CN105657319B - The method and system of candidate vector penalty value are controlled in ME based on feature dynamic - Google Patents
The method and system of candidate vector penalty value are controlled in ME based on feature dynamic Download PDFInfo
- Publication number
- CN105657319B CN105657319B CN201610134440.5A CN201610134440A CN105657319B CN 105657319 B CN105657319 B CN 105657319B CN 201610134440 A CN201610134440 A CN 201610134440A CN 105657319 B CN105657319 B CN 105657319B
- Authority
- CN
- China
- Prior art keywords
- vector
- block
- candidate motion
- apl
- penalty
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0135—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes
- H04N7/014—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level involving interpolation processes involving the use of motion vectors
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/144—Movement detection
- H04N5/145—Movement estimation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/01—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level
- H04N7/0127—Conversion of standards, e.g. involving analogue television standards or digital television standards processed at pixel level by changing the field or frame frequency of the incoming video signal, e.g. frame rate converter
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Television Systems (AREA)
- Image Analysis (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides the method and system for controlling candidate vector penalty value in a kind of ME based on feature dynamic, comprising: candidate motion vector obtaining step: obtains all candidate motion vectors of current calculation block;Matching cost calculates step: calculating the matching cost of each candidate motion vector one by one;Compare screening step: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block selects motion vector of the smallest candidate motion vector of matching cost as current calculation block.The present invention is a kind of scheme that vector penalty value adaptively adjusts, and can adjust the penalty value of vector according to practical convergence state, and the only fixed value that is determined by candidate blocks position in unorthodox method, therefore is more advantageous to the convergence of estimation.
Description
Technical field
The present invention relates to estimation of motion vectors fields, and in particular, to is punished in ME based on feature dynamic control candidate vector
The method and system of penalties.
Background technique
Frame per second converts (Frame Rate Conversion, FRC) for realizing video source from a kind of frame per second to another frame
The control of rate is converted, and is now widely used in TV chip, effectively solve and improve shake when video content plays and
Liquid crystal trailing phenomenon when high definition television is watched.Frame per second transfer algorithm based on motion estimation motion compensation is current frame per second conversion
The mainstream implementation of technology, wherein object between two frames is obtained by calculation in estimation (Motion Estimation, ME)
Motion vector, for motion compensated interpolation (Motion Compensation) provide motion information.
Three-dimensional recursive search method is motion estimation implementing method general in current hardware realization.Having realization technology is:
Image is divided by the block of M × N (value of M, N are generally 2,4,8,16,32 etc.) size, in the movement for estimating some block
When vector, by specifying in its neighborhood several pieces, their motion vector (Motion Vector, MV) is used as and needs to calculate
Current calculation block candidate motion vector, using optimal motion vector that the matching result of more each candidate vector is selected as working as
The MV of preceding block completes the estimation of the block.Wherein, the measurement of candidate vector matching superiority and inferiority, three-dimensional recursive search method
In be usually measured using the weighting of image similarity and the intrinsic penalty value of vector, this vector penalty value is according to candidate
The reliability of the position of the relatively current calculation block of block determines where vector, a usually fixed value pre-defined, and
Only controlled by candidate vector block position.Further, from the point of view of the result of estimation, in many scenes, movement is estimated
Meter cannot obtain correct motion vector in the edge of prospect.
Summary of the invention
For the defects in the prior art, the object of the present invention is to provide be based on feature dynamic in a kind of ME to control candidate arrow
Measure the method and system of penalty value.
The method that candidate vector penalty value is controlled based on feature dynamic in a kind of ME provided according to the present invention, comprising:
Candidate motion vector obtaining step: all candidate motion vectors of current calculation block are obtained;
Matching cost calculates step: calculating the matching cost of each candidate motion vector one by one;
Compare screening step: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block,
Select motion vector of the smallest candidate motion vector of matching cost as current calculation block.
Preferably, the matching cost calculates step, includes the following steps:
Block SAD and feature extraction step: it is directed to each candidate motion vector Vector_k, previous frame image is obtained and works as
Block block_pre, block_cur of corresponding two same sizes of prior image frame, according to the block block_pre of acquirement,
Block_cur calculates SAD (absolute difference and the Sum of Absolute of candidate motion vector Vector_k
) and feature Differences;
Reliability judgment step: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates obtaining step: according to SAD and adaptive vector penalty value, obtaining of candidate motion vector
With cost.
Preferably, described piece of SAD and feature extraction step, specifically:
SAD_k=∑J=1~bHt∑I=1~bWdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bWdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bWdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates average picture brightness (APL, the average pixel of candidate motion vector Vector_k
Level) feature, APL_pre indicate that the average picture brightness of block block_pre, APL_cur indicate block block_cur's
Average picture brightness;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bWdabs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bWdabs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates details (Dtl, detail) feature of candidate motion vector Vector_k, and Dtl_pre is indicated
The minutia of block block_pre, Dtl_cur indicate the minutia of block block_cur.
Preferably, it in the reliability judgment step, is calculated according to average picture brightness, minutia adaptive
The method of vector penalty value be one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_k indicates the vector penalty value of candidate motion vector Vector_k, and Penalty_apl indicates to wait
The average picture brightness of motion vector Vector_k is selected, α _ apl indicates the weighting of predefined average picture brightness
Coefficient, Penalty_dtl indicate the minutia of candidate motion vector Vector_k, and α _ dtl indicates predefined minutia
Weighting coefficient, Grey_medium indicates image gray-scale level intermediate value, and Penalty_max indicates the predefined vector penalty value upper limit.
Preferably, the comprehensive obtaining step of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, α _ pealty indicate that the vector of candidate motion vector Vector_k is punished
The predefined weighting coefficient of penalties.
The system that candidate vector penalty value is controlled based on feature dynamic in a kind of ME provided according to the present invention, comprising:
Candidate motion vector acquisition device: all candidate motion vectors of current calculation block are obtained;
Matching cost computing device: the matching cost of each candidate motion vector is calculated one by one;
Compare screening plant: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block,
Select motion vector of the smallest candidate motion vector of matching cost as current calculation block.
Preferably, the matching cost computing device, including following device:
Block SAD and feature acquisition device: it is directed to each candidate motion vector Vector_k, previous frame image is obtained and works as
Block block_pre, block_cur of corresponding two same sizes of prior image frame, according to the block block_pre of acquirement,
Block_cur calculates SAD (absolute difference and the Sum of Absolute of candidate motion vector Vector_k
) and feature Differences;
Reliability judgment means: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates acquisition device: according to SAD and adaptive vector penalty value, obtaining of candidate motion vector
With cost.
Preferably, described piece of SAD and feature acquisition device, specifically:
SAD_k=∑J=1~bHt∑I=1~bWdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bWdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bWdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates average picture brightness (APL, the average pixel of candidate motion vector Vector_k
Level) feature, APL_pre indicate that the average picture brightness of block block_pre, APL_cur indicate block block_cur's
Average picture brightness;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bWWabs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bWdabs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates details (Dtl, detail) feature of candidate motion vector Vector_k, and Dtl_pre is indicated
The minutia of block block_pre, Dtl_cur indicate the minutia of block block_cur.
Preferably, it in the reliability judgment means, is calculated according to average picture brightness, minutia adaptive
The method of vector penalty value be one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_k indicates the vector penalty value of candidate motion vector Vector_k, and Penalty_apl indicates to wait
The average picture brightness of motion vector Vector_k is selected, α _ apl indicates the weighting of predefined average picture brightness
Coefficient, Penalty_dtl indicate the minutia of candidate motion vector Vector_k, and α _ dtl indicates predefined minutia
Weighting coefficient, Grey_medium indicates image gray-scale level intermediate value, and Penalty_max indicates the predefined vector penalty value upper limit.
Preferably, the comprehensive acquisition device of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, α _ pealty indicate that the vector of candidate motion vector Vector_k is punished
The predefined weighting coefficient of penalties.
Compared with prior art, the present invention have it is following the utility model has the advantages that
The present invention is a kind of scheme that vector penalty value adaptively adjusts, and can adjust vector according to practical convergence state
Penalty value, and the fixed value only determined by candidate blocks position in unorthodox method, therefore it is more advantageous to the convergence of estimation.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is the step process of the method based on feature dynamic control candidate vector penalty value in ME provided by the invention
Figure.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, several changes and improvements can also be made.These belong to the present invention
Protection scope.
The method that candidate vector penalty value is controlled based on feature dynamic in the ME provided according to the present invention, comprising:
Candidate motion vector obtaining step: all candidate motion vectors of current calculation block are obtained;The Candidate Motion arrow
Amount includes any of time domain candidate motion vector, airspace candidate motion vector, random candidate motion vector and zero vector
Or appoint multiple vectors;Random candidate motion vector is randomly generated within the scope of a 0~N-1 by Generating Random Number
Numerical value, a new candidate vector of composition, the value range of N is positive integer.Specifically, the generation of random candidate motion vector
Method are as follows: from current calculation block BlkijAll time domain candidate motion vectors and airspace candidate motion vector in selection one or
Several candidate motion vectors, it is raw that the horizontal direction speed and vertical speed of these candidate motion vectors add random number respectively
- (2 be randomly generated at algorithmk)~2kNumerical value in -1 range constitutes new candidate motion vector, as current calculation block
BlkijRandom candidate motion vector.Wherein, k can be equal to arbitrary positive integer, be generally set to 1~4.
Matching cost calculates step: calculating the matching cost of each candidate motion vector one by one;
Compare screening step: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block,
Select motion vector of the smallest candidate motion vector of matching cost as current calculation block.
In preference, the matching cost calculates step, includes the following steps:
Block SAD and feature extraction step: it is directed to each candidate motion vector Vector_k, previous frame image is obtained and works as
Block block_pre, block_cur of corresponding two same sizes of prior image frame, according to the block block_pre of acquirement,
Block_cur calculates SAD (absolute difference and the Sum of Absolute of candidate motion vector Vector_k
) and feature Differences;
Reliability judgment step: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates obtaining step: according to SAD and adaptive vector penalty value, obtaining of candidate motion vector
With cost;
Further, described piece of SAD and feature extraction step, specifically:
SAD_k=∑J=1~bHt∑I=1~bWdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bWdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bWdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates average picture brightness (APL, the average pixel of candidate motion vector Vector_k
Level) feature, APL_pre indicate that the average picture brightness of block block_pre, APL_cur indicate block block_cur's
Average picture brightness;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bWdabs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bWdabs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates details (Dtl, detail) feature of candidate motion vector Vector_k, and Dtl_pre is indicated
The minutia of block block_pre, Dtl_cur indicate the minutia of block block_cur;
In the reliability judgment step, specifically:
The value of average picture brightness is closer to 0 or maximum gray value, then it is assumed that candidate motion vector Vector_k
More unreliable, corresponding vector penalty value is bigger;Conversely, then vector penalty value is smaller;
The value of minutia is smaller, then it is assumed that candidate motion vector Vector_k is more unreliable, and corresponding penalty value will
It is bigger;Conversely, then vector penalty value is smaller.
In the reliability judgment step, adaptive vector is calculated according to average picture brightness, minutia
The method of penalty value is one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_k indicates the vector penalty value of candidate motion vector Vector_k, and Penalty_apl indicates to wait
The average picture brightness of motion vector Vector_k is selected, α _ apl indicates the weighting of predefined average picture brightness
Coefficient, Penalty_dtl indicate the minutia of candidate motion vector Vector_k, and α _ dtl indicates predefined minutia
Weighting coefficient, Grey_medium indicates image gray-scale level intermediate value, and Penalty_max indicates the predefined vector penalty value upper limit;
The comprehensive obtaining step of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, α _ pealty indicate that the vector of candidate motion vector Vector_k is punished
The predefined weighting coefficient of penalties.
The system that candidate vector penalty value is controlled based on feature dynamic in a kind of ME provided according to the present invention, comprising:
Candidate motion vector acquisition device: all candidate motion vectors of current calculation block are obtained;
Matching cost computing device: the matching cost of each candidate motion vector is calculated one by one;
Compare screening plant: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block,
Select motion vector of the smallest candidate motion vector of matching cost as current calculation block.
Preferably, the matching cost computing device, including following device:
Block SAD and feature acquisition device: it is directed to each candidate motion vector Vector_k, previous frame image is obtained and works as
Block block_pre, block_cur of corresponding two same sizes of prior image frame, according to the block block_pre of acquirement,
Block_cur calculates SAD (absolute difference and the Sum of Absolute of candidate motion vector Vector_k
) and feature Differences;
Reliability judgment means: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates acquisition device: according to SAD and adaptive vector penalty value, obtaining of candidate motion vector
With cost.
Preferably, described piece of SAD and feature acquisition device, specifically:
SAD_k=∑J=1~bHt∑I=1~bWdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bWdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bWdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates average picture brightness (APL, the average pixel of candidate motion vector Vector_k
Level) feature, APL_pre indicate that the average picture brightness of block block_pre, APL_cur indicate block block_cur's
Average picture brightness;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bWdabs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bWdabs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates details (Dtl, detail) feature of candidate motion vector Vector_k, and Dtl_pre is indicated
The minutia of block block_pre, Dtl_cur indicate the minutia of block block_cur.
Preferably, it in the reliability judgment means, is calculated according to average picture brightness, minutia adaptive
The method of vector penalty value be one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_k indicates the vector penalty value of candidate motion vector Vector_k, and Penalty_apl indicates to wait
The average picture brightness of motion vector Vector_k is selected, α _ apl indicates the weighting of predefined average picture brightness
Coefficient, Penalty_dtl indicate the minutia of candidate motion vector Vector_k, and α _ dtl indicates predefined minutia
Weighting coefficient, Grey_medium indicates image gray-scale level intermediate value, and Penalty_max indicates the predefined vector penalty value upper limit.
Preferably, the comprehensive acquisition device of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, α _ pealty indicate that the vector of candidate motion vector Vector_k is punished
The predefined weighting coefficient of penalties.
One skilled in the art will appreciate that in addition to realizing system provided by the invention in a manner of pure computer readable program code
And its other than each device, completely can by by method and step carry out programming in logic come so that system provided by the invention and its
Each device is in the form of logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and embedded microcontroller etc.
To realize identical function.So system provided by the invention and its every device are considered a kind of hardware component, and it is right
The device for realizing various functions for including in it can also be considered as the structure in hardware component;It can also will be for realizing each
The device of kind function is considered as either the software module of implementation method can be the structure in hardware component again.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make a variety of changes or modify within the scope of the claims, this not shadow
Ring substantive content of the invention.In the absence of conflict, the feature in embodiments herein and embodiment can any phase
Mutually combination.
Claims (2)
1. a kind of method for controlling candidate vector penalty value based on feature dynamic in ME characterized by comprising
Candidate motion vector obtaining step: all candidate motion vectors of current calculation block are obtained;
Matching cost calculates step: calculating the matching cost of each candidate motion vector one by one;
Compare screening step: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block, selection
Motion vector of the smallest candidate motion vector of matching cost as current calculation block;
The matching cost calculates step, includes the following steps:
Block SAD and feature extraction step: it is directed to each candidate motion vector Vector_k, obtains previous frame image and present frame
Block block_pre, block_cur of corresponding two same sizes of image, according to block block_pre, block_ of acquirement
Cur calculates the SAD and feature of candidate motion vector Vector_k;
Reliability judgment step: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates obtaining step: according to SAD and adaptive vector penalty value, obtaining the matching generation of candidate motion vector
Valence;
Described piece of SAD and feature extraction step, specifically:
SAD_k=∑J=1~bHt∑I=1~bwdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bwdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bwdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates that the average picture brightness of candidate motion vector Vector_k, APL_pre indicate block block_
The average picture brightness of pre, APL_cur indicate the average picture brightness of block block_cur;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bwd abs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bwd abs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates the minutia of candidate motion vector Vector_k, and Dtl_pre indicates the details of block block_pre
Feature, Dtl_cur indicate the minutia of block block_cur;
The comprehensive obtaining step of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, Penalty_k indicate the vector of candidate motion vector Vector_k
Penalty value, α _ pealty indicate the predefined weighting coefficient of the vector penalty value of candidate motion vector Vector_k;
In the reliability judgment step, adaptive vector is calculated according to average picture brightness, minutia and is punished
The method of value is one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_apl indicates that the average picture brightness of candidate motion vector Vector_k subtracts Grey_medium
Difference and α _ apl product, α _ apl indicates the weighting coefficient of predefined average picture brightness, Penalty_dtl table
Show that Penalty_max subtracts the difference of the minutia of candidate motion vector Vector_k, α _ dtl indicates that predefined details is special
The weighting coefficient of sign, Grey_medium indicate image gray-scale level intermediate value, and Penalty_max is indicated in predefined vector penalty value
Limit.
2. the system for controlling candidate vector penalty value based on feature dynamic in a kind of ME characterized by comprising
Candidate motion vector acquisition device: all candidate motion vectors of current calculation block are obtained;
Matching cost computing device: the matching cost of each candidate motion vector is calculated one by one;
Compare screening plant: for current calculation block, the matching cost of more current all candidate motion vectors of calculation block, selection
Motion vector of the smallest candidate motion vector of matching cost as current calculation block;
The matching cost computing device, including following device:
Block SAD and feature acquisition device: it is directed to each candidate motion vector Vector_k, obtains previous frame image and present frame
Block block_pre, block_cur of corresponding two same sizes of image, according to block block_pre, block_ of acquirement
Cur calculates the SAD and feature of candidate motion vector Vector_k;
Reliability judgment means: according to the SAD and feature of candidate motion vector Vector_k, judge candidate motion vector
Whether Vector_k is reliable, and carries out vector penalty value to candidate motion vector Vector_k according to judging result and adaptively adjust
It is whole, obtain adaptive vector penalty value, wherein more reliable then vector penalty value is smaller, more unreliable then vector penalty value more
Greatly;
Matching cost integrates acquisition device: according to SAD and adaptive vector penalty value, obtaining the matching generation of candidate motion vector
Valence;
Described piece of SAD and feature acquisition device, specifically:
SAD_k=∑J=1~bHt∑I=1~bwdabs(G_prei,j-G_curi,j)
Wherein, SAD_k indicate candidate motion vector Vector_k absolute difference and, abs () expression seek absolute value, G_
prei,jIndicate the gray level information of the i-th row jth column pixel in block block_pre, G_curi,jIndicate the i-th row in block block_cur
The gray level information of jth column pixel;The size of block is bWd × bHt, and bWd indicates that the Horizontal number of pixels of block, bHt indicate the vertical of block
Pixel number;
APL_k=(APL_pre+APL_cur)/2
APL_pre=(∑J=1~bHt∑I=1~bwdG_prei,j)/(bHt*bWd)
APL_cur=(∑J=1~bHt∑I=1~bwdG_curi,j)/(bHt*bWd)
Wherein, APL_k indicates that the average picture brightness of candidate motion vector Vector_k, APL_pre indicate block block_
The average picture brightness of pre, APL_cur indicate the average picture brightness of block block_cur;
Dtl_k=(Dtl_pre+Dtl_cur)/2
Dtl_pre=∑J=1~bHt∑I=1~bwd abs(G_prei,j-APL_pre)
Dtl_cur=∑J=1~bHt∑I=1~bwd abs(G_curi,j-APL_cur)
Wherein, Dtl_k indicates the minutia of candidate motion vector Vector_k, and Dtl_pre indicates the details of block block_pre
Feature, Dtl_cur indicate the minutia of block block_cur;
The comprehensive acquisition device of the matching cost, specifically:
Cost_k=SAD_k × α _ sad+Penalty_k × α _ pealty
Wherein, Cost_k indicates the matching cost of candidate motion vector Vector_k, and α _ sad indicates candidate motion vector
The predefined weighting coefficient of the absolute difference sum of Vector_k, Penalty_k indicate the vector of candidate motion vector Vector_k
Penalty value, α _ pealty indicate the predefined weighting coefficient of the vector penalty value of candidate motion vector Vector_k;
In the reliability judgment means, adaptive vector is calculated according to average picture brightness, minutia and is punished
The method of value is one-dimensional map method, specifically:
Penalty_k=Penalty_apl × α _ apl+Penalty_dtl × α _ dtl
Penalty_apl=(APL_k-Grey_medium) × α _ apl
Penalty_dtl=Penalty_max-Dtl_k
Wherein, Penalty_apl indicates that the average picture brightness of candidate motion vector Vector_k subtracts Grey_medium
Difference and α _ apl product, α _ apl indicates the weighting coefficient of predefined average picture brightness, Penalty_dtl table
Show that Penalty_max subtracts the difference of the minutia of candidate motion vector Vector_k, α _ dtl indicates that predefined details is special
The weighting coefficient of sign, Grey_medium indicate image gray-scale level intermediate value, and Penalty_max is indicated in predefined vector penalty value
Limit.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610134440.5A CN105657319B (en) | 2016-03-09 | 2016-03-09 | The method and system of candidate vector penalty value are controlled in ME based on feature dynamic |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610134440.5A CN105657319B (en) | 2016-03-09 | 2016-03-09 | The method and system of candidate vector penalty value are controlled in ME based on feature dynamic |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105657319A CN105657319A (en) | 2016-06-08 |
CN105657319B true CN105657319B (en) | 2018-12-04 |
Family
ID=56493349
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610134440.5A Active CN105657319B (en) | 2016-03-09 | 2016-03-09 | The method and system of candidate vector penalty value are controlled in ME based on feature dynamic |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105657319B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106993108B (en) * | 2017-04-07 | 2020-08-28 | 上海顺久电子科技有限公司 | Method and device for determining random quantity of video image in motion estimation |
CN111526369B (en) * | 2020-04-29 | 2022-09-09 | Oppo广东移动通信有限公司 | Video motion estimation method and device, electronic equipment and storage medium |
CN112001943B (en) * | 2020-08-31 | 2024-07-30 | Oppo广东移动通信有限公司 | Motion estimation method and device, computer readable medium and electronic equipment |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101422047A (en) * | 2006-04-14 | 2009-04-29 | Nxp股份有限公司 | Motion estimation at image borders |
CN102447892A (en) * | 2010-09-30 | 2012-05-09 | 宝利微电子系统控股公司 | Motion estimation method and device |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090304293A1 (en) * | 2008-06-08 | 2009-12-10 | Te-Hao Chang | Motion estimation method and related apparatus for efficiently selecting motion vector |
-
2016
- 2016-03-09 CN CN201610134440.5A patent/CN105657319B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101422047A (en) * | 2006-04-14 | 2009-04-29 | Nxp股份有限公司 | Motion estimation at image borders |
CN102447892A (en) * | 2010-09-30 | 2012-05-09 | 宝利微电子系统控股公司 | Motion estimation method and device |
Also Published As
Publication number | Publication date |
---|---|
CN105657319A (en) | 2016-06-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
TWI455588B (en) | Bi-directional, local and global motion estimation based frame rate conversion | |
CN103888767B (en) | A kind of frame per second method for improving that UMH block-based motion estimations are combined with optical flow field estimation | |
US20110109796A1 (en) | Frame Rate Conversion Using Motion Estimation and Compensation | |
CN102291578B (en) | Apparatus and method for frame rate conversion | |
CN105657319B (en) | The method and system of candidate vector penalty value are controlled in ME based on feature dynamic | |
KR20100114499A (en) | Image interpolation with halo reduction | |
US10410358B2 (en) | Image processing with occlusion and error handling in motion fields | |
US9197891B1 (en) | Systems and methods for periodic structure handling for motion compensation | |
US9552623B1 (en) | Variable frame rate interpolation | |
CN102868879A (en) | Method and system for converting video frame rate | |
CN102014281A (en) | Methods and systems for motion estimation with nonlinear motion-field smoothing | |
CN102123283A (en) | Interpolated frame acquisition method and device in video frame rate conversion | |
CN111052184A (en) | Moving image processing device, display device, moving image processing method, and control program | |
US8692933B1 (en) | Method and apparatus for buffering anchor frames in motion compensation systems | |
US20120176536A1 (en) | Adaptive Frame Rate Conversion | |
CN105681806B (en) | Method and system based on logo testing result control zero vector SAD in ME | |
US8942503B1 (en) | Global motion vector calculation using phase plane correlation | |
US20120274845A1 (en) | Image processing device and method, and program | |
CN112532907A (en) | Video frame frequency improving method, device, equipment and medium | |
CN107124617B (en) | Method and system for generating random vector in motion estimation motion compensation | |
Ye et al. | Towards anytime optical flow estimation with event cameras | |
Park et al. | Covariance-based adaptive deinterlacing method using edge map | |
CN104780362A (en) | Video static logo detecting method based on local feature description | |
CN101860746B (en) | Motion estimation method | |
Aysu et al. | Low cost FPGA design and implementation of a stereo matching system for 3D-TV applications |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20200117 Address after: 266101 No. 399, Laoshan District Pine Ridge Road, Qingdao of Shandong province Patentee after: Qingdao Xinxin Microelectronics Technology Co., Ltd. Address before: 201203 Shanghai City Chenhui Road, Zhangjiang High Tech Park of Pudong New Area No. 88 Building 1 316 Patentee before: Hongyou Image Technology (Shanghai) Co., Ltd. |
|
TR01 | Transfer of patent right |