moment, reach 416 data.The block of supposing to conform to most in the step 2 is block 416, then then in step 3, centered by block 416, is written into newly-increased block 422,424,426,428, and 430 and repeat above-mentioned comparison operation.The block of supposing to conform to most in the step 3 is block 428, then follows in step 4, repeats aforesaid operations centered by block 428.Block is block 428 if the block position that conforms to most in the step 4 at the central point of square figure, even conforms to most, then can stop the square search to this present block.

Get back to Fig. 3, in this embodiment, at first begin comparison from the block 312 of corresponding block 302 positions, and centered by block 312 by 9 zones that block was consisted of in seek the block (being the block of SAD minimum) that conforms to most.For instance, can use as shown by the arrows in Figure 3 the path sequentially to compare for 9 blocks.Suppose in 9 blocks that block 314 has minimum sad value, then centered by block 314, repeat above-mentioned steps, until the central point of block at 9 blocks comparing that conform to most.

All blocks in the picture all can carry out above-mentioned mobile appraisal procedure at present, in reference picture, to find out respectively the most close corresponding block, and the order of carrying out mobile assessment will affect a certain particular block when carrying out mobile assessment, and which block around it had been carried out mobile assessment.For instance, Fig. 5 shows an example that uses grating (raster) scanning, in this example, scans from left to right, from top to bottom all blocks in the picture 500.Therefore, in the time will doing mobile assessment for a certain block (such as block 510), its left (511), upper left side (512), top (513), and the block in upper right side (514) program of all having carried out mobile assessment, namely the motion-vector of these adjacent block and corresponding sad value thereof are all known.The adjacent block related data cooperation of passing through to obtain and the spatial coherence (spatial correlation) between the adjacent block, the motion-vector of measurable present block also can be in order to set the SAD critical value of present block dynamically to adjust the search window scope.

Take embodiment shown in Figure 3 as example, prediction search window scope and the predetermined critical of block 302 all can obtain according to the comparison result of adjacent block at present, are described below.At first, obtain motion-vector and the comparison data of the rear gained of adjacent block comparison, wherein compare the sad value (hereinafter referred to as residual value data (residual data)) that data are its coupling block that searches, the predetermined critical of then setting present block is as follows:

α _n＝(2×LEFT _SAD+2×TOP _SAD+TOP-RIGHT _SAD+TOP-LEFT _SAD+ε)/6

LEFT wherein _SADRepresent the residual value data of present block left side block, TOP _SADRepresent the residual value data of present block top block, TOP-RIGHT _SADRepresent the residual value data of present block upper right corner block, TOP-LEFT _SADRepresent the residual value data of present block upper left corner block, ε represents the constant factor of present block.ε is the penalty coefficient that can finely tune, and it is chosen as the application of the rule of thumb, and the designer can adjust according to practical application.In this embodiment, consider the distance of distance between block, give 2 times weight with the left side and top block, so in other embodiments, the weight of each adjacent block can adjust according to practical application.Should be noted, the present invention is not limited to use raster scan order as shown in Figure 5, other similarly is that zigzag (zigzag) scanning sequency also is applicable to the present invention, but should be noted that different scanning sequencies will affect a certain block can obtain the mobile assessment result of which adjacent block as the usefulness of prediction search window scope.

In the embodiments of figure 3, when moving assessment for block 302, load at first first the block 312 of the position of corresponding block 302 in the reference picture 310 in internal storage, then compare block 302 and block 312 and calculate its sad value.If sad value is less than predetermined critical α _n, can finish the mobile assessment (motion-vector is (0,0)) of block 302, carry out the mobile assessment of next block.Thus, when the film of static state, memory band width (Memory Bandwidth) only can be reduced to and need about 11%.

If the sad value between block 302 and the block 312 is greater than predetermined critical α _n, then then load the search window scope predicted to carry out follow-up comparison.The prediction of search window scope can utilize the adjacent block related data that obtains to cooperate and adjacent block between spatial coherence (spatialcorrelation) decide.In this embodiment, utilize the motion-vector of upper left corner block 1, top block 2, upper right corner block 3 and the left side block 4 of present block, dope the motion-vector of present block.At first, the central point respective coordinates that defines present block is (0,0), and the central point respective coordinates of upper left corner block 1, top block 2, upper right corner block 3 and the left side block 4 of block is respectively (16 at present, 16), (0,16), (16,16), (16,0).Then, according to least squares method, utilize the motion-vector of upper left corner block 1, top block 2, upper right corner block 3 and left side block 4 to ask for the suitableeest plane of regression (regression plane): z=c-ax-by with relevant coordinate, to predict the motion-vector of present block.Coordinate and motion-vector substitution with four adjacent block can get:

$E=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{((c-{\mathrm{ax}}_i-{\mathrm{by}}_i)-{\mathrm{MV}}_i)}^2$

Get respectively the partial differential of a, b, c:

$\∂E/\∂a=0;$

$\∂E/\∂b=0;$

$\∂E/\∂c=0;$

By above 3 formulas, can obtain a, b, c numerical value is as follows:

a＝1/32(MV ₁-MV ₃)；

b＝1/96(-5MV ₁-2MV ₂+MV ₃+6MV ₄)；

c＝1/2(-MV ₁+MV ₃+2MV ₄)；

With a, b, c numerical value and present block coordinate (0,0) substitution, namely measurable present block motion-vector is:

MV＝1/2(-MV ₁+MV ₃+2MV ₄)

Motion-vector and its corresponding square graphics field according to the present block of predicting are measurable search window scope, so can a download section divide the required data of using, and avoid downloading the data of known monoblock search window.After loading the search window after dynamically adjusting, the mode with above-mentioned square search in search window is compared.To sum up state, the present invention only loads the data of the corresponding reference picture in home position of the block of processing at the beginning, and how many data volumes just dynamically adjust after the comparison needs to load to internal storage, and namely the size of search window is that capable of dynamic determines.Therefore, the present invention can reduce the data volume of required loading internal memory, not only can reduce time and the consumed power of transfer of data, also can reduce required internal storage size and reduces hardware cost.

Except the method for usage data prediction, in storage management, the present invention is application data recycling framework also, be temporary in the internal storage by the data that will reuse, and the number of times that reduction memory access and data shift.In other words, after the reusing of analyzing data, avoid some data of repeated access by adding internal storage, and then reduce the memory band width demand.Utilize the associated description of framework for Data duplication, can with reference to by people such as D.X.Li in IEEE Trans.ConsumerElectron., vol.53, no.3, pp.1053-1060, " architecture design (Architecture Design forH.264/AVC Integer Motion Estimation with Minimum Memory Bandwidth) with H.264/AVC integrated moving assessment of minimized memory frequency range " delivered among the Aug.2007, by people such as J.C.Tuan in IEEE Trans.Circuits Syst.Video Technol., vol.12, no.1, pp.61-72, " the searching Data duplication utilization and the memory band width analysis (On the data reuse and memory bandwidth analysis forfull-search block-matching VLSI architecture) of block matching VLSI framework fully " of delivering among the Jan.2002, by people such as C.Y.Chen in IEEETrans.Circuits Syst.Video Technol., vol.16, no.4, pp.553-558, " being used for having the mobile grade C+ Data duplication use framework (LevelC+data reuse scheme for motion estimation with corresponding coding orders) of assessing of corresponding coded sequence " of delivering among the Apr.2006, and, by people such as T.C.Chen in IEEE Trans.Circuits Syst.Video Technol., vol.17, no.2, pp.242-247, " the multiple reference picture H.264/AVC moves the multiple present macro zone block framework of single reference picture (Single Reference Frame Multiple CurrentMacroblocks Scheme for Multiple Reference Frame Motion Estimation inH.264/AVC) of assessment " of delivering among the Feb.2007, content will be incorporated this paper into as a reference on it.

Data duplication utilizes the usefulness of framework to be assessed by following two factors: the size of internal storage and redundant access parameter Ra, wherein internal storage can use and the temporary required memory size of reference data for Data duplication in order to expression, redundant access parameter Ra then can be in order to assess the external memory storage frequency range, and it is defined as follows:

The degree that Data duplication utilizes is lower, and the Ra value is larger, and needs more memory band widths, otherwise the degree that Data duplication utilizes is higher, and the Ra value is less, and required memory band width is fewer.Total memory frequency range BW can be expressed as follows:

BW=f * W * H * (Ra _{Present picture})+f * W * H * (Ra _{Reference picture})

Wherein f is frame updating speed, and W is picture width, and H is picture width.

In general, memory band width is to depend on frame updating speed (frame rate), picture size, search window size, and Ra value etc., and for specific video compression applications, frame updating speed and picture size are generally fixed value, therefore the present invention is by selecting the less Data duplication of Ra value to utilize framework and usage data Forecasting Methodology to reduce the size of search window, and then effectively with the low memory frequency range.

Concerning present picture, on average each block can be accessed SR _HX SR _VInferior, namely

But as long as the adding size is the internal storage of N * N, just the Ra of present picture can be reduced to 1, as follows:

And concerning reference picture, Fig. 6 show needle is utilized framework to the grade A of reference picture to four kinds of Data duplications of D, and its bend is the data for reusing partly.Grade A and B are respectively the recycling of the data in single search window 610,620, and grade C and D are the recycling in the data of different search windows.In detail, be the block of N * N pixel to the size in present picture, grade A is that the size in the recycling reference picture is (N+SR _H-1) * (N+SR _V-1) candidate block 612 that two horizontal directions in the single search window 610 are continuous and the pixel of 614 overlappings, grade B then is two continuous row's candidate block 622 of the vertical direction of recycling in search window 620 and the pixel of 624 overlappings.Grade C be recycling in reference picture 630 two continuous blocks of horizontal direction the pixel of each self-corresponding search window 632 and 634 overlappings, and grade D two row's blocks that to be recycling vertical direction in reference picture 640 continuous the pixel of each self-corresponding search window 642 and 644 overlappings.As above-mentioned, the total memory frequency range depends on Ra, and grade A can be calculated as follows to the Ra of D framework:

Grade A:

Grade B:

Grade C:

Grade D:

On the other hand, as seen from Figure 6, grade A is as follows to the required internal storage size of D framework:

The recycling grade	The internal storage size
		A	N×(N-1)
B	(N+SR H-1)×(N-1)
		C	(SR H-1)×(SR V+N-1)

D	W×(SR V-1)

From the above, the internal storage size is less, and the demand of memory band width is larger (such as grade A) then, otherwise though the memory band width demand of grade D framework can significantly reduce, relatively needs larger internal storage size.

Grade C can obtain better balance in internal storage size and external memory storage frequency range, but along with the lifting of video resolution, the Data duplication of grade C uses framework not apply the needs of practical application.Therefore, the present invention combines Data duplication and uses and two kinds of functions of data prediction, in grade C framework, predict and dynamically adjust the search window size of present block according to the search window size of adjacent block, replace script grade C framework and need load the separately overlapping area of corresponding monoblock search window of adjacent two continuous blocks of horizontal direction, and only need load the search window scope of its corresponding prediction.So, not only can effectively reduce its internal storage size requirements, and the required memory band width of C framework that can further downgrade.

Fig. 7 shows that the execution of one embodiment of the invention moves the method flow diagram of assessment.In general, when the execution block is compared mobile assessment algorithm, can first present picture segmentation be become a plurality of blocks, and determine a plurality of onblock executing are moved the scanning sequency of assessment.In this embodiment, adopt raster scan order sequentially each block to be moved assessment, and the square search algorithm of mobile assessment algorithm system's employing of each block.At first, in step S700, select wherein a block moving assessment, and obtain the dynamic scan window ranges of its adjacent block, the motion vector prediction value of adjacent block and the comparison data of adjacent block.Then, in step S710, according to the predetermined critical α of the present block of residual value data definition of adjacent block _n, for instance, predetermined critical α _nCan be the numerical value of the residual value data of adjacent block being got addition gained after the different weights.Should be noted predetermined critical α _nCalculating can select to do suitable adjustment according to practical application is required.Then, in step S720, load the corresponding at present data of block position in the reference picture, and carry out once true comparison calculation with present block, to calculate its sad value.

Then, in step S730, the predetermined critical α that comparison step S710 sets _nAnd the sad value of step S720 gained.If in step S730, the sad value of comparison gained is less than predetermined critical α _n, then program proceeds to step S740, finishes the mobile assessment of this block, and then proceeds to step S750, all finishes mobile assessment to judge whether all blocks.If judge that at step S750 all blocks all finish mobile assessment, then program proceeds to step S760, finishes the mobile assessment of present picture, if judge not to be that all blocks are all finished mobile assessment at step S750, then program is got back to step S700, selects next block to continue to carry out mobile assessment.

In step S730, if sad value is greater than predetermined critical α _n, then program proceeds to step S770, utilizes the motion vector prediction of known adjacent block to go out the motion-vector of present block.In this embodiment, predictor formula is according to motion-vector and the coordinate of upper left side, top, upper right side and left block, asks for plane of regression by least squares method and obtains.Should be noted that predictor formula may adjust with the difference of scanning sequency.Then, in step S780, the square Search Area of the present block motion-vector that correspondence is predicted is from external memory storage loading internal memory.Then, in step S790, in the search window that loads, with square search algorithm selected onblock executing is moved assessment, to seek the block that mates most.Find the block that mates most in step S790 after, program is got back to step S750 to repeat above-mentioned steps, until finish the mobile assessment of present picture.Should be noted, if in step S790, can't find with square search the block of coupling, the block that then can in the search window of prediction, select to have minimum sad value as the coupling block of present block or can load other or larger search window scope with the operation of again comparing, so the present invention does not limit feasible in the case mode.

The present invention utilizes the residual value data (residual data) of adjacent block and the result who truly compares for the first time, reduce first required memory band width and internal storage size, and the motion vector prediction of block goes out the search window scope of present block around utilizing.The present invention replaces grade C framework need download the search area of adjacent two blocks of horizontal direction, and only need to download the search window scope of predicting.Therefore, the present invention only need to use 30%～60% of internal storage originally, and it is about 40%～80% that memory band width only needs, and reduces significantly internal storage and memory band width.Proposed by the invention utilize the mobile evaluation method of framework in conjunction with data prediction and Data duplication, can be applicable in the software and hardware storage management of various real-time embedded multimedia system.

The above is preferred embodiment of the present invention only, is not to limit claim scope of the present invention; All other do not break away from the equivalence of finishing under the disclosed spirit and changes or modification, all should be included in the above-mentioned claim scope.

Claims (8)

1. one kind in order to carry out the method for mobile assessment, comprises following steps:

A. in present picture, select present block;

B. obtain the motion-vector and the residual value data that are positioned at this present block a plurality of adjacent block on every side;

C. according to this residual value data setting predetermined critical of this a plurality of adjacent block;

D. compare this present block with initial reference block in reference picture and obtain initial comparison result, and relatively this predetermined critical and this initial comparison result;

If e. this initial comparison result is greater than this predetermined critical, then determine the prediction motion-vector of this present block according to this motion-vector of these a plurality of adjacent block; And

F. in to the search window scope that should predict motion-vector, carry out block and compare, with the corresponding reference block of seeking and this present block is complementary,

Wherein this predetermined critical is set as follows:

α _n＝(2×LEFT _SAD+2×TOP _SAD+TOP-RIGHT _SAD+TOP-LEFT _SAD+ε)/6；

LEFT wherein _SADBe the residual value data of this present block left side adjacent block, TOP _SADBe the residual value data of this present block top adjacent block, TOP-RIGHT _SADBe the residual value data of this present block upper right corner adjacent block, TOP-LEFT _SADBe the residual value data of this present block upper left corner adjacent block, ε is constant coefficient.

2. method according to claim 1, wherein this initial reference block position in this reference picture is to the present position of block in this present picture.

3. method according to claim 1, wherein step f uses square search algorithm to carry out the block comparison.

4. method according to claim 1 is wherein in steps d, if this initial comparison result is less than this predetermined critical, then take this initial reference block as this correspondence reference block.

5. method according to claim 4 also comprises for all block repeating step a to f in this present picture, and wherein step a selects this present block with grating scanning mode.

6. method according to claim 1, wherein this initial comparison result is the function of the absolute error sum total between this initial reference block and this present block.

7. method according to claim 1 also comprises following steps:

G. from an external memory storage this search window scope is loaded, and this search window scope is stored in the internal storage;

H. determine that the horizontal direction of the present block that this is selected is to the corresponding prediction search window of right continuous block scope; And

I. the newly-increased prediction search window scope that step h is loaded compared to step g, this external memory storage is loaded in this internal storage certainly.

8. method according to claim 1, wherein this prediction motion-vector of this present block is:

MV＝1/2(-MV ₁+MV ₃+2MV ₄)；

MV wherein ₁Motion-vector, MV for this present block upper left corner adjacent block ₃Motion-vector, MV for this present block upper right corner adjacent block ₄Motion-vector for this present block left side adjacent block.

Images