CN102263944A - Video coding system, mobile estimation device and calculation method for mobile estimation - Google Patents
Video coding system, mobile estimation device and calculation method for mobile estimation Download PDFInfo
- Publication number
- CN102263944A CN102263944A CN2010101970219A CN201010197021A CN102263944A CN 102263944 A CN102263944 A CN 102263944A CN 2010101970219 A CN2010101970219 A CN 2010101970219A CN 201010197021 A CN201010197021 A CN 201010197021A CN 102263944 A CN102263944 A CN 102263944A
- Authority
- CN
- China
- Prior art keywords
- candidate
- motion
- location searching
- vector
- cost
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a mobile estimation device which is applied to a video coding system. The mobile estimation device comprises a first calculation module and a second calculation module. When the calculation of the mobile estimation moves to a second candidate searching position from a first candidate searching position along one search path, the first calculation module estimates a first motion vector cost difference according to one piece of search path information corresponding to the search path; and the second calculation module selectively adds or subtracts the first motion vector cost difference from an original motion vector cost value to obtain the first motion vector cost value corresponding to the second candidate searching position.
Description
Technical field
The present invention is relevant with video coding, especially, need not utilize complicated multiplying can promptly estimate corresponding to the video coding system of the motion-vector cost function of different candidate's location searchings, mobile estimating device and How It Works thereof relevant for a kind of.
Background technology
In video coding system (video encoding system), move estimation (motion estimation) technology in order to the motion-vector (motionvector) of an image in the decision video streaming with respect to its contiguous image.In many video coding systems (for example with H.264 or the system of MPEG-4 protocol-compliant), can adopt mobile estimation algorithm to assist to find out dependence between different pictures usually, and then reach application such as image compression or image content analysis.
Generally speaking, present image processing technique mainly sees through the method for block comparison and finds out optimal motion-vector, represents the relative displacement of zone (comparison form) between two different pictures of a certain desire comparison.See also Fig. 1 (A), Fig. 1 (A) illustrates the schematic diagram of motion-vector V in the prior art.Shown in Fig. 1 (A), motion-vector V can be used for representing between previous picture and the present picture, the level of corresponding particular block and the vector of vertical displacement (as the vectorial V among the figure one).In actual applications, the most normal block matching difference that is used (block-matching discrepancy) computational methods for calculate absolute value difference sum (Sum of Absolute Differences, SAD).
In the evaluation method of various motion-vectors, a kind of at last preferable evaluation method of motion-vector cost function (motion vector costfunction), its formula is as follows:
Motion-vector cost function=A * (mvbits[cx-px]+mvbits[cy-py])
Wherein A is scale factor (scaling factor); [cx, cy] is the motion-vector of candidate's location searching (searchcandidate); [px, py] is for estimating the motion-vector of position (predictor); Mvbits[cx-px] be the bit estimated value of the x durection component difference of above-mentioned two motion-vectors; Mvbits[cy-py] then be the bit estimated value of the y durection component difference of above-mentioned two motion-vectors.With mvbits[k] be example, its bit number and | the corresponding relation between the k| is shown in Fig. 1 (B).
Please refer to Fig. 1 (C), Fig. 1 (C) has illustrated the circuit framework of traditional motion-vector cost function generator.Shown in Fig. 1 (C), motion-vector cost function generator 3 comprises first subtrator 30, second subtrator 31, the first motion-vector bit evaluation unit 32, the second motion-vector bit evaluation unit 33, adder unit 34 and multiplication unit 35.
At first, first subtrator 30 is according to the motion-vector [cx, cy] and the x component cx of the motion-vector [px, py] of estimating the position and the difference [cx-px] that px calculates both of candidate's location searching; Second subtrator 31 is according to the motion-vector [cx, cy] and the y component cy of the motion-vector [px, py] of estimating the position and the difference [cy-py] that py calculates both of candidate's location searching.Then, the first motion-vector bit evaluation unit 32 and the second motion-vector bit evaluation unit 33 calculate motion-vector bit mvbits[cx-px according to [cx-px] respectively] and according to [cy-py] calculating motion-vector bit mvbits[cy-py].Afterwards, adder unit 34 is with mvbits[cx-px] and mvbits[cy-py] addition and obtain (mvbits[cx-px]+mvbits[cy-py]).At last, multiplication unit 35 multiplies each other scale factor A and (mvbits[cx-px]+mvbits[cy-py]) and the motion-vector cost function that obtains being asked again.
Yet the disadvantage of the circuit framework of above-mentioned traditional motion-vector cost function generator 3 is to carry out complicated multiplying by the multiplication unit 35 of higher-order, thereby causes problems such as time restriction and logic gate number increase.
Therefore, main purpose of the present invention is to provide a kind of video coding system, mobile estimating device and How It Works thereof, to address the above problem.
Summary of the invention
According to first specific embodiment of the present invention, be a kind of mobile estimating device.In this embodiment, this moves estimating device and is applied in the video coding system.This mobile estimating device comprises first computing module and second computing module.When the calculating of moving estimation along a search path from first candidate's location searching (x
1, y
1) move to second candidate's location searching (x
2, y
2) time, this first computing module goes out the first motion-vector cost difference according to the search path data-evaluation corresponding to this search path.This second computing module optionally adds an original motion-vector value at cost or deduct this first motion-vector cost difference according to a preset rules, to obtain corresponding to this second candidate location searching (x
2, y
2) one first motion-vector value at cost.Wherein, x
1And x
2Be respectively this first candidate location searching and the pairing trunnion axis coordinate values of this second candidate location searching; y
1And y
2Be respectively this first candidate location searching and the pairing vertical axis coordinate values of this second candidate location searching.
According to second specific embodiment of the present invention, be a kind of computational methods of mobile estimation.In this embodiment, this method is applied in the video coding system.At first, when the calculating of moving estimation along a search path from first candidate's location searching (x
1, y
1) move to second candidate's location searching (x
2, y
2) time, this method goes out the first motion-vector cost difference according to the search path data-evaluation corresponding to this search path.Then, this method optionally adds an original motion-vector value at cost or deduct this first motion-vector cost difference according to a preset rules, to obtain corresponding to this second candidate location searching (x
2, y
2) one first motion-vector value at cost.Wherein, x
1And x
2Be respectively this first candidate location searching and the pairing trunnion axis coordinate values of this second candidate location searching; y
1And y
2Be respectively this first candidate location searching and the pairing vertical axis coordinate values of this second candidate location searching.
In sum, because mobile estimating device of the present invention does not need can be rapidly and estimate the motion-vector cost function exactly by the multiplying of complexity, so problems such as time restriction, logic gate number increase and the mobile estimation efficient that can effectively avoid prior art to suffer from are not good, especially when this moved estimating device and carries out the mobile estimation of a plurality of candidate's location searchings simultaneously, it is more remarkable that it promotes the effect that moves estimation efficient.
Can be further understood by the following detailed description and accompanying drawings about the advantages and spirit of the present invention.
Description of drawings
Fig. 1 (A) has illustrated the schematic diagram of motion-vector V in the prior art; Fig. 1 (B) has illustrated mvbits[k] the bit number and | the corresponding relation between the k|; Fig. 1 (C) has illustrated the circuit framework of traditional motion-vector cost function generator.
Fig. 2 has illustrated the functional block diagram according to the mobile estimating device of first specific embodiment of the present invention.
Fig. 3 (A) has illustrated information provides the information between the module and first computing module to transmit situation; Fig. 3 (B) has illustrated the detailed functions calcspar of first computing module; Fig. 3 (C) has illustrated information provides the information between the module and second computing module to transmit situation; The information that Fig. 3 (D) has illustrated between first computing module, second computing module and the 3rd computing module is transmitted situation.
Fig. 4 has illustrated the detailed functions calcspar of second computing module.
Fig. 5 (A)～Fig. 5 (C) has illustrated the search path of different types.
Fig. 6 (A) and Fig. 6 (B) have illustrated the search path that sees through the pattern of crawling and have searched the schematic diagram of scanning; Fig. 6 (C) has illustrated mark (pixel) and has moved an example of estimation.
Fig. 7 has illustrated the flow chart according to the mobile estimating device How It Works of second specific embodiment of the present invention.
The main element symbol description
S10～S17: process step
1: mobile estimating device 10: information provides module
14: the second computing modules of 12: the first computing modules
31: the second subtrators of 30: the first subtrators
34: adder unit 35: multiplication unit
3: motion-vector cost function generator
32: the first motion-vector bit evaluation units
33: the second motion-vector bit evaluation units
140: the first multiplexer 141:x component processors
143: the second multiplexers of 142:y component processor
122,144: adder 145: motion-vector bit evaluation unit
146: subtracter 147: motion-vector cost difference computational unit
V: motion-vector MVC
0: original motion-vector value at cost
A: scale factor Δ MVC
1: the first motion-vector cost difference
C: middle position H:1/2 candidate location searching
Q:1/4 candidate location searching
[px, py]: corresponding to the motion-vector of estimating the position
[cx, cy]: the motion-vector of candidate's location searching
[cx
0, cy
0]: corresponding to first candidate's location searching (x
1, y
1) motion-vector
[cx
1, cy
1]: corresponding to second candidate's location searching (x
2, y
2) motion-vector
MVC
1: corresponding to second candidate's location searching (x
2, y
2) the motion-vector value at cost
120: motion-vector bit generator 124: accumulator
1202: the second subtracters of 1200: the first subtracters
1204: the first motion-vector bit evaluation units
1206: the second motion-vector bit evaluation units
Embodiment
Main purpose of the present invention is to propose a kind of the need and can promptly estimates corresponding to the video coding system of the motion-vector cost function of different candidate's location searchings, mobile estimating device and How It Works thereof by the multiplying of complexity.
According to first specific embodiment of the present invention, be a kind of mobile estimating device.In this embodiment, this moves estimating device and applies in the video coding system.Please refer to Fig. 2, Fig. 2 has illustrated the functional block diagram of this mobile estimating device.As shown in Figure 2, mobile estimating device 1 comprises information provides module 10, first computing module 12, second computing module 14, the 3rd computing module 16 and storage module 18.Wherein, information provides module 10 to be coupled to first computing module 12, second computing module 14 and storage module 18; First computing module 12 and second computing module 14 all are coupled to the 3rd computing module 16.
In this embodiment, when the calculating of moving estimation as yet not from first candidate's location searching (x
1, y
1) begin to carry out along a search path before, first computing module 12 will calculate earlier corresponding to first candidate's location searching (x
1, y
1) original motion-vector value at cost MVC
0Shown in Fig. 3 (A), suppose to be expressed as [cx corresponding to the motion-vector of first candidate's location searching
0, cy
0] and be expressed as [px, py] corresponding to the motion-vector of estimating the position, information provides module 10 that [cx will be provided
0, cy
0], information to the first computing module 12 such as [px, py] and scale factor A, make the computing module 12 of winning obtain original motion-vector value at cost MVC according to those information calculations
0
It should be noted that, for the multiplier that improves prior art employing higher-order carries out complicated multiplying, cause problems such as time restriction and logic gate number increase, first computing module 12 goes out original motion-vector value at cost MVC by the accumulator computes that only comprises a plurality of adder units
0, directly do not adopt the multiplier of high-order to carry out complicated multiplying, use the cost of saving circuit element effectively.Please refer to Fig. 3 (B), Fig. 3 (B) has illustrated the detailed functions calcspar of first computing module 12.Shown in Fig. 3 (B), first computing module 12 comprises motion-vector bit generator 120, adder 122 and accumulator 124.Wherein, motion-vector bit generator 120 comprises first subtracter 1200, second subtracter 1202, the first motion-vector bit evaluation unit 1204 and the second motion-vector bit evaluation unit 1206; Accumulator 124 comprises a plurality of adder units.
When first subtracter 1200 receives cx
0When reaching px, be sent to the first motion-vector bit evaluation unit, 1204, the first motion-vector bit evaluation units 1204 after first subtracter 1200 subtracts each other it and produce the first motion-vector bit more according to this; When second subtracter 1202 receives cy
0When reaching py, be sent to the second motion-vector bit evaluation unit, 1206, the second motion-vector bit evaluation units 1206 after second subtracter 1202 subtracts each other it and produce the second motion-vector bit more according to this.Then, when adder 122 respectively when the first motion-vector bit evaluation unit 1204 and the second motion-vector bit evaluation unit 1206 receive the first motion-vector bit and the second motion-vector bit, adder 122 is sent to accumulator 124 after with the first motion-vector bit and the second motion-vector bit addition, and accumulator 124 sees through the result of calculation that its a plurality of adder units obtain multiply by routine factors A.
It should be noted that at this, though first computing module 12 calculates original motion-vector value at cost MVC
0The time, still need multiply by scale factor A, yet, because scale factor A only is a constant, so need not seeing through the multiplier of high-order, first computing module 12 do not carry out complicated multiplying, only need to realize multiply by the simple multiplication computing of a constant, and then try to achieve original motion-vector value at cost MVC by a plurality of adder units that accumulator 124 is comprised
0
When first computing module 12 calculates corresponding to first candidate's location searching (x
1, y
1) original motion-vector value at cost MVC
0After, the calculating of moving estimation can begin respectively each candidate's location searching of process on the way to be carried out in regular turn along this search path.Certainly, before the calculating of moving estimation began to carry out, this search path needed to be determined earlier, in order to follow-up mobile estimation.In fact, this search path can be the systemic presupposition value or be set by the user.
Suppose to move the calculating of estimation by first candidate's location searching (x
1, y
1) move to second candidate's location searching (x along this search path
2, y
2), and corresponding to second candidate's location searching (x
2, y
2) motion-vector be expressed as [cx
1, cy
1], shown in Fig. 3 (C), information provides module 10 that [cx will be provided
1, cy
1], [px, py] and give second computing module 14 corresponding to the search path information of this search path.By this, second computing module 14 can go out the first motion-vector cost difference DELTA MVC according to those information calculations
1According to the present invention, in searching the process of scanning, usually can very big (for example pixel) owing to move distance that the calculating of estimation moves each time, and be generally moving of horizontal direction or vertical direction, therefore, second computing module 14 calculates the first motion-vector cost difference DELTA MVC
1Load and little, so can effectively save system's calculation resources.
Then, shown in Fig. 3 (D), because first computing module 12 has calculated corresponding to first candidate's location searching (x
1, y
1) original motion-vector value at cost MVC
0, and second computing module 14 has calculated the first motion-vector cost difference DELTA MVC
1, therefore, the 3rd computing module 16 can be with MVC
0With Δ MVC
1Addition obtains corresponding to second candidate's location searching (x
2, y
2) motion-vector value at cost MVC
1It should be noted that the 3rd computing module 16 according to a preset rules optionally with MVC
0Add or deduct Δ MVC
1To obtain MVC
1, and this preset rules is estimated first candidate's location searching (x by the 3rd computing module 16 according to this search path information
1, y
1) and this second candidate location searching (x
2, y
2) between coordinate values difference determine that is the poor (x of x coordinate of this preset rules and two location searchings
2-x
1) and the poor (y of y coordinate
2-y
1) be that plus or minus is relevant.
Next, how put up with second computing module 14 according to [cx
1, cy
1], [px, py] and corresponding to the search path information calculations first motion-vector cost difference DELTA MVC of this search path
1Carry out more deep discussion.Please refer to Fig. 4, Fig. 4 has illustrated the detailed functions calcspar of second computing module 14.As shown in Figure 4, second computing module 14 comprises first multiplexer 140, x component processor 141, y component processor 142, second multiplexer 143, adder 144, motion-vector bit evaluation unit 145, subtracter 146 and motion-vector cost difference computational unit 147.
In this embodiment, first multiplexer 140 in order to reception information provide that module 10 imported corresponding to the motion-vector of estimating the position [px, py] and search path information, and output [px, py] is to motion-vector bit evaluation unit 145.As for information provide that module 10 imported corresponding to second candidate's location searching (x
2, y
2) motion-vector [cx
1, cy
1], its x component cx
1With y component cy
1After passing through the processing of x component processor 141 and y component processor 142 respectively, will be transferred into second multiplexer 143, and second multiplexer 143 will be exported MV
1[cx
1, cy
1] to motion-vector bit evaluation unit 145.It should be noted that between second multiplexer 143 and the motion-vector bit evaluation unit 145 to be provided with adder 144, in order to according to search path information in MV
1Add a displacement (displacement), make it become MV according to the mobile situation of reality
1'.In fact, MV
1' may be [cx
1', cy
1], [cx
1, cy
1'] or [cx
1', cy
1'], look closely the distance size of each moved further in the different search paths and decide.For example, if search path is serpentine path, then MV
1' should be [cx
1', cy
1] or [cx
1, cy
1']; If search path is diamond-shaped path, then MV
1' [cx may appear
1', cy
1'].
Then, motion-vector bit evaluation unit 145 is promptly respectively according to different MV
1And MV
1' the generation first motion-vector bit and the second motion-vector bit.Afterwards, subtracter 146 deducts the second motion-vector bit the first motion-vector bit again to produce a motion-vector bit difference.At last, motion-vector cost difference computational unit 147 calculates the first motion-vector cost difference DELTA MVC that is asked according to this motion-vector bit difference
1
In addition, pattern about above-mentioned search path there is no certain restriction, as Fig. 5 (A) to shown in Fig. 5 (C), the pattern of search path can be serpentine path (serpentine path), helical-like path (spiralpath) and diamond-shaped path (diamond path), but not as limit.In fact, these have the search path of different types and corresponding search path information thereof (for example search path all location searchings of process and searching sequence etc.) on the way and can be stored in the storage module 18, when second computing module 14 needs these information when carrying out computing, information provide module 10 promptly can to the required information of storage module 18 accesses to offer second computing module 14.
Next, will crawl a concrete instance that search path searches scanning to all pixels on the picture as an illustration with employing.Please refer to Fig. 6 (A) and Fig. 6 (B), Fig. 6 (A) and Fig. 6 (B) have illustrated the search path that sees through the pattern of crawling and have searched the schematic diagram of scanning.Shown in Fig. 6 (A), suppose that comparison zone (comparison form) is that figure bend part and the coordinate of estimating the position are (0,0), search path system is begun according to (2 by first candidate's location searching (2 ,-2),-1), (2,0) ..., (2,4), (1,4), (1,3) ... order crawl always and search till (4,4).
Before the calculating of moving estimation did not also really begin along search path to carry out from (2 ,-2), first computing module 12 will be earlier according to [cx
0, cy
0], [px, py] and scale factor A calculate the original motion-vector value at cost MVC corresponding to (2 ,-2)
0
In this example, [cx
0, cy
0]=[-2 ,-2] and [px, py]=[0,0].Then, when the calculating of moving estimation moved to (2 ,-1) along search path by (2 ,-2), second computing module 14 promptly can be according to [cx
1, cy
1], [px, py] and go out the first motion-vector cost difference DELTA MVC corresponding to the search path information calculations of this search path
1In this example, [cx
1, cy
1]=[-2 ,-1] and [px, py]=[0,0], that is cy is only arranged
1With cy
0Difference, and cx
1Then with cx
0Identical, so needn't seeing through complex calculations, second computing module 14 can obtain Δ MVC
1Afterwards, the 3rd computing module 16 can be according to preset rules optionally with MVC
0Add or deduct Δ MVC
1To obtain MVC
1In this example, when moving to (2 ,-1) owing to the calculating of moving estimation oneself (2 ,-2), y
1-y
0=(1)-(2)=1＞0, representative are moved the calculating of estimation and are moved toward positive direction on the y direction, so the 3rd computing module 16 will be with MVC
0Add Δ MVC
1To obtain MVC
1In like manner, the calculating of moving estimation is also similar by (2 ,-1) situation when moving to (2,0) along search path, does not give unnecessary details separately at this.
Shown in Fig. 6 (B), when the calculating of moving estimation continued to desire to move to (1,4) by (2,4) along search path, situation can be different.Suppose that in the resultant motion-vector cost corresponding to (2,4) of previous step be MVC
6And its [cx
6, cy
6]=[-2,4], second computing module 14 promptly can be according to [cx
7, cy
7], [px, py] and go out the 7th motion-vector cost difference DELTA MVC corresponding to the search path information calculations of this search path
7In this example, [cx
7, cy
7]=[-1,4] and [px, py]=[0,0], that is only be cx
7With cx
6Difference, and cy
7Then with cy
6Identical, so needn't seeing through complex calculations, second computing module 14 can obtain Δ MVC
7Afterwards, the 3rd computing module 16 can be according to preset rules optionally with MVC
6Add or deduct Δ MVC
7To obtain MVC
7In this example, when moving to (1,4) owing to the calculating of moving estimation oneself (2,4), x
7-x
6=(1)-(2)=1＞0, representative are moved the calculating of estimation and are moved toward positive direction on the x direction, so the 3rd computing module 16 will be with MVC
6Add Δ MVC
7To obtain MVC
7
In actual applications, mobile estimating device according to the present invention is except being applied to mobile estimation (the Integer Motion Estimation of the integer (pixel) as the foregoing description, IME) outside, also can be applicable to mobile estimation (Factional Motion Estimation, situation FME) of mark (pixel).For example, shown in Fig. 6 (C), in the mobile estimation of mark (pixel), candidate's location searching can be divided into middle position C, 1/2 candidate's location searching H and 1/4 candidate's location searching Q.If mobile estimating device is positioned at middle position C, the calculating of then next moving estimation can see through and once move to those 1/2 candidate location searchings H.As for those 1/4 candidate location searchings Q, the calculating of moving estimation need see through twice and move and could move to those 1/4 candidate location searchings Q from middle position C, or is to see through once to move to move to those 1/4 candidate location searchings Q from those 1/2 candidate location searchings H.
The computational methods that second specific embodiment according to the present invention is a kind of mobile estimation.In this embodiment, this method is applied in the video coding system.Please refer to Fig. 7, Fig. 7 has illustrated this and has moved the flow chart of estimation computational methods.As shown in Figure 7, at first, in step S10, before the calculating of moving estimation did not begin to carry out as yet, this method will be calculated earlier corresponding to first candidate's location searching (x
1, y
1) original motion-vector value at cost MVC
0In fact, MVC
0According to motion-vector [cx corresponding to first candidate's location searching
0, cy
0], corresponding to information calculations such as the motion-vector of estimating the position [px, py] and scale factor A and get.It should be noted that, for the multiplier that improves prior art employing higher-order carries out complicated multiplying, cause problems such as time restriction and logic gate number increase, this method only needs can realize multiply by the simple multiplication computing of a constant by a plurality of adder units that accumulator comprised, and then tries to achieve original motion-vector value at cost MVC
0, directly do not adopt complicated multiplying.
Then, in step S11, the calculating of moving estimation begins along a search path from first candidate's location searching (x
1, y
1) move to second candidate's location searching (x
2, y
2).Then, this method execution in step S12 goes out the first motion-vector cost difference DELTA MVC according to the search path data-evaluation corresponding to this search path
1In fact, the first motion-vector cost difference DELTA MVC
1According to search path information, corresponding to second candidate's location searching (x
2, y
2) motion-vector [cx
1, cy
1] and estimated corresponding to the motion-vector of estimating the position [px, py].Wherein, x
1And x
2Be respectively this first candidate location searching and the pairing trunnion axis coordinate values of this second candidate location searching; y
1And y
2Be respectively this first candidate location searching and the pairing vertical axis coordinate values of this second candidate location searching.
In fact, this search path can be a serpentine path, a helical-like path or a diamond-shaped path, and can be the systemic presupposition value or set by the user, there is no certain restriction.In this embodiment, suppose that this search path is in regular turn through first candidate's location searching (x
1, y
1), second candidate's location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3), and this search path information comprises first candidate's location searching (x
1, y
1), second candidate's location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3) coordinate values.
Then, in step S13, this method according to a preset rules optionally with original motion-vector value at cost MVC
0Add or deduct the first motion-vector cost difference DELTA MVC
1, to obtain corresponding to second candidate's location searching (x
2, y
2) motion-vector value at cost MVC
1In practical application, this preset rules and this first candidate location searching (x
1, y
1) and this second candidate location searching (x
2, y
2) the coordinate values difference of asking relevant, that is the poor (x of x coordinate of this preset rules and two location searchings
2-x
1) and the poor (y of y coordinate
2-y
1) be that plus or minus is relevant.
Calculating corresponding to second candidate's location searching (x
2, y
2) motion-vector value at cost MVC
1After, this method execution in step S14 judges whether second candidate's location searching is the terminal point of search path.If the judged result of step S14 is for being that the terminal point that the calculating of estimating has moved to search path is moved in representative, so the calculation procedure of whole mobile estimation is finished; If the judged result of step S14 is not, this method execution in step S15, the calculating of moving estimation is from second candidate's location searching (x
2, y
2) continue to move to the 3rd candidate's location searching (x along this search path
3, y
3).
At this moment, this method will execution in step S16, goes out one second motion-vector cost difference DELTA MVC according to the search path data-evaluation
2Wherein, x
3And y
3Be respectively pairing trunnion axis coordinate values of the 3rd candidate's location searching and vertical axis coordinate values.Then, in step S17, this method optionally will be corresponding to second candidate's location searching (x according to this preset rules
2, y
2) motion-vector value at cost MVC
1Add or deduct the second motion-vector cost difference DELTA MVC
2, to obtain corresponding to the 3rd candidate's location searching (x
3, y
3) the second motion-vector value at cost MVC
2
Next, this method will judge whether the 3rd candidate's location searching is the terminal point of search path, if on behalf of search path, judged result do not finish as yet for not, this method promptly can repeat above-mentioned steps, till the evaluation work of the mobile estimation of whole search path is finished.By this, see through this method and can obtain motion-vector value at cost corresponding to each the candidate's location searching in the search path, so can judge according to this have minimum motion-vector value at cost candidate's location searching why.
In sum, because mobile estimating device of the present invention does not need to carry out complicated multiplying by the high-order multiplier, can estimate the motion-vector cost function rapidly and exactly, so can effectively avoid time restriction, logic gate number that prior art suffers to increase and move problems such as estimation efficient is not good.Especially when mobile estimating device carried out the mobile estimation of a plurality of candidate's location searchings simultaneously, it is more remarkable that it promotes the effect that moves estimation efficient.In addition, because the distance that mobile estimating device moves in searching scanning process each time not quite and greatly mostly is moving of level or vertical direction usually, therefore, the load of mobile estimating device calculating motion-vector cost difference is also little, so can effectively save system's calculation resources.
By the above detailed description of preferred embodiments, hope can be known description feature of the present invention and spirit more, and is not to come category of the present invention is limited with the above-mentioned preferred embodiment that is disclosed.On the contrary, its objective is that hope can contain in the category of claim of being arranged in of various changes and tool equality institute of the present invention desire application.
Claims (13)
1. a mobile estimating device is applied in the video coding system, and this moves estimating device and comprises:
First computing module, when this calculating of moving estimation along a search path from first candidate's location searching (x
1, y
1) move to second candidate's location searching (x
2, y
2) time, this first computing module goes out the first motion-vector cost difference according to the search path data-evaluation corresponding to this search path; And
Second computing module is coupled to this first computing module, in order to optionally an original motion-vector value at cost is added or deducts this first motion-vector cost difference according to a preset rules, to obtain corresponding to this second candidate location searching (x
2, y
2) the first motion-vector value at cost;
Wherein, x
1And x
2Be respectively this first candidate location searching and the pairing trunnion axis coordinate values of this second candidate location searching, y
1And y
2Be respectively this first candidate location searching and the pairing vertical axis coordinate values of this second candidate location searching.
2. mobile estimating device as claimed in claim 1, it is characterized in that, further comprise the 3rd computing module, be coupled to this first computing module and this second computing module, the 3rd computing module only is made of adder, in order to calculate corresponding to this first candidate location searching (x
1, y
1) this original motion-vector value at cost.
3. mobile estimating device as claimed in claim 1 is characterized in that, this preset rules and this first candidate location searching (x
1, y
1) and this second candidate location searching (x
2, y
2) between coordinate values difference relevant, this first candidate location searching (x
1, y
1) and this second candidate location searching (x
2, y
2) between coordinate values difference estimate according to this search path information.
4. mobile estimating device as claimed in claim 1 is characterized in that, this search path is a serpentine path, a helical-like path or a diamond-shaped path.
5. mobile estimating device as claimed in claim 1 is characterized in that, when the calculating of moving estimation from this second candidate location searching (x
2, y
2) continue to move to the 3rd candidate's location searching (x along this search path
3, y
3) time, this first computing module goes out one second motion-vector cost difference according to this search path data-evaluation, this second computing module optionally adds this first motion-vector value at cost or deduct this second motion-vector cost difference according to this preset rules, to obtain corresponding to the 3rd candidate's location searching (x
3, y
3) second motion-vector value at cost, the wherein x
3And y
3Be respectively pairing trunnion axis coordinate values of the 3rd candidate's location searching and vertical axis coordinate values, this preset rules system and this second candidate location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3) between coordinate values difference relevant.
6. mobile estimating device as claimed in claim 5 is characterized in that, this search path information comprises this search path this first candidate location searching (x of process in regular turn
1, y
1), this second candidate location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3) coordinate values.
7. mobile estimating device as claimed in claim 5, it is characterized in that, comprise a storage module, in order to store this search path information, this original motion-vector value at cost, this first motion-vector cost difference, this first motion-vector value at cost, this second motion-vector cost difference and this second motion-vector value at cost.
8. the method for estimation is moved in a calculating one, and this method is applied in the video coding system, and this method comprises the following step:
When the calculating of moving estimation along a search path from first candidate's location searching (x
1, y
1) move to second candidate's location searching (x
2, y
2) time, go out the first motion-vector cost difference according to a search path data-evaluation corresponding to this search path; And
Optionally an original motion-vector value at cost is added or deducts this first motion-vector cost difference according to a preset rules, to obtain corresponding to this second candidate location searching (x
2, y
2) the first motion-vector value at cost;
X wherein
1And x
2Be respectively this first candidate location searching and the pairing trunnion axis coordinate values of this second candidate location searching; y
1And y
2Be respectively this first candidate location searching and the pairing vertical axis coordinate values of this second candidate location searching.
9. method as claimed in claim 8 is characterized in that, this original motion-vector value at cost is corresponding to this first candidate location searching (x
1, y
1) and this original motion-vector value at cost only see through add operation and get.
10. method as claimed in claim 8 is characterized in that, this preset rules and this first candidate location searching (x
1, y
1) and this second candidate location searching (x
2, y
2) between coordinate values difference relevant.
11. method as claimed in claim 8 is characterized in that, this search path is a serpentine path, a helical-like path or a diamond-shaped path.
12. method as claimed in claim 8 is characterized in that, further comprises the following step:
When the calculating of moving estimation from this second candidate location searching (x
2, y
2) continue to move to the 3rd candidate's location searching (x along this search path
3, y
3) time, go out the second motion-vector cost difference according to this search path data-evaluation; And
Optionally this first motion-vector value at cost is added or deducts this second motion-vector cost difference according to this preset rules, to obtain corresponding to the 3rd candidate's location searching (x
3, y
3) the second motion-vector value at cost;
X wherein
3And y
3Be respectively pairing trunnion axis coordinate values of the 3rd candidate's location searching and vertical axis coordinate values, this preset rules and this second candidate location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3) between coordinate values difference relevant.
13. method as claimed in claim 12 is characterized in that, this search path information comprises this search path this first candidate location searching (x of process in regular turn
1, y
1), this second candidate location searching (x
2, y
2) and the 3rd candidate's location searching (x
3, y
3) coordinate values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010197021.9A CN102263944B (en) | 2010-05-27 | 2010-05-27 | Video coding system, mobile estimation device and calculation method for mobile estimation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010197021.9A CN102263944B (en) | 2010-05-27 | 2010-05-27 | Video coding system, mobile estimation device and calculation method for mobile estimation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102263944A true CN102263944A (en) | 2011-11-30 |
| CN102263944B CN102263944B (en) | 2014-07-02 |
Family
ID=45010376
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010197021.9A Expired - Fee Related CN102263944B (en) | 2010-05-27 | 2010-05-27 | Video coding system, mobile estimation device and calculation method for mobile estimation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102263944B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103458255A (en) * | 2012-05-31 | 2013-12-18 | 晨星软件研发(深圳)有限公司 | Image coding method and related device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040064036A1 (en) * | 2002-09-26 | 2004-04-01 | Zuhua Mao | Methods and systems for motion tracking |
| TW200516991A (en) * | 2003-11-11 | 2005-05-16 | Mediatek Inc | Method and related apparatus for motion estimation |
| CN1829325A (en) * | 2005-03-01 | 2006-09-06 | 凌阳科技股份有限公司 | Method and system for dynamically adjusting motion estimation |
| TW200635378A (en) * | 2005-03-21 | 2006-10-01 | Pixart Imaging Inc | Improved diamond search and dynamic estimation method |
| CN101715131A (en) * | 2008-09-30 | 2010-05-26 | 原相科技股份有限公司 | Block movement estimation method |
-
2010
- 2010-05-27 CN CN201010197021.9A patent/CN102263944B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040064036A1 (en) * | 2002-09-26 | 2004-04-01 | Zuhua Mao | Methods and systems for motion tracking |
| TW200516991A (en) * | 2003-11-11 | 2005-05-16 | Mediatek Inc | Method and related apparatus for motion estimation |
| CN1829325A (en) * | 2005-03-01 | 2006-09-06 | 凌阳科技股份有限公司 | Method and system for dynamically adjusting motion estimation |
| TW200635378A (en) * | 2005-03-21 | 2006-10-01 | Pixart Imaging Inc | Improved diamond search and dynamic estimation method |
| CN101715131A (en) * | 2008-09-30 | 2010-05-26 | 原相科技股份有限公司 | Block movement estimation method |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103458255A (en) * | 2012-05-31 | 2013-12-18 | 晨星软件研发(深圳)有限公司 | Image coding method and related device |
| CN103458255B (en) * | 2012-05-31 | 2016-04-06 | 晨星软件研发(深圳)有限公司 | The method of image coding and relevant apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102263944B (en) | 2014-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5044568B2 (en) | Motion estimation using predictive guided decimation search | |
| US8139634B2 (en) | Methods and apparatuses of fast cross motion estimation for video encoding | |
| CN104067619A (en) | Video decoder, video encoder, video decoding method, and video encoding method | |
| JP2007502053A (en) | Motion estimation method and apparatus | |
| CN102075760A (en) | Quick movement estimation method and device | |
| JP6394876B2 (en) | Encoding circuit and encoding method | |
| CN104754180A (en) | Method and device for reducing video frame jitter | |
| WO2019072246A1 (en) | Motion estimation method and device in video coding, electronic apparatus and storage medium | |
| CN109348234A (en) | A kind of efficient sub-picture element movement estimating method and system | |
| CN101860747B (en) | Sub-pixel movement estimation system and method | |
| US20060285597A1 (en) | Reusing interpolated values in advanced video encoders | |
| EP3596698B1 (en) | Motion estimation method and apparatus for plurality of frames | |
| CN106658024B (en) | A kind of quick method for video coding | |
| CN102263944B (en) | Video coding system, mobile estimation device and calculation method for mobile estimation | |
| CN112261413A (en) | Video encoding method, encoding device, electronic device, and storage medium | |
| TWI451767B (en) | Video encoding system, motion estimation apparatus, and motion estimation method | |
| CN100366092C (en) | Search method for video frequency encoding based on motion vector prediction | |
| KR100490730B1 (en) | Method and apparatus for motion estimation using of adaptive search pattern for video sequence compression | |
| CN109889851A (en) | Block matching method, device, computer equipment and the storage medium of Video coding | |
| CN113115038B (en) | Motion estimation method and device, electronic equipment and readable storage medium | |
| CN115546515A (en) | Depth information acquisition method and device | |
| JP6390275B2 (en) | Encoding circuit and encoding method | |
| CN202931463U (en) | Characteristic block based video image stabilization device | |
| WO2012028480A1 (en) | Global motion vector estimation | |
| Tao et al. | Motion estimation based on an improved block matching technique |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140702 Termination date: 20190527 |