CN109391821A - Method for estimating and movement estimation apparatus - Google Patents
Method for estimating and movement estimation apparatus Download PDFInfo
- Publication number
- CN109391821A CN109391821A CN201710650632.6A CN201710650632A CN109391821A CN 109391821 A CN109391821 A CN 109391821A CN 201710650632 A CN201710650632 A CN 201710650632A CN 109391821 A CN109391821 A CN 109391821A
- Authority
- CN
- China
- Prior art keywords
- motion vector
- block
- current block
- present frame
- candidate
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/513—Processing of motion vectors
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/57—Motion estimation characterised by a search window with variable size or shape
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/573—Motion compensation with multiple frame prediction using two or more reference frames in a given prediction direction
Abstract
A kind of means for motion estimation and method.Movement estimation apparatus includes Candidate Set generation circuit and motion vector decision circuitry.The motion vector of Candidate Set generation circuit selection adjacent block is added to the motion vector candidate collection of the current block of present frame.One of current block and adjacent block belong to identical row (or identical column), and one of described adjacent block is the block of present frame or the block of previous frame.Motion vector decision circuitry is coupled to Candidate Set generation circuit to receive motion vector candidate collection.Motion vector decision circuitry judges the motion vector of the current block in present frame according to motion vector candidate collection.
Description
Technical field
The present invention is suitable for parallel about a kind of method for estimating and exercise estimator, and particularly with regard to one kind
The method for estimating and exercise estimator of processing.
Background technique
Estimation (motion estimation) is at frame-rate conversion (frame rate up conversion, FRC)
In be considerable technology.Estimation, which refers to, finds best or the second optimal motion vectors (Motion Vector).Usually
Picture frame (image frame) may be divided into most blocks (blocks).Motion vector may indicate that the movement of block
Direction and move distance.Usual consecutive frame (neighboring frames) shows strong correlation in the time and space
Property, and usually there is similar or identical part to each other.3-d recursive search (three is presented in the relevant technologies
Dimension recursive, 3DRS) block adaptation, there is the upper left for being set as current block (current block)
With two candidate motion vectors of the optimal motion vectors of upper right adjacent block, wherein corresponding convergence direction or search side
To being from left to right and from top to bottom.
Fig. 1 is please referred to, Fig. 1 is the schematic diagram of the Candidate Set of 3-d recursive search (3DRS) estimation traditionally.
Traditional 3DRS method for estimating is calculated by the motion vector of Candidate Set is used at present frame (current frame)
In current block B0 motion vector.Fig. 1 is please referred to, the block B0 of present frame has Candidate Set, including in the current frame, when
The spatial candidate motion vector Sa of the upper left side block 104 of preceding block B0 works as proparea in previous frame (previous frame)
The spatial candidate motion vector Sb of the upper right side block 106 of block B0, in previous frame, the time candidate motion vector of block 108
Ta, in previous frame, the time candidate motion vector Tb of block 110 and in previous frame, the time Candidate Motion of block 102
Vector T 0.It note that the block 102 of previous frame and the block B0 in present frame are identical position.The block 102 of previous frame,
108,110 and the block 104,106 of present frame there is previous respective motion estimation process.As motion vector T0, Ta and Tb
Referred to as time motion vector (because its Correlation block be set in previous frame) when, motion vector Sa, Sb are referred to as space fortune
Moving vector (because block 104 and 106 is set in frame identical with block B0).
Fig. 2 is before illustrating to (forward) motion estimation algorithm and backward (backward) motion estimation algorithm.It please be same
When referring to Fig.1 and Fig. 2, from the respective motion estimation process of the block of the correlation of the present frame with reference frame to be transported
Moving vector Sa, Sb.It can be the previous frame (example in forward motion estimation according to design and use demand, the reference frame
Frame F-1 in this way) or future frame (e.g. frame F+1) in backward motion estimation.Reverse estimation obtains motion vector
Mvb and propulsion estimation acquisition motion vector mvf.In the same manner, from reference frame F-2 (forward motion estimation) or reference
In the previous frame F-1 of frame F (backward motion estimation), respective associated block is to obtain motion vector T0, Ta, Tb.
Fig. 3 is the motion vector candidate collection of traditional 3DRS estimation.In addition to motion vector as shown in Figure 1, from frame F,
Except the block of F-1, the motion vector candidate of block B0 further includes the big random file of spatial motion vector Sa, Sb and small
Random file, and further include zero motion vector or global motion vector (global motion vector, GMV).
Candidate spatial motion vector is from not in same a line of current block B0 or the block of same row.This will be for row
Or the parallel processing between column causes difficulty.Since the motion estimation process for current block B0 can not be activated, until tool
There is candidate spatial motion vector (i.e. block 104,106) to complete estimation to be coupled, and acquisition respective motion vector Sa,
Sb.Since candidate motion vector especially spatial motion vector is from the block divided with current block B0, this can be used for
To obtain accurate motion vector when current block B0, while significantly reducing convergence rate.For the complexity fortune in an image
It is dynamic, in traditional 3DRS technology, it will lead to lower convergence rate and by motion vector prediction error.
Summary of the invention
The present invention provides a kind of movement estimation apparatus with faster convergence rate and more accurate motion vector
With method.
An embodiment according to the present invention provides a kind of for estimating the estimation side of the motion vector of the block of present frame
Method.The method for estimating includes: the fortune that current block of the motion vector of multiple adjacent blocks to present frame is added in selection
Moving vector Candidate Set, wherein one of them of the current block and those adjacent blocks belong to same row or same a line, and described
One of them of those adjacent blocks is a block of the present frame or previous frame;And it determines according to motion vector candidate collection
The motion vector of the current block of settled previous frame.
An embodiment according to the present invention provides a kind of estimation dress for estimating the motion vector of the block of present frame
It sets.Movement estimation apparatus includes that Candidate Set generation circuit and motion vector determine circuit.The selection of Candidate Set generation circuit is added
The motion vector of multiple adjacent blocks to present frame current block motion vector candidate collection, wherein current block and these
One of them of adjacent block belongs to same row or same a line, and one of them of these adjacent blocks is present frame or elder generation
One block of previous frame.Motion vector determines that circuit is coupled to Candidate Set generation circuit, to receive motion vector candidate collection.Movement
Vector determines that circuit can determine the motion vector of the current block of present frame according to motion vector candidate collection.
An embodiment according to the present invention provides a kind of for estimating the estimation side of the motion vector of the block of present frame
Method.The method for estimating includes: to select for the motion vector of the current block of previous frame to be added to present frame to work as proparea
The motion vector candidate collection of block;And according to motion vector candidate collection come determine present frame current block motion vector.
An embodiment according to the present invention provides a kind of estimation dress for estimating the motion vector of the block of present frame
It sets.Movement estimation apparatus includes that Candidate Set generation circuit and motion vector determine circuit.The selection of Candidate Set generation circuit will be first
The motion vector of the current block of previous frame is added to the motion vector candidate collection of the current block of present frame.Motion vector determines electricity
Road is coupled to Candidate Set generation circuit, to receive motion vector candidate collection.Motion vector determines that circuit can be according to motion vector
Candidate Set determines the motion vector of the current block of present frame.
Based on above-mentioned, motion vector candidate collection that movement estimation apparatus described in the embodiment of the present invention and its method are introduced
Different from motion vector candidate collection used in existing 3DRS technology, so that movement estimation apparatus described in the embodiment of the present invention has
There are faster convergence rate and more accurate predicted motion vector.
To make the foregoing features and advantages of the present invention clearer and more comprehensible, special embodiment below, and it is detailed to cooperate attached drawing to make
Carefully it is described as follows.
Detailed description of the invention
Fig. 1 is the schematic diagram of the Candidate Set of 3-d recursive search (3DRS) estimation traditionally.
Fig. 2 is to illustrate propulsion estimating algorithm and reverse estimating algorithm.
Fig. 3 is the motion vector candidate collection of traditional 3DRS estimation.
Fig. 4 is the schematic diagram for illustrating exercise estimator according to one embodiment of the invention.
Fig. 5 is the flow chart for illustrating step according to one embodiment of the invention.
Fig. 6 is the motion vector candidate collection for illustrating the current block of present frame according to one embodiment of the invention.
Fig. 7~Figure 10 is the schematic diagram for illustrating Candidate Set according to one embodiment of the invention.
Figure 11 is the Candidate Set CSe for illustrating a kind of current block according to one embodiment of the invention, indicate search direction SD with
And according to from top to bottom and the processing sequence of left-to-right.
Figure 12 be illustrate according to one embodiment of the invention it is a kind of the fourth line parallel processing the case where.
Figure 13~Figure 16 is to illustrate a kind of schematic diagram for indicating to search direction according to one embodiment of the invention.
Figure 17 is the schematic diagram for illustrating the block corresponding to spatial candidate motion vector according to one embodiment of the invention.
Figure 18 is the circuit block diagram for illustrating movement estimation apparatus according to another embodiment of the present invention.
Figure 19 is the flow chart for illustrating method for estimating according to another embodiment of the present invention.
Figure 20 is the schematic diagram illustrated according to the present invention relative to an embodiment of Fig. 6.
Figure 21 is the schematic diagram that one embodiment of the invention illustrates motion vector candidate collection CS.
Figure 22 is the flow chart for illustrating method for estimating according to another embodiment of the present invention.
Figure 23 is the flow chart for illustrating the parallel processing process of method for estimating according to another embodiment of the present invention.
Figure 24 is the schematic diagram that one embodiment of the invention illustrates frame-rate conversion (frame rate up conversion).
Figure 25 is the flow chart for illustrating method for estimating according to one embodiment of the invention.
Figure 26 is the example for projecting (projected) motion estimation process.
Figure 27 is the process estimated for the project motion of another embodiment.
Figure 28 is to indicate to select forward motion vector and backward motion vectors using as candidate motion vector.
Figure 29 is to illustrate that the possibility Candidate Set of the current block Bc of frame includes from the same block position of current block Bc
The adjacent block of present frame shown in the multiple time motion vectors and Figure 20 of more than one consecutive frame and previous frame to
Amount.
Figure 30 is the possibility Candidate Set CS for backward motion estimation processing.
Figure 31 is the possibility Candidate Set for forward motion estimation method.
Figure 32 is the source for illustrating Candidate Set CS according to another embodiment of the present invention.
Figure 33 be according to the present invention another embodiment illustrate Candidate Set processing schematic diagram.
Figure 34 is the circuit block diagram for illustrating movement estimation apparatus according to the embodiment of Figure 18.
Figure 35 is the circuit block diagram for illustrating movement estimation apparatus 1800 according to another embodiment of Figure 18.
Figure 36~Figure 41 is the part of Candidate Set CS_f1~CS_p2 in Figure 35 respectively.
Figure 42 is the flow chart for illustrating the method for estimation according to another embodiment of the present invention.
Symbol description
20: exercise estimator
30: program
102、104、106、108、110、410、510、610、710、1202、mvb、mvf、B0、Ba、Bv、Bb、B10、B11、
B20, B21, B30, B40, B41, B50, B51, B60, B70, Bc: block
202,1820: Candidate Set generation circuit
204: searching direction generation circuit
206,1830: motion vector decision circuitry
300~308, S1910, S1920, S2210~S2250, S2510~S2540: step
910,1110,1140: backward motion estimation
920,1120,1150: forward motion estimation
930,1130,1160: project motion estimation
1810: controller
CSa, CSb, CSc, CSd, CSe, CS1~CSn, CS_f1~CS_f2, CS_b1~CS_b2, CS_p1~CS_p2:
Candidate Set
4th row: fourth line
F-2, F-1, F, F+1, Fint, Fint1, Fint2, Fint ': frame
ME1~MEn: motion estimation process
SD: direction is searched
T0~T4, T01~T05, Ta, Tb, S1~S4, Sa, Sb, B1~B4, Bc, MV, mv_f, mv_p, mv_p ', mv_b,
Mv_p ", mv_f ', mv_b ', MV_1~MV_5, MV_p0, MV_b0: motion vector
VS: vision signal
Specific embodiment
" coupling (or connection) " word used in this case specification full text (including claim) can refer to appoint
What direct or indirect connection means.For example, if it is described herein that first device coupling (or connection) is then answered in second device
This be construed as the first device can be directly connected to the second device or the first device can by other devices or
Certain connection means and be coupled indirectly to the second device.In addition, all possible places, use phase in the drawings and embodiments
Component/component/step with label represents same or like part.Identical label is used in different embodiments or uses identical use
Component/component/step of language can be with cross-referenced related description.
Referring to figure 4., Fig. 4 is the schematic diagram for illustrating exercise estimator 20 according to one embodiment of the invention.In order to facilitate saying
It is bright, it is assumed that vision signal (video signal) includes most frames (frames).Wherein, each frame is divided into most areas
Block, these blocks are arranged in multiple rows and multiple column.Each block includes one or more pixels (pixels).Movement
The motion vector (motion vectors) of block of the estimator 20 to estimate each frame.Exercise estimator 20 includes candidate
Collect generation circuit 202, search direction generation circuit 204 and motion vector decision circuitry 206.Candidate Set generation circuit 202 is used
To select Candidate Set associated with present frame and previous frame for each block of present frame.For each of present frame
Block, Candidate Set may include from at least spatial motion vector of the block in present frame and from the block in previous frame
Most time motion vectors.It searches direction generation circuit 204 and searches direction to select according to Candidate Set.Motion vector is sentenced
Deenergizing 206 is to judge the motion vector of the block of present frame according to direction is searched and sentence according to respective Candidate Set
The motion vector of disconnected each block in the current frame.
Fig. 5 is the flow chart according to one embodiment of the invention read-me (procedure) 30.Referring to figure 5., program 30
It can be used to estimate the motion vector of each block.Program 30 in Fig. 5 is mainly the exercise estimator 20 corresponded in Fig. 4
Operation.It note that those skilled in the art should be noted that the method in Fig. 5 may include other intermediate steps or many steps
Suddenly single step can be merged into.Program 30 can comprise the following steps that
Step 300: starting.
Step 302: for each block of present frame, selecting Candidate Set associated with present frame and previous frame.
Step 304: selection search direction associated with each block and Candidate Set.
Step 306: judging the motion vector of present frame according to direction and Candidate Set is searched.
Step 308: terminating.
Motion vector according to the corresponding present frame of program 30 can be estimated.In step s 302, for current
Each block of frame, Candidate Set generation circuit 202 select Candidate Set associated with present frame and previous frame.Each block
Candidate Set may include at least one spatial motion vector from block in the current frame and in previous frame
The majority time motion vector of block.
In the present embodiment, Candidate Set generation circuit 202 is from adjacent with the space of each block of present frame direct neighbor
In block, at least one spatial motion vector is selected.Each spatial neighbors can be configured in the mutually colleague of present frame
Or using as respective block on column.Temporally adjacent block direct neighbor of the Candidate Set generation circuit 202 from previous frame
In reference block, most time motion vectors are selected.Reference block is configured in previous frame, and space corresponds to currently
Other block in frame.That is, the reference block of previous frame and the block out of the ordinary of present frame are on identical position.Institute
State the relationship between at least one spatial neighbors and the most temporally adjacent blocks may include in following at least
One: (a) when on the mutually colleague that an at least spatial neighbors are configured in present frame using as each respective area
Block, the most temporally adjacent blocks are configured in the same column of previous frame using as reference block, and (b) when at least one
When a spatial neighbors are configured in the same row of present frame using as other block, a most temporally adjacent blocks are matched
It is placed in same a line of previous frame using as reference block.
Fig. 6 is the motion vector candidate collection for illustrating the current block of present frame according to one embodiment of the invention.Current block
Bc has the candidate motion vector from four moment adjacent block B1~B4.Wherein, each of current block Bc is configured in
The adjacent region at top, bottom, left side and right side.The motion vector of each of block B1~B4 can be the time move to
Measure (in the motion vector of its block for being configured in previous frame) or spatial motion vector (its block for being assigned in present frame F
In motion vector).In either one or two of block B1~B4 in present frame F, before the motion estimation process of current block Bc
Motion estimation process is completed, to obtain spatial motion vector as the candidate motion vector of current block Bc.
For example, Fig. 7 is the schematic diagram for illustrating Candidate Set according to one embodiment of the invention.Fig. 7 indicates a kind of and works as proparea
The case where Candidate Set of block Bc, the processing sequence about direction is searched and according to search direction SD from left to right.Please join simultaneously
It is selected according to Fig. 6 and Fig. 7, the Candidate Set CSa about current block Bc.Candidate Set CSa includes spatial motion vector S3, time
Motion vector T1, time motion vector T2 and time motion vector T4.Spatial neighbors B3 is adjacent to current block Bc's
Left edge.In previous frame, temporally adjacent block B1, B2 top edges adjacent to reference block 410 and lower section respectively
Edge.In previous frame, right edge of the temporally adjacent block B4 adjacent to reference block 410.It can judge current block
Before the motion vector of Bc, the motion vector S3 and temporally adjacent block of spatial neighbors B3 in the current frame are estimated
Motion vector T1, T2, T4 in B1, B2, B4.Therefore, the candidate motion vector S3 of spatial neighbors B3 and temporally adjacent
Candidate motion vector T1, T2, T4 of block B1, B2, B4 are selected to the Candidate Set CSa in current block Bc.
Fig. 8 is the Candidate Set CSb for illustrating a kind of current block Bc according to direction SD according to one embodiment of the invention, is searched
Seek the case where direction SD is with processing sequence from right to left.Candidate referring to Fig. 6 and Fig. 8, about current block Bc
Collection CSb is selected.Candidate Set CSb includes spatial motion vector S4, time motion vector T1, time motion vector T2 and time
Motion vector T3.Spatial motion vector S4 is adjacent to the right edge in current block Bc.In previous frame, temporally adjacent block
B1, B2 are respectively adjacent to the top edges and lower edge of reference block 510.In previous frame, temporally adjacent block B3 phase
Adjacent to the left edge of reference block 510.In the same manner, the motion vector S4 and temporally adjacent block of spatial neighbors B4
Motion vector T1, T2, T3 of B1, B2, B3 can be used as the candidate motion vector in current block Bc.
Fig. 9 is the Candidate Set CSc for illustrating a kind of current block Bc according to direction SD according to one embodiment of the invention, is searched
Seek the case where direction SD is with processing sequence from top to bottom.Candidate referring to Fig. 6 and Fig. 9, about current block Bc
Collection CSc is selected.Candidate Set CSc include spatial candidate motion vector S1, time candidate motion vector T2, time Candidate Motion to
Measure T3 and time candidate motion vector T4.Spatial neighbors B1 is adjacent to the top edges in current block Bc.Previous
In frame, temporally adjacent block B2 is adjacent to the lower edge in reference block 610.In previous frame, temporally adjacent block B3, B4
Respectively adjacent to the left edge of reference block 610 and right edge.In the same manner, the motion vector S1 of spatial neighbors B1
And motion vector T2, T3, T4 of temporally adjacent block B2, B3, B4 can be used as the candidate motion vector in current block Bc.
Figure 10 is the Candidate Set CSd for illustrating a kind of current block Bc according to one embodiment of the invention, indicates to search direction SD
And according to processing sequence from bottom to up.Referring to Fig. 6 and Figure 10, for the Candidate Set CSd in current block Bc
It is selected.Candidate Set CSd includes spatial candidate motion vector S2, time candidate motion vector T1, time candidate motion vector T3
And time candidate motion vector T4.Spatial neighbors B2 is adjacent to the lower edge in current block Bc.In previous frame,
Temporally adjacent block B1 is adjacent to the top edges in reference block 710.In previous frame, temporally adjacent block B3, B4 difference
Adjacent to the left edge and right edge in reference block 710.In the same manner, the motion vector S2 in spatial neighbors B2
And motion vector T1, T3, T4 in temporally adjacent block B1, B3, B4 can be used as the Candidate Motion in current block Bc to
Amount.
Figure 11 is the Candidate Set CSe for illustrating a kind of current block Bc according to one embodiment of the invention, indicates to search direction SD
And according to from top to bottom and the processing sequence of left-to-right.Referring to Fig. 6 and Figure 11, in current block Bc
Candidate Set CSe is selected.Candidate Set CSe includes spatial candidate motion vector S1, spatial candidate motion vector S3, time candidate fortune
Moving vector T2, time candidate motion vector T4.Spatial neighbors B1, B3 are respectively adjacent to the top edges in current block Bc
And left edge.In previous frame, temporally adjacent block B2 and B4 respectively adjacent to the lower edge in reference block with
And right edge.In the same manner, in motion vector S1, S3 in spatial neighbors B1, B3 and temporally adjacent block B2, B4
Motion vector T2, T4 can be used as the candidate motion vector in current block Bc.
Figure 12 be illustrate according to one embodiment of the invention it is a kind of the fourth line parallel processing the case where so that space adjacent region
The motion vector in motion vector and spatial neighbors B3 in block B2 is selected as the space in current block Bc
Candidate motion vector S2, S3, and in previous frame, the motion vector in temporally adjacent block B1, B4 can be used as current block
Candidate motion vector T1, T4 in Bc.
In the case where Fig. 7 to Figure 12, each corresponding Candidate Set can further include spatial motion vector it is big with
Seat in the plane move with small random file, and it is similar should zero/global motion vector (zero/global motion in 3DRS
Vector) to obtain preferable coverage area.
In step s 304, direction generation circuit 204 is searched according to the Candidate Set for estimating the motion vector of present frame
To select to search direction.An other block and spatial neighbors can be associated in by searching direction.In the present embodiment,
The direction of searching is the direction from corresponding spatial neighbors to a other block.For example, Fig. 7 is please referred to, when
Spatial neighbors B3 adjacent to current block Bc left edge when, search direction SD by search direction generation circuit 204
It is selected from left to right.In the same manner, Fig. 8 is please referred to, when right edge of the spatial neighbors B4 adjacent to current block Bc,
It searches direction SD and is selected from right to left by direction generation circuit 204 is searched.Referring to Fig. 9 and Figure 10, difference base
In spatial neighbors and other current block, select search direction SD.
On the other hand, searching direction generation circuit 204 can choose search direction, and be selected based on search circuit, wait
Selected works generation circuit 202 can select to search direction by direction generation circuit 204 is searched.For example, when search direction
Generation circuit 204 selects the direction of left-to-right as direction is searched, then to be selected to make Candidate Set to generate electricity based on search direction
Road 202 selects Candidate Set CSa (as shown in Figure 7).
In step S306, motion vector decision circuitry 206 is selected by search direction generation circuit 204 and foundation
Direction is searched to judge the motion vector of present frame, and by the Candidate Set generation circuit 202 for each other block
And other motion vector is judged according to the Candidate Set distinctly selected.About present frame, in the search direction selected,
Motion vector decision circuitry 206 can judge the motion vector of present frame by direction generation circuit 204 is searched.For example, asking
Referring to Fig.1 3, due in Fig. 7 search direction and Candidate Set CSa selected, motion vector decision circuitry 206 is sequentially
The motion vector of the block of ground judgement from left to right.When estimation is used for the motion vector of the first row of present frame, motion vector
Decision circuitry 206 judges the motion vector of the block in the first row (that is, from the first block of the first row to first in order
Last capable block).In the same manner, when estimation is used for the motion vector of the second row of present frame, motion vector decision circuitry
206 judge the block on the second row in order motion vector it is (that is, last from the first block of the second row to the second row
One block).
In the same manner, please refer to Figure 14, due in Fig. 8 search direction and Candidate Set CSb selected, then motion vector is sentenced
Deenergizing 206 judges the motion vector of block from right to left in order.When estimation for present frame every a line movement to
When amount, motion vector decision circuitry 206 judges the motion vector of the block in every a line (that is, most from every a line in order
Latter block to every a line the first block).Please refer to Figure 15, due in Fig. 9 search direction and Candidate Set CSc it is selected
It selects, then motion vector decision circuitry 206 judges the motion vector of block from top to bottom in order.When estimation is for present frame
When the motion vector of first row, motion vector decision circuitry 206 judges the motion vector of the block on first row (also in order
I.e. from the first block of first row to last block of first row).In the same manner, the fortune when estimation for the secondary series of present frame
When moving vector, motion vector decision circuitry 206 judges the motion vector of the block on secondary series (that is, from secondary series in order
The first block to secondary series last block).Please refer to Figure 16, due in Figure 10 search direction and Candidate Set CSd
It is selected, then motion vector decision circuitry 206 judges the motion vector of block from bottom to up in order.When estimation is for current
When the motion vector of each column of frame, motion vector decision circuitry 206 judge in order the movement of the block on each column to
It measures (that is, from last block of each column to the first block of each column).
About each block of present frame, motion vector decision circuitry 206 judges distinctly according to other Candidate Set
Motion vector.For example, Fig. 7 is please referred to, when being selected for the Candidate Set CSa in current block Bc (such as Fig. 7 institute
Show), motion vector decision circuitry 206 calculate current block Bc and correspond respectively to Candidate Set CSa candidate motion vector S3,
Difference between the block of T1, T2, and according to the difference come one of select in candidate motion vector S3, T1, T2 with
Motion vector as current block Bc.
Each difference can be in pixel and the Candidate Motion in corresponding Candidate Set CSa in current block Bc
The absolute difference and value (a sum-of-the-absolute difference, SAD) of the pixel of the block of vector.In the present embodiment
In, about spatial candidate motion vector S3, motion vector decision circuitry 206 calculate pixel in current block Bc with corresponding
The absolute difference and value SAD1 between pixel in the block of candidate motion vector S3.
In addition, the candidate motion vector S3 that block corresponds to Candidate Set CSa can be in previous frame or in present frame
In next frame (that is, by use backward motion estimation process).Since forward motion estimation process carrys out estimation space adjacent region
The motion vector of block B3, according to candidate motion vector S3 and by the shift position of current block Bc, so that corresponding to candidate fortune
The block of moving vector can be configured in previous frame.Due to calculated using backward motion estimation process spatial candidate motion to
S3 is measured, according to candidate motion vector S3 and by the shift position of current block Bc, so that corresponding to candidate motion vector S3's
Block can be configured in next frame of present frame.
Figure 17 is the schematic diagram for illustrating the block corresponding to spatial candidate motion vector S3 according to one embodiment of the invention.It lifts
For example, it is assumed that backward motion estimation process is utilized, so that Candidate Set CSa is selected and the motion vector of spatial neighbors B3
It is calculated, then selects the motion vector of spatial neighbors B3 using the Candidate Motion of the Candidate Set CSa as current block Bc
Vector.About spatial candidate motion vector S3, the block 1202 corresponding to candidate motion vector S3 can be selected.Please refer to figure
17, by with spatial neighbors B3 (that is, candidate motion vector S3), so that block 1202 is configured in next frame.
Accordingly, with respect to spatial candidate motion vector S3, motion vector decision circuitry 206 can calculate the pixel in current block Bc
Absolute difference and value SAD1 between the pixel corresponding to the block of candidate motion vector S3.
About time candidate motion vector T1, motion vector decision circuitry 206 calculate current block Bc with corresponding
For the absolute difference and value SAD2 between the pixel of the block of candidate motion vector T1.Block corresponding to candidate motion vector T1
Can by for the candidate motion vector T1 in current block Bc to be displaced block.It is transported in current block Bc with candidate is corresponded to
Offset between the block of moving vector T1 is equal to the motion vector of temporally adjacent block B1.In the same manner, due to utilizing rear Xiang Yun
Dynamic estimation process calculates the motion vector of temporally adjacent block B1, according to temporally adjacent block B1 motion vector and by working as
The displaced position of preceding block Bc is configured in the block corresponding to candidate motion vector T1 in previous frame.Due to benefit
With it is preceding to estimation process calculate the motion vector of temporally adjacent block B1, according to temporally adjacent block B1 movement to
It measures and by the displaced position of current block Bc, the block corresponding to candidate motion vector T1 is allowed to be configured in present frame
Next frame in.
About time candidate motion vector T2, motion vector decision circuitry 206 calculates the pixel in current block Bc
With corresponding to candidate motion vector T2 pixel between absolute difference and value SAD3.Block corresponding to candidate motion vector T2 can
With by the candidate motion vector T2 for current block Bc so that block is displaced by.It is transported in current block Bc with candidate is corresponded to
Deviant between the block of moving vector T2 is equal to the motion vector of temporally adjacent block B2.In the same manner, due to utilizing rear Xiang Yun
Dynamic estimation process calculates the motion vector of temporally adjacent block B2, according to temporally adjacent block B2 motion vector and by
The displaced position of current block Bc is configured in the block corresponding to candidate motion vector T2 in present frame.Due to
The motion vector that temporally adjacent block B2 is calculated using backward motion estimation process, the movement according to temporally adjacent block B2
Vector and by the displaced position of current block Bc is configured in the block corresponding to candidate motion vector T2 currently
In next frame of frame.
In addition, motion vector decision circuitry 206 compares absolute difference and value SAD1, SAD2, SAD3, and select absolute difference and value
SAD1,SAD2,SAD3.For example, if absolute difference and value SAD2 are the smallest absolute difference and value, motion vector decision circuitry 206
Select the motion vector of temporally adjacent block T1 using as the motion vector in current block Bc.In other words, due to being flowed previously
Cheng Zhong, it has been estimated that the motion vector of each spatial neighbors, the motion vector of each spatial neighbors and
Temporally adjacent block can be utilized using the candidate motion vector as current estimation block.Motion vector decision circuitry 206 can
With from correspond to Candidate Set in selection have minimum absolute difference and value a candidate motion vector, and judge have minimum absolute difference
Motion vector with the candidate motion vector of value as current block Bc.
Figure 18 is the circuit block diagram for illustrating movement estimation apparatus according to another embodiment of the present invention.Movement estimation apparatus dress
Setting 1800 includes controller 1810, Candidate Set generation circuit 1820 and motion vector decision circuitry 1830.Candidate Set generates electricity
Road 1820 generates the Candidate Set of each of vision signal VS block according to design requirement.1830 coupling of motion vector decision circuitry
Candidate Set generation circuit 1820 is connected to receive candidate motion vector, receiving includes a series of picture frame, by each frame area
It is divided into block, and according to each corresponding candidate motion vector received by the Candidate Set generation circuit 1820, it is every to obtain
The motion vector MV of a block.The motion vector MV of block is also provided to Candidate Set generation circuit 1820, using as present frame
Other blocks candidate motion vector, and as candidate spatial motion vector in candidate time motion vector or continuous
Frame.
Controller 1810 is coupled to Candidate Set generation circuit 1820 and motion vector decision circuitry 1830.Controller 1810
Judge the search direction in each block, and controls the disposed of in its entirety process of estimation.According to the demand of design, movement is estimated
Counter device 1800 can realize by circuit, e.g. integrated circuit or the general processor of one or more core architectures
The combination of (such as central processing unit) or hardware circuit and software program code, such as the ASIC with constructed in hardware block.
Figure 19 is the flow chart for illustrating method for estimating according to another embodiment of the present invention.Referring to Figure 18 and
Figure 19, in step S1910, Candidate Set generation circuit 1820 motion vector of most adjacent blocks can be added to current
The motion vector candidate collection of the current block of frame.Wherein, one of them in adjacent block and current block belong to same a line (or
Same row), and one of them in adjacent block is block in the current frame or the block in previous frame.Motion vector
Decision circuitry 1830 is coupled to Candidate Set generation circuit 1820 to receive motion vector candidate collection.In step S1920, move to
Amount decision circuitry 1830 can judge the motion vector of the current block of present frame according to motion vector candidate collection.
Figure 20 is the schematic diagram illustrated according to the present invention relative to an embodiment of Fig. 6.The possible time of current block Bc
Selected works, the time motion vector T0 of the corresponding block of block Bc from the previous frame F-1 being added is as present frame F's
Current block Bc.
Figure 21 is the schematic diagram that one embodiment of the invention illustrates motion vector candidate collection CS.For generating the time of Candidate Set CS
Select motion vector, and combine the motion estimation process of current block Bc, with execute n motion estimation process ME1, ME2 ...,
MEn.Each of motion estimation process ME1~Men include corresponding Candidate Set (that is, motion vector candidate collection CS1~
CSn).Motion estimation process ME1~ME4 is configured to along search from top to bottom and handles direction, to execute block respectively
The estimation of B1~B4.Therefore, motion estimation process ME1 generates motion vector S1 using the estimation as current block Bc
Handle ME1.Above-mentioned situation is simulated in Fig. 9.
Figure 22 is the flow chart for illustrating method for estimating according to another embodiment of the present invention.In step S2210, image
Frame is divided into multiple blocks.In step S2220, for each block, Candidate Set generation circuit 1820 is selected for working as
The Candidate Set CS of preceding block Bc.In step S2230, the selection of controller 1810 searches direction SD to handle the fortune of current block Bc
Dynamic estimation.In step S2240, motion vector decision circuitry 1830 calculates the sad value of the current block with reference block,
Each of these is corresponding to one of in the candidate motion vector in Candidate Set.In step S2250, motion vector is sentenced
Deenergizing 1830 judges the best match motion vector in candidate motion vector, using the motion vector as current block Bc.
Figure 23 is the flow chart for illustrating the parallel processing process of method for estimating according to another embodiment of the present invention.It please join
According to Figure 23, motion vector decision circuitry 1830 can execute parallel processing on most rows.Picture frame be divided into M row with it is each
The N number of block of row.The motion estimation process of motion vector decision circuitry 1830 executes parallel line by line.In the present embodiment, when
The Candidate Set CS of preceding block Bc under the movement of processing, arrange from mutually colleague on adjacent block candidate spatial motion vector with
As block (that is, the case where Fig. 7, Fig. 8 and/or Figure 12).In the same manner, parallel processing, which can be configured in, arranges (that is, Fig. 9
And/or the case where Figure 10).The configuration of parallel processing can accelerate to handle and promote convergence rate, due to the candidate of current block Bc
Collection CS (that is, the case where Figure 12) can have more spatial motion vectors from adjacent block before the processing of block itself starting.
Referring to Fig. 6~Figure 12 and Figure 20~Figure 21, at the estimation of the current block Bc of present frame
Reason is obvious.Candidate Set CS may include in the identical row or column from the current block Bc being configured in present frame
At least one spatial candidate motion vector of adjacent block.This mode can improve motion vector candidate collection CS in 3DRS method
Coverage area, to promote convergence rate to obtain the accurate motion vector of current block.In addition, motion vector candidate collection CS
The parallel processing of motion estimation process is also contributed to, and then promotes the speed of processing.
Figure 24 is that one embodiment of the invention illustrates showing for frame-rate conversion (frame rate up conversion, FRC)
It is intended to.Referring to figure 2. 4, generate one or more frames between two original frames, for example, frame Fint is continuous
Interpolation frame between frame F and F+1, and frame Fint1 and Fint2 are the interpolation frames between original frame F-1 and F-2.
Figure 25 is the flow chart for illustrating method for estimating according to one embodiment of the invention.Referring to Figure 24 and figure
25, the judging result of interpolation frame Fint is generated from original frame F and F+1, about each block Bv of interpolation frame Fint, for
The block Bv of interpolation frame Fint selects motion vector candidate collection CS (step S2510).Each of motion vector candidate collection CS
Candidate motion vector is displaced parallel, to be transferred to the block Bv (step S2520) of interpolation frame Fint.Execute the candidate with displacement
The motion estimation process of each of motion vector, to obtain optimal matching motion vector, from the block for being configured at frame F
Ba, and another block Bb (step S2530) for being configured at another frame F+1 is directed toward via block Bv.By block Ba, Bb
Combination is to calculate block Bv (step S2540).Processing performed in step S2530 is defined as projection (projected)
Motion estimation process, and optimal matching motion vector is known as project motion vector.
Figure 26 is the example for project motion estimation processing.The Candidate Set vector M V_f of motion vector candidate collection CS is taken
Out.Motion vector MV_f executes previous forward estimation from the frame F with reference frame F+1, for block B10, and in frame F+1
Block B20 obtain optimal matching.Motion vector MV_f is displaced parallel to transmit the block B30 in interpolation frame Fint, with
Obtain the motion vector MV_p of displacement.Therefore, motion vector MV_p is directed toward the block B21 in frame F+1 from the block B11 in frame F.
It is verified by backward motion estimation with similar block B11, B21, backward motion estimation is associated with block B30 one
Kind project motion estimation, and if motion vector MV_p is best match in the candidate motion vector of motion vector candidate collection CS
One, then motion vector MV_p is known as to the project motion vector of block B30.By the combination of block B11, B21 to generate area
Block B30.The bilinear combinations (bi-linear combination) of application block B11, B21, such as B30=r*B11+ (1-
R) * B21, wherein r, which can be, to be determined by many parameters with design requirement.For example, the considerations of parameter r, may include
The relative time distance of frame Fint to each of frame F-1, F.
In the same manner, it can use another candidate motion vector to realize above-mentioned process.Figure 27 is for another implementation
The process of the project motion estimation of example.The motion vector MV_b of motion vector candidate collection CS be can use to realize above-mentioned stream
Journey.Motion vector MV_b executes forward motion estimation from the frame F+1 with reference frame F, for block B40, and in frame F
Block B50 obtains optimal matching.Motion vector MV_b is displaced parallel to transmit the block B30 in interpolation frame Fint, to obtain
The motion vector MV_p ' that must be displaced.Therefore, motion vector MV_p ' is directed toward the block B51 in frame F from the block B41 in frame F+1.
It is verified by another forward motion estimation with similar block B41, B51, forward motion estimation is related to block B30
Connection a kind of project motion estimation, and if motion vector MV_p ' be in the candidate motion vector of motion vector candidate collection CS most
Good matched one, then motion vector MV_p ' is known as to the project motion vector of block B30.By the combination of block B41, B51
To generate block B30.
The candidate motion vector of block B30 in interpolation frame Fint can be from the movement between frame F, the F+1 previously executed
It is selected in estimation processing.Such as shown in Figure 26 and Figure 27, and between frame F-1, F during generation interpolation frame, movement
Estimation processing is performed in previously, including forward, backward and project motion estimation processing.For example, referring to figure 2. 8, choosing
Select the forward motion vector MV_f ' and backward motion vectors MV_ that each motion estimation process is executed between frame frame F-1, F
B ', using as the candidate motion vector for the block B30 in interpolation frame Fint.Figure 28 is the interpolation frame between frame F-1, F
The project motion vector M V_p " of Fint ', is also selected as the candidate motion vector of block B30.The Candidate Motion of block B30 to
The candidate motion vector for measuring the block of same position of the option and installment in each frame, using as the block for being located at interpolation frame Fint
The position of B30.For example, Figure 26, Figure 27 and Figure 28 please be respectively refer to, block B40, the frame of the block B10 of frame F, frame F+1
The block B60 of the block B70 and frame Fint ' of F-1 are configured at the same position in the interpolation frame of the block B30 in each frame.
Facilitate the convergence of the accurate motion vector of acquisition block B30.
Figure 29 is to illustrate that the possibility Candidate Set of the current block Bc of frame includes from the same block position of current block Bc
The adjacent block of present frame shown in the multiple time motion vectors and Figure 20 of more than one consecutive frame and previous frame to
Amount.Consecutive frame may include original and interpolation frame.It please respectively refer to shown in Figure 26~Figure 28, be handled for project motion estimation
Motion vector candidate collection CS the case where, wherein motion vector T01~T05 correspond to movement to MV_f ', MV_p ", MV_f,
MV_b',MV_b.Motion vector candidate collection CS further includes the big random file and small random file and such as of spatial motion vector
Zero motion vector (or GMV) is used in 3DRS to obtain better coverage rate.Although selecting motion vector candidate collection CS's
Candidate motion vector has many possibilities, but in the considerations of selecting, treatment effeciency is mostly important.For example, it transports
Moving vector MV_p " and motion vector MV_f ' and MV_b ' are highly relevant, in fact, not needing all to be selected as motion vector time
Selected works CS.Selected Candidate Set CS is given, still can choose some Candidate Motions that can be influenced from adjacent block
The search direction of the estimation process of vector, such as shown in Fig. 7~Figure 12, with the change for searching direction, with adjacent region
The associated some time motion vectors of block translate into spatial motion vector.
Figure 30 is the possibility Candidate Set CS for backward motion estimation processing.Area as shown in Figure 26~Figure 28, for frame F
The backward motion estimation of block B10, that is to say, that in such a case, current block Bc (that is, the frame F in Figure 26~Figure 28
Block B10) be frame F block B10.The candidate motion vector of motion vector candidate collection CS may include from current block Bc
The vector T 01 of the block adjacent to present frame of identical block locations, T02, S0.Wherein, motion vector T01, T02, S0 points
Not Dui Yingyu motion vector MV_f ' (forward motion vector of the block B70 in Figure 28), (area in Figure 28 motion vector MV_p "
The project motion vector of block B60) and motion vector MV_b ' (backward motion vectors of the block B10 in Figure 28).Motion vector
Candidate Set CS can also include from the spatial motion vector of adjacent block and time motion vector, spatial motion vector it is big random
Displacement and small random file and previously described zero motion vector (or GMV).
Figure 31 is the possibility Candidate Set for forward motion estimation method, in the case, 6~Figure 28 referring to figure 2., frame
The forward motion estimation of the block B40 of F+1, that is to say, that current block Bc is in the block B40 of frame F+1.Motion vector is waited
The candidate motion vector of selected works CS may include from the movement of the block of the consecutive frame of the identical block locations of current block Bc to
It measures T01~T03 (that is, block B40 of the frame F+1 in Figure 27~Figure 28), wherein motion vector T01~T03 is corresponded respectively to
Motion vector MV_f (forward motion vector of the block B10 in Figure 26), motion vector MV_b ' is (after the block B10 in Figure 28
To motion vector) and motion vector MV_p " (the project motion vector of the block B60 in Figure 28).Motion vector candidate collection CS
Can also include from the spatial motion vector of adjacent block and time motion vector, the big random file of spatial motion vector with it is small
Random file and previously described zero motion vector (or GMV).
Figure 32 is the source for illustrating Candidate Set CS according to another embodiment of the present invention.In embodiment as shown in figure 32, hold
5 of candidate motion vector of the row for generating the motion vector candidate collection CS in conjunction with the motion estimation process of current block Bc
Motion estimation process ME1~ME5.Each of motion estimation process ME1~ME5 all includes corresponding Candidate Set, that is,
Motion vector candidate collection CS1~CS5.In the embodiment of Figure 29, i.e., in project motion estimation scene, motion estimation process
ME1~ME5 is configured in respectively to the same position of each of consecutive frame to execute estimation.Wherein, in consecutive frame
The position of same position block and the position of the current block Bc in frame Fint be identical.Motion estimation process ME1~ME5 points
Not Shu Chu candidate motion vector T01~T05, using as current block Bc project motion estimation processing motion vector candidate collection
The candidate motion vector of CS.Shown in another embodiment of Figure 30, that is, in backward motion estimation scene, motion estimation process
ME1~ME3 is configured to the same position block in each of consecutive frame and present frame, to execute at estimation
Reason, wherein the position of the same position block in consecutive frame and the position of the current block Bc in frame F are identical.Estimation
Processing ME1~ME3 exports candidate motion vector T01, T02, S0 respectively, using the backward motion estimation processing as current block Bc
Candidate Set CS candidate motion vector.In another embodiment of Figure 31, that is, in forward motion estimation scene, movement is estimated
Meter processing ME1~ME3 is configured to the same position block in each of consecutive frame consecutive frame, to execute estimation,
Wherein, the position of the same position block in consecutive frame and the current block Bc in frame F+1 are identical.Motion estimation process ME1
~ME3 exports candidate motion vector T01~T03 respectively, using the Candidate Set of the forward motion estimation processing as current block Bc
The candidate motion vector of CS.
Referring to Figure 29 and Figure 32, the motion estimation process of the current block Bc for frame, Candidate Set CS can be with
More than one candidate motion vector including the block from the current and/or consecutive frame of current block Bc same position.
The method can improve the coverage rate of the Candidate Set CS in 3DRS method, and obtain the more accurate motion vector of current block Bc and
Convergence rate is promoted simultaneously.
Figure 33 be according to the present invention another embodiment illustrate Candidate Set processing schematic diagram.In the embodiment of Figure 33,
Associated time can be handled to improve current kinetic estimation by the first motion estimation process of previously performed one or more
Selected works, and the object motion vector of current kinetic estimation processing can be improved.Referring to figure 3. 3, motion estimation process ME2~
ME4 previous motion estimation process ME1 exports its object motion vector MV1, using the time as motion estimation process ME2~ME4
Select motion vector.Therefore, Candidate Set CS2 is set as Candidate Set CS4.In the estimation that motion estimation process ME3~ME4 is previous
ME2 output vector MV2 is handled, using the candidate motion vector as motion estimation process ME3~ME4.Therefore, Candidate Set CS3 into
One step is set as Candidate Set CS4.Motion estimation process ME4 previous movement thus meter processing ME3 exports motion vector MV3, to make
For the candidate motion vector of motion estimation process ME4.Candidate Set CS4 is set again.Finally, motion estimation process ME4 joins
It examines and the Candidate Set CS4 after processing ME1~ME3 setting is estimated by previous movement, to obtain the motion vector MV4 of best match, and
Further promote the accuracy of object motion vector.
In the case where certain applications, the final goal fortune of the last motion estimation process in estimation chain can be provided
Moving vector, so that the Candidate Set of setting in motion the estimation even Candidate Set of subsequent motion estimation process (not drawing) carry out carefully
Change, to search the motion estimation process in direction with or without change with the restarting of estimation chain, so that refinement is every
The motion vector of one motion estimation stage, and then refine final goal motion vector.For example, as shown in figure 33, motion vector
MV4 can be used for refining the Candidate Set CS1 of the second motion estimation process with estimation chain.Normally, it repeats
Motion estimation process is by the search direction for changing each motion estimation stage, and by from the first running adjacent block
Processing result, Lai Gengxin candidate motion vector, further promoted final goal motion vector accuracy.
Figure 34 is the circuit block diagram for illustrating movement estimation apparatus according to the embodiment of Figure 18.Wherein, frame-rate conversion
The example of (frame rate up conversion, FRC), the thin of Candidate Set is explained by previous motion estimation process
Change, to improve object motion vector.Referring to figure 3. 4, the motion vector candidate Ji Bao as provided by Candidate Set generation circuit 1820
Include the first motion vector candidate collection (such as Candidate Set CS_f), the second motion vector candidate collection (such as Candidate Set CS_b) and
Three motion vector candidate collection (such as Candidate Set CS_p), and motion vector decision circuitry 1830 includes the first exercise estimator circuit
(such as circuit 910 for executing backward motion estimation), the second exercise estimator circuit (such as estimate for executing reverse
The circuit 920 of meter) and third exercise estimator circuit (such as circuit 930 for executing project motion estimation) difference correlation
It is coupled to Candidate Set CS_f, CS_b, CS_p.Backward motion estimation 910 is held using the Candidate Set CS_f of the current block of present frame
The first estimation of row acts (such as backward motion estimation), to obtain the first motion vector (example of the current block of present frame
Such as motion vector MV_1).The motion vector MV_1 as provided by backward motion estimation 910 is injected towards the current block of present frame
Candidate Set CS_b, CS_p in.Forward motion estimation 920 executes second using the Candidate Set CS_b of the current block of present frame
Estimation acts (such as forward motion estimation), (such as is moved with obtaining the second motion vector of the current block of present frame
Vector M V_2).The motion vector MV_2 as provided by forward motion estimation 920 is injected towards the candidate of the current block of present frame
Collect in CS_p, CS_f.Project motion estimation 930 is executed third movement using the Candidate Set CS_p of the current block of present frame and estimated
Meter acts (such as project motion estimation), to obtain third motion vector (such as the motion vector MV_ of the current block of present frame
3).As project motion estimation 930 provided by motion vector MV_3 be injected towards present frame current block Candidate Set CS_f,
In CS_b.Operation estimation processing 910,920,930 forms iteration chain, by the estimation of progress to obtain motion estimation vector
With the Candidate Set for refining two subsequent motion estimation processing.That is, motion vector MV_1 is for refining Candidate Set CS_
B, CS_p, motion vector MV_2 are for refining Candidate Set CS_p, CS_f, and motion vector MV_3 is for refining Candidate Set CS_f, CS_
b。
Referring to Figure 26, Figure 28, block B30 is generated in interpolation frame Fint, according to from being configured at previous interpolation frame
The Candidate Set CS_f of project motion vector M V_p " in the project motion estimation 930 of block B60 in Fint ', and in frame F
Block B10 execute backward motion estimation 910, and for frame F block B10 forward motion estimation 920 reverse to
Measure MV_b '.During project motion vector M V_p " and backward motion vectors MV_b ' is all the interpolation frame between frame F-1, F
The motion vector of generation.According to above-mentioned Candidate Set CS_f, backward motion estimation 910 generates motion vector MV_f to optimize candidate
Collect CS_b, CS_p.For the block B40 in frame F+1 to execute forward motion estimation 920, to generate motion vector MV_b.It provides
Motion vector MV_b is to refine Candidate Set CS_p, CS_f.Then, for the block B30 in interpolation frame Fint to execute projection fortune
Dynamic estimation 930, to generate project motion vector M V_p.Project motion vector M V_p is defined as the picture for generating block B30
The final motion vector of the block B30 of element, or it is provided for refinement Candidate Set CS_f, CS_b, it is used for from backward motion estimation
The another an iteration of 910 motion estimation process started.
Although Figure 34 has backward motion estimation performed before forward motion estimation before project motion estimation,
But the reference sequence of backward motion estimation and forward motion estimation can be exchanged.That is, Xiang Yun before being carried out before project motion estimation
Forward motion estimation is executed before dynamic estimation.The sequence that forward direction is handled with forward motion estimation can carry out in duplicate processing
Exchange can be exchanged in running sequence for the first time with second of running sequence.
Based on above-mentioned, in order to carry out interpolation to from the block B30 of the frame Fint in frame F, F+1 in FRC is applied, for frame
The backward motion estimation of the block B10 of F and the forward motion estimation of block 40, frame F+1 are configured to provide the projection of block B30
The candidate motion vector of the Candidate Set of estimation.In addition, executing the mode of corresponding estimation successively to improve candidate
Motion vector, due to project motion estimation final motion vector be can not be received, estimation chain can be weighed
It is multiple.The repeating motion estimation of final the be supplied to Candidate Set refinement of motion vector, to improve the convergence of final goal motion vector.
Figure 35 is the circuit block diagram for illustrating movement estimation apparatus 1800 according to another embodiment of Figure 18.Wherein, Figure 34
In two iteration example for obtaining final goal motion vector.Referring to figure 3. 5, by Candidate Set generation circuit 1820
Provided motion vector candidate collection includes the first motion vector candidate collection (such as Candidate Set CS_f1), the second motion vector time
Selected works (such as Candidate Set CS_b1), third motion vector candidate collection (such as Candidate Set CS_p1), the 4th motion vector candidate collection
(such as Candidate Set CS_f2), the 5th motion vector candidate collection (such as Candidate Set CS_b2) and the 6th motion vector candidate collection
(such as Candidate Set CS_p2), and motion vector decision circuitry 1830 includes that the first exercise estimator circuit (such as estimate by propulsion
Meter is 1110), the second exercise estimator circuit (such as forward motion estimation 1120), third exercise estimator circuit (such as project
Estimation 1130), the 4th exercise estimator circuit (such as backward motion estimation 1140), the 5th exercise estimator circuit (example
Such as forward motion estimation 1150) and the 6th exercise estimator circuit (such as project motion estimation 1160).Circuit 1110,1140
For the backward motion estimation in different search directions, reverse of the circuit 1120,1150 for different search directions is estimated
Meter, and project motion estimation of the circuit 1130,1160 for different search directions.
Backward motion estimation device 1110 executes the first estimation using the Candidate Set CS_f1 of the current block of present frame
It acts (such as backward motion estimation), to obtain the first motion vector (such as motion vector) of the current block of present frame.By
The motion vector MV_1 as provided by backward motion estimation 1110 be injected towards the current block of present frame Candidate Set CS_b1,
In CS_p1.It is dynamic that forward motion estimation 1120 executes the second estimation using the Candidate Set CS_b1 of the current block of present frame
Make (such as forward motion estimation), to obtain the second motion vector (such as motion vector MV_2) of the current block of present frame.
The Candidate Set CS_ of the current block of present frame is injected towards by motion vector MV_2 provided by forward motion estimation 1120
In p1, CS_f2.Project motion estimation 1130 executes third estimation using the Candidate Set CS_p1 of the current block of present frame
Act (such as project motion estimation), with obtain the Candidate Set of the current block of present frame third motion vector (such as move to
Measure MV_3).The time of the current block of present frame is injected towards by motion vector MV_3 provided by project motion estimation 1130
In selected works CS_f2, CS_b2.Backward motion estimation 1140 executes the 4th using the Candidate Set CS_f2 of the current block of present frame
Estimation act (such as backward motion estimation), with obtain the Candidate Set of the current block of present frame the 4th vector (such as
Motion vector MV_4).Work as proparea by what motion vector MV_4 provided by backward motion estimation 1140 was injected towards present frame
In Candidate Set CS_b2, CS_p2 of block.Forward motion estimation 1150 is held using the Candidate Set CS_b1 of the current block of present frame
The 5th estimation of row acts (such as forward motion estimation), to obtain the 5th vector of the Candidate Set of the current block of present frame
(such as motion vector MV_5).Working as present frame is injected towards by motion vector MV_5 provided by forward motion estimation 1150
In the Candidate Set CS_p2 of preceding block.Project motion estimation 1160 is executed using the Candidate Set CS_p2 of the current block of present frame
6th estimation acts (such as project motion estimation), to obtain the six-way amount of the Candidate Set of the current block of present frame
(such as motion vector MV_6).
Candidate Set CS_f1 has motion vector MV_b0, MV_p0 to be modified, and motion vector MV_p0 is from for producing
The motion estimation process of raw previous interpolation frame.Using motion vector MV_p0 with from the motion vector MV_ of backward motion estimation 1110
Selected works CS_b1 after 1 modification.Utilize the motion vector MV_1 from backward motion estimation 1110 and the fortune from forward motion estimation 1120
Selected works CS_p1 after moving vector MV_2 modification.Estimate using from the motion vector MV_2 of forward motion estimation 1120 with from project motion
The motion vector MV_3 of meter 1130 is come selected works CS_f2 after modifying.Candidate Set CS_b2 utilizes the movement from project motion estimation 1130
Vector M V_3 modifies with the motion vector MV_4 from backward motion estimation 1140.Utilize the fortune from backward motion estimation 1140
Moving vector MV_4 with from the motion vector MV_5 of forward motion estimation 1150 come selected works CS_p2 after modifying.Finally, from iterative motion
In estimation processing, project motion vector is modified, on project motion vector M V_3 to obtain final object vector MV_6.
Figure 36~Figure 41 is the part of Candidate Set CS_f1~CS_p2 in Figure 35 respectively.Figure 36 is search side from top to bottom
To.Wherein, the b/p of the position current block Bc corresponds respectively to motion vector MV_b0, MV_p0 of Figure 35.Figure 37 is from left to right
Search direction.Wherein, the f/p of the position current block Bc corresponds respectively to the motion vector of the backward motion estimation 1110 of Figure 35
MV_1.The embodiment of Figure 38~Figure 41 and the embodiment of Figure 36~Figure 37 are identical, and details are not described herein.
3~Figure 41 referring to figure 3., the current block Bc for frame be it will be apparent that for it is current in consecutive frame
Block Bc is located at the block of same position, successively executes most motion estimation process, wherein at the most estimations
There is each of reason the previous movement as motion vector to estimate, to generate the Candidate Set CS of subsequent motion estimation processing,
To refine the final goal motion vector of current block Bc.Final object motion vector can be provided to refine Candidate Set CS,
Processing is estimated with setting in motion to restart most motion estimation process, makes the final mesh for further refining current block Bc
Mark motion vector.The repetition of the most motion estimation process facilitates the convergence of final goal motion vector.Most fortune
Each of dynamic estimation processing institute's repetition motion estimation process also can change search direction with merge new Candidate Motion to
Amount, to improve the motion vector of best match, and then facilitates the convergence of final goal motion vector.For Candidate Set CS and
The thinning method of object motion vector is particularly conducive to the application of FRC.In the application of FRC, most motion estimation process packets
Include backward motion estimation, forward motion estimation and project motion estimation.Xiang Yun before being executed before project motion estimation processing
Dynamic estimation and forward motion estimation.One of backward motion estimation and forward motion estimation are executed, to generate the first fortune
Moving vector, to modify another Candidate Set CS in backward motion estimation and forward motion estimation.Later generate second move to
Amount.Second motion vector can also provide the first motion vector, to refine the Candidate Set CS of project motion estimation processing.With above-mentioned
Mode, Lai Shixian from produced by project motion estimation processing and for final goal between two original frames move to
Amount, to reach good convergence.
In some other embodiments, Candidate Set CS_b1 can also include other candidate motion vectors.For example (but
Be not intended to limit), the candidate motion vector of Candidate Set CS_b1 can be the movement of the adjacent block of the current block of present frame to
Amount.The first adjacent block and current block Bc in the current frame belongs to identical row (or identical column).Another Candidate Set
The candidate motion vector of CS_b1 can be the motion vector of adjacent block in the current frame.Second in previous frame is adjacent
Block and current block Bc in the current frame belong to identical column (or identical row).The time of another Candidate Set CS_b1
The motion vector for the adjacent block for selecting motion vector to can be in the current frame.The second adjacent block in previous frame and
Current block Bc in present frame belongs to identical column (or identical row).The candidate motion vector of another Candidate Set CS_b1
It can be the motion vector of another adjacent block in previous frame.Third adjacent block in previous frame in present frame
In current block Bc belong to identical row and column.The candidate motion vector of another Candidate Set CS_b1 can be present frame and add
The motion vector of first adjacent block of the current block of upper first random vector.By the primary vector in random selection vector
Range obtains the first random vector.The candidate motion vector of another Candidate Set CS_b1 can be present frame plus second with
The motion vector of first adjacent block of the current block of machine vector.By random selection secondary vector range come obtain second with
Machine vector, and secondary vector range is better than primary vector range.First and secondary vector range can according to design requirement come
It determines.The candidate motion vector of another Candidate Set CS_b1 can be null vector.The Candidate Motion of another Candidate Set CS_b1
Vector can be global motion vector (global motion vector).Description relevant to Candidate Set CS_b1 can be referred to
Derive other Candidate Sets, not described here any more.
Figure 42 is the flow chart for illustrating the method for estimation according to another embodiment of the present invention.Please refer to Figure 18 and figure
42, in step S4210, Candidate Set generation circuit 1820 can choose block in the current frame.Work as proparea in previous frame
The motion vector of block is injected towards the motion vector candidate collection CS of the current block Bc in present frame.Motion vector decision circuitry
1830 are coupled to Candidate Set generation circuit 1820 to receive motion vector candidate collection.In step S4220, motion vector judgement electricity
Road 1830 can judge the motion vector of the current block of present frame according to motion vector candidate collection.Step S4210 and step
The relevant action details of rapid S4220, please refers to the relevant paragraph illustrated about Figure 29~Figure 31.
In conclusion exercise estimator circuit and its method different described in the embodiment of the present invention.From existing
3DRS technology uses motion vector candidate collection by exercise estimator circuit, so that movement estimation apparatus of the invention has
Faster convergence rate and more accurate predicted motion vector.
Although the present invention is disclosed as above with embodiment, however, it is not to limit the invention, any technical field
Middle tool usually intellectual, without departing from the spirit and scope of the present invention, when can make some changes and embellishment, thus it is of the invention
Protection scope when view the attached claims institute defender subject to.
Claims (22)
1. a kind of for estimating the method for estimating of the motion vector of the block of present frame, comprising:
A motion vector candidate collection of a current block of the motion vector of multiple adjacent blocks to the present frame is added in selection,
Wherein, one of them of the current block and the adjacent block belong to same row or same a line, and the adjacent block
It is described one of them be a block of the present frame or a previous frame;And
A motion vector of the current block of the present frame is determined according to the motion vector candidate collection.
2. method for estimating as described in claim 1, wherein the motion vector candidate collection includes one first motion vector
Candidate Set and one second motion vector candidate collection determine the described current of the present frame according to the motion vector candidate collection
The step of motion vector of block includes:
First motion vector candidate by one first exercise estimator circuit according to the current block of the present frame
Collection acts to execute one first estimation, to obtain one first motion vector of the current block of the present frame;
First motion vector provided by the first exercise estimator circuit is added to described in the present frame and is worked as
The second motion vector candidate collection of preceding block;And
Second motion vector candidate by one second exercise estimator circuit according to the current block of the present frame
Collection acts to execute one second estimation, to obtain one second motion vector of the current block of the present frame.
3. method for estimating as claimed in claim 2, wherein the motion vector candidate collection further include a third move to
Candidate Set is measured, the motion vector of the current block of the present frame is determined according to the motion vector candidate collection
Step further include:
Second motion vector provided by the second exercise estimator circuit is added to described in the present frame and is worked as
The third motion vector candidate collection of preceding block;And
The third motion vector candidate by a third exercise estimator circuit according to the current block of the present frame
Collection acts to execute a third estimation, to obtain a third motion vector of the current block of the present frame.
4. method for estimating as claimed in claim 3, wherein being determined according to the motion vector candidate collection described current
The step of motion vector of the current block of frame further include:
First motion vector provided by the first exercise estimator circuit is added to described in the present frame and is worked as
The third motion vector candidate collection of preceding block.
5. method for estimating as claimed in claim 3, wherein being determined according to the motion vector candidate collection described current
The step of motion vector of the current block of frame further include:
Second motion vector provided by the second exercise estimator circuit is added to described in the present frame and is worked as
The first motion vector candidate collection of preceding block.
6. method for estimating as claimed in claim 3, wherein being determined according to the motion vector candidate collection described current
The step of motion vector of the current block of frame further include:
The third motion vector provided by the third exercise estimator circuit is added to described in the present frame and is worked as
The first motion vector candidate collection of preceding block.
7. method for estimating as claimed in claim 3, wherein being determined according to the motion vector candidate collection described current
The step of motion vector of the current block of frame further include:
The third motion vector provided by the third exercise estimator circuit is added to described in the present frame and is worked as
The second motion vector candidate collection of preceding block.
8. method for estimating as claimed in claim 3, wherein it is described move to candidate quantity set further include one the 4th move to
Candidate Set, one the 5th motion vector candidate collection and one the 6th motion vector candidate collection are measured, according to the motion vector candidate collection
The step of motion vector of the current block to determine the present frame further include:
The third motion vector provided by the third exercise estimator circuit is added to described in the present frame and is worked as
The 4th motion vector candidate collection of preceding block;
The 4th motion vector candidate by one the 4th exercise estimator circuit according to the current block of the present frame
Collection acts to execute one the 4th estimation, to obtain one the 4th motion vector of the current block of the present frame;
4th motion vector provided by the 4th exercise estimator circuit is added to described in the present frame and is worked as
The 5th motion vector candidate collection of preceding block;
The 5th motion vector candidate by one the 5th exercise estimator circuit according to the current block of the present frame
Collection acts to execute one the 5th estimation, to obtain one the 5th motion vector of the current block of the present frame;
5th motion vector provided by the 5th exercise estimator circuit is added to described in the present frame and is worked as
The 6th motion vector candidate collection of preceding block;And
The 6th motion vector candidate by one the 6th exercise estimator circuit according to the current block of the present frame
Collection acts to execute one the 6th estimation, to obtain one the 6th motion vector of the current block of the present frame.
9. method for estimating as claimed in claim 2, wherein second fortune of the current block of the present frame
Moving vector Candidate Set includes:
By first movement of the current block of the present frame provided by the first exercise estimator circuit
Vector;
One motion vector of one first adjacent block of the current block of the present frame, wherein first adjacent block
Belong to same row or same a line with the current block;And
One motion vector of one second adjacent block in the previous frame, wherein second adjacent block and described current
Block belongs to same row or same a line.
10. method for estimating as claimed in claim 9, wherein second fortune of the current block of the present frame
Moving vector Candidate Set further include:
One motion vector of the third adjacent block in the previous frame, wherein third phase adjacent area block and described current
Block belongs to same row or same a line;
The motion vector of first adjacent block of the current block of the present frame plus one first at random to
Amount, wherein by one vector of random selection to obtain first random vector from a primary vector range;
The motion vector of first adjacent block of the current block of the present frame plus one second at random to
Amount, wherein by one vector of random selection to obtain second random vector, and described second from a secondary vector range
Ranges of vectors is greater than the primary vector range;And
One null vector or a global motion vector.
11. a kind of for estimating the movement estimation apparatus of the motion vector of the block of present frame, comprising:
One Candidate Set generation circuit, to select the motion vector that multiple adjacent blocks are added to work as proparea to the one of the present frame
One motion vector candidate collection of block, wherein one of adjacent block of the current block and the adjacent block belongs to together
One column or same a line, and one of adjacent block of the adjacent block is an area of the present frame or a previous frame
Block;And
One motion vector decision circuitry, is coupled to the Candidate Set generation circuit to receive the motion vector candidate collection, to
A motion vector of the current block of the present frame is determined according to the motion vector candidate collection.
12. movement estimation apparatus as claimed in claim 11, wherein the motion vector candidate collection include one first move to
Amount Candidate Set and one second motion vector candidate collection, the motion vector decision circuitry include:
One first exercise estimator circuit, the first motion vector candidate collection of the current block according to the present frame
It is acted to execute one first estimation, to obtain one first motion vector of the current block of the present frame, wherein
First motion vector provided by the first exercise estimator circuit is added to described the Candidate Set generation circuit
The second motion vector candidate collection of the current block of present frame;And
One second exercise estimator circuit, the second motion vector candidate collection of the current block according to the present frame
It is acted to execute one second estimation, to obtain one second motion vector of the current block of the present frame.
13. movement estimation apparatus as claimed in claim 12, wherein the motion vector candidate collection further includes third movement
Vec-tor candidate collection, the Candidate Set generation circuit is by second motion vector provided by the second exercise estimator circuit
It is added to the third motion vector candidate collection of the current block of the present frame, the motion vector decision circuitry is also
Include:
One third exercise estimator circuit, the third motion vector candidate collection of the current block according to the present frame
It is acted to execute a third estimation, to obtain a third motion vector of the current block of the present frame.
14. movement estimation apparatus as claimed in claim 13, wherein the Candidate Set generation circuit estimates first movement
First motion vector provided by estimator circuit is added to the third movement of the current block of the present frame
Vec-tor candidate collection.
15. movement estimation apparatus as claimed in claim 13, wherein the Candidate Set generation circuit estimates second movement
Second motion vector provided by estimator circuit is added to first movement of the current block of the present frame
Vec-tor candidate collection.
16. movement estimation apparatus as claimed in claim 13, wherein the Candidate Set generation circuit estimates third movement
The third motion vector provided by estimator circuit is added to first movement of the current block of the present frame
Vec-tor candidate collection.
17. movement estimation apparatus as claimed in claim 13, wherein the Candidate Set generation circuit estimates third movement
The third motion vector provided by estimator circuit is added to second movement of the current block of the present frame
Vec-tor candidate collection.
18. movement estimation apparatus as claimed in claim 13, wherein the motion vector candidate collection further includes one the 4th movement
Vec-tor candidate collection, one the 5th motion vector candidate collection and one the 6th motion vector candidate collection, the Candidate Set generation circuit will
The third motion vector provided by the third exercise estimator circuit is added to the current block of the present frame
The 4th motion vector candidate collection, the motion vector decision circuitry further include:
One the 4th exercise estimator circuit, the 4th motion vector candidate collection of the current block according to the present frame
It is acted to execute one the 4th estimation, to obtain one the 4th motion vector of the current block of the present frame, wherein
4th motion vector provided by the 4th exercise estimator circuit is added to described the Candidate Set generation circuit
The 5th motion vector candidate collection of the current block of present frame;
One the 5th exercise estimator circuit is waited by the 5th motion vector for the current block for using the present frame
Selected works act to execute one the 5th estimation, to obtain one the 5th motion vector of the current block of the present frame,
Wherein the Candidate Set generation circuit by the 5th motion vector provided by the 5th exercise estimator circuit be added to
The 6th motion vector candidate collection of the current block of the present frame;And
One the 6th exercise estimator circuit, the 6th motion vector candidate collection of the current block according to the present frame
It is acted to execute one the 6th estimation, to obtain one the 6th motion vector of the current block of the present frame.
19. movement estimation apparatus as claimed in claim 12, wherein described the second of the current block of the present frame
Motion vector candidate collection includes:
Described the first of the current block of the present frame provided by the first exercise estimator circuit move to
Amount;
One motion vector of one first adjacent block of the current block of the present frame, wherein first adjacent block
Belong to same row or same a line with the current block;And
One motion vector of one second adjacent block in the previous frame, wherein second adjacent block and described current
Block belongs to same row or same a line.
20. movement estimation apparatus as claimed in claim 19, wherein described the second of the current block of the present frame
Motion vector candidate collection further include:
One motion vector of the third adjacent block in the previous frame, wherein third phase adjacent area block and described working as
Preceding block belongs to same row or same a line;
The motion vector of first adjacent block of the current block of the present frame plus one first at random to
Amount, wherein by one vector of random selection to obtain first random vector from a primary vector range;
The motion vector of first adjacent block of the current block of the present frame plus one second at random to
Amount, wherein by one vector of random selection to obtain second random vector, and described second from a secondary vector range
Ranges of vectors is greater than the primary vector range;And
One null vector or a global motion vector.
21. a kind of for estimating the method for estimating of the motion vector of the block of present frame, comprising:
One motion vector of one current block of one previous frame is added to the one of the current block of the present frame for selection
Motion vector candidate collection;And
A motion vector of the current block of the present frame is determined according to the motion vector candidate collection.
22. a kind of for estimating the method for estimating of the motion vector of the block of present frame, comprising:
One Candidate Set generation circuit, to select the motion vector by a current block of a previous frame to be added to described current
One motion vector candidate collection of the current block of frame;And
One motion vector decision circuitry, is coupled to the Candidate Set generation circuit to receive the motion vector candidate collection, to
A motion vector of the current block of the present frame is determined according to the motion vector candidate collection.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710650632.6A CN109391821B (en) | 2017-08-02 | 2017-08-02 | Motion estimation method and motion estimation device |
US15/674,522 US20180184107A1 (en) | 2016-12-28 | 2017-08-11 | Motion estimation method and motion estimation apparatus |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710650632.6A CN109391821B (en) | 2017-08-02 | 2017-08-02 | Motion estimation method and motion estimation device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109391821A true CN109391821A (en) | 2019-02-26 |
CN109391821B CN109391821B (en) | 2020-08-28 |
Family
ID=65412099
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710650632.6A Active CN109391821B (en) | 2016-12-28 | 2017-08-02 | Motion estimation method and motion estimation device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109391821B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113542743A (en) * | 2020-04-22 | 2021-10-22 | 瑞昱半导体股份有限公司 | Image processing method and image processing apparatus |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8144775B2 (en) * | 2004-03-30 | 2012-03-27 | Stmicroelectronics S.A. | Method and device for generating candidate motion vectors from selected spatial and temporal motion vectors |
CN102917197A (en) * | 2011-08-04 | 2013-02-06 | 想象技术有限公司 | External vectors in a motion estimation system |
CN103238319A (en) * | 2010-11-29 | 2013-08-07 | 联发科技股份有限公司 | Method and apparatus of extended motion vector predictor |
US20130230109A1 (en) * | 2001-07-24 | 2013-09-05 | Sandip Bhaumik | Motion estimation technique for digital video encoding applications |
US20170195686A1 (en) * | 2014-09-26 | 2017-07-06 | Polycom, Inc. | Motion estimation |
-
2017
- 2017-08-02 CN CN201710650632.6A patent/CN109391821B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130230109A1 (en) * | 2001-07-24 | 2013-09-05 | Sandip Bhaumik | Motion estimation technique for digital video encoding applications |
US8144775B2 (en) * | 2004-03-30 | 2012-03-27 | Stmicroelectronics S.A. | Method and device for generating candidate motion vectors from selected spatial and temporal motion vectors |
CN103238319A (en) * | 2010-11-29 | 2013-08-07 | 联发科技股份有限公司 | Method and apparatus of extended motion vector predictor |
CN102917197A (en) * | 2011-08-04 | 2013-02-06 | 想象技术有限公司 | External vectors in a motion estimation system |
US20170195686A1 (en) * | 2014-09-26 | 2017-07-06 | Polycom, Inc. | Motion estimation |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113542743A (en) * | 2020-04-22 | 2021-10-22 | 瑞昱半导体股份有限公司 | Image processing method and image processing apparatus |
Also Published As
Publication number | Publication date |
---|---|
CN109391821B (en) | 2020-08-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6370194B2 (en) | Motion vector estimation circuit and method | |
US8929451B2 (en) | External vectors in a motion estimation system | |
CN102461150B (en) | Method and device for motion estimation in a sequence of video images | |
CN111200736B (en) | Method and device for acquiring motion information in video coding/decoding | |
RU2013157151A (en) | DEVICE FOR PROCESSING IMAGES AND METHOD FOR PROCESSING IMAGES | |
JP2003533800A (en) | Motion estimator to reduce halo in MC upconversion | |
US20140010307A1 (en) | Method of and apparatus for complexity scalable frame rate up-conversion | |
CN106534858B (en) | True motion estimation method and device | |
KR20110022133A (en) | Move estimating method and image processing apparauts | |
CN101924936A (en) | Image frame interpolation device, image frame interpolation method and image frame interpolation program | |
CN101883286A (en) | Calibration method and device, and motion estimation method and device in motion estimation | |
CN109391821A (en) | Method for estimating and movement estimation apparatus | |
JP5566786B2 (en) | Error absolute value sum estimation system and estimation method | |
JP6321365B2 (en) | Corresponding point search method and distance measuring device | |
TWI444049B (en) | Frame prediction system and prediction method thereof | |
JP2004356747A (en) | Method and apparatus for matching image | |
JP2009187204A (en) | Image processing apparatus | |
CN100366092C (en) | Search method for video frequency encoding based on motion vector prediction | |
US20180184107A1 (en) | Motion estimation method and motion estimation apparatus | |
JP2008226067A (en) | Image processor and image processing method | |
JPWO2004012460A1 (en) | Motion vector detection apparatus and motion vector detection method | |
Wang et al. | Tracking facial feature points with prediction-assisted view-based active shape model | |
CN100416476C (en) | Hand held direction indicator and method for evaluating displacement quantity | |
KR20090017296A (en) | Method of image processing for generating an intermediate frame image and apparatus thereof | |
JP3437381B2 (en) | Motion vector detection device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |