CN104811723B - Local motion vector modification method in MEMC technologies - Google Patents

Local motion vector modification method in MEMC technologies Download PDF

Info

Publication number
CN104811723B
CN104811723B CN201510203610.6A CN201510203610A CN104811723B CN 104811723 B CN104811723 B CN 104811723B CN 201510203610 A CN201510203610 A CN 201510203610A CN 104811723 B CN104811723 B CN 104811723B
Authority
CN
China
Prior art keywords
motion vector
block
local
memc
motion
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
Application number
CN201510203610.6A
Other languages
Chinese (zh)
Other versions
CN104811723A (en
Inventor
姜建德
余横
袁嘉林
刘广智
马琰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qingdao Xinxin Microelectronics Technology Co Ltd
Original Assignee
HONGYOU IMAGE TECHNOLOGY (SHANGHAI) Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by HONGYOU IMAGE TECHNOLOGY (SHANGHAI) Co Ltd filed Critical HONGYOU IMAGE TECHNOLOGY (SHANGHAI) Co Ltd
Priority to CN201510203610.6A priority Critical patent/CN104811723B/en
Publication of CN104811723A publication Critical patent/CN104811723A/en
Application granted granted Critical
Publication of CN104811723B publication Critical patent/CN104811723B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Abstract

Local motion vector modification method in a kind of MEMC technologies of present invention offer, including step are as follows:1) scope of some FX around it is specified to each block;2) accuracy of all block motion vector and the motion vector in the range of this is obtained;3) selection calculates a reliable motion vector as replacement motion vector from those motion vectors;4) analysis judges whether current block motion vector is reliable, calculates a weighted average coefficients;5) motion vector of block scripts and its replacement motion vector are weighted into out a new motion vector as current block final motion vector by the use of the weighted average coefficients.Inventive algorithm can replace bad motion vector by using the relatively good motion vector on periphery, the interpolated frame generated using smooth motion vector, visual effect is preferable, easily covers phenomena such as picture is broken when motion search is local not good enough.

Description

Local motion vector modification method in MEMC technologies
Technical field
The present invention relates to image processing techniques, more particularly to local motion vector amendment side in a kind of MEMC technologies Method.
Background technology
Motion estimation motion compensates (Motion estimation and motion compensation, MEMC) technology It is a kind of frame rate conversion generally used at present, applied in the video systems such as television equipment, the technology can effectively disappear Remove or improve shake and the tailing problem of motion picture.Estimation is one of its key technology, correct motion estimation result The interpolated frame of generation can give good visual experience, and the motion estimation result of mistake can influence viewing effect on the contrary, than original Begin to shake and bring even worse viewing effect.The effect of estimation associates greatly with the scene content of primitive frame, there is many scenes Correct motion search result can not be obtained under in general MEMC technologies, for example compound movement, wisp motion, is faded in light Go out scene etc..
Prior art is realized, the result compensated according to motion estimation motion, judges to transport under the scene by various indexs Whether dynamic estimation fails, and obtained motion vector is simultaneously unreliable, and then interpolated frame no longer uses the result of motion compensation, but selects Select primitive frame or primitive frame weighted blend (blending) result.But some scenes only have that local motion estimation effect is bad, After whole frame fallback (moving back as original input frame or primitive frame weighted results, rather than the interpolation result of motion estimation motion compensation) Although avoid that wrong motion estimation may bring is broken, visual experience is still not good enough.
The content of the invention
The present invention is for technical problem present in above-mentioned prior art, there is provided local motion vector in a kind of MEMC technologies Modification method, it can be replaced when motion search is local not good enough by using the relatively good motion vector on periphery bad Motion vector, although being also not necessarily correct motion vector even if the motion vector replaced, sweared using smooth motion The interpolated frame of generation is measured, visual effect is preferable, easily covers phenomena such as picture is broken.
To reach above-mentioned purpose, the technical solution adopted in the present invention is as follows:
Local motion vector modification method in a kind of MEMC technologies, including step are as follows:
1) some FX scope around it is specified to each block;
2) accuracy of all block motion vector and the motion vector in the range of this is obtained;
3) selection calculates an accurately and reliably motion vector conduct replacement motion vector from those motion vectors;
4) whether accurately and reliably the motion vector that analysis judges current block, calculates a weighted average coefficients;
5) with the weighted average coefficients by the motion vector of block scripts and its replace motion vector weight out one it is new Final motion vector of the motion vector as current block.
The step 1), MEMC technologies are based on block and carry out estimation, and image is split by block, specify current calculation block One region for containing N × M block of surrounding, the region should also contains current block, and N, M are integer, realized according to hardware, Algorithm requirement is selected.
Described piece of Similarity matching degree is carried out according to the index of the similarity degree for the corresponding blocks that can show that motion vector Judge.
The index includes absolute difference and criterion, mean-square error criteria, or according to position, candidate motion vector type, A penalty coefficient is added to the result of calculation of absolute difference sum criterion or mean-square error criteria, obtained basis for estimation refers to Mark.
The absolute difference and criterion, i.e. sad value, sad value is smaller, represents motion vector more accurately and reliably, and sad value is bigger, Block- matching is poorer, and the accuracy of the motion vector is poorer, and its calculation formula is as follows:
SAD=Σ (| pa(i)-pb(i)|)
PaAnd P (i)b(i) the ith pixel point in the block a and block b being compared is referred to.
Accurately and reliably motion vector refers in the scope represented by N × M block, and block Similarity matching degree is relatively good, can With the vector of the approximate representation range motion.
Selection block matching degree best one or several motion vectors from N × M motion vector, final weighting meter A motion vector is calculated as replacement motion vector.
In N × M vector, compare the value being calculated by above-mentioned error criterion, select X smaller vector of value, it is accurate Really reliable motion vector refers to the vector that X vector weighting obtains, and X can be any integer less than N × M, X=1,2, 3,...。
Step 4) the analysis judges whether current block motion vector is reliable, is to refer to the part comprising current block What the convergence situation in zonule determined, this region can be the scope for including a Z × Z-block centered on current block; What weighted average coefficients reacted is exactly the convergence situation of this zonule.
Matching criterior error meets the number of the block of certain condition, or often row minimum match error is cumulative and can table Show the convergence situation in region, map to obtain weighted average coefficients with these indexs.
In the step 5), weighted average coefficients can be mapped to obtain by the index of conversion zone convergence situation (for example to be passed through One-dimensional linear curve is mapped), gained coefficient should be an integer, be easy to two motion vectors to be weighted averaging, when Mapping relations determine that the scope of coefficient is also determined;
Final motion vector=(Vec × (Range-alpha)+FB_Vec × alpha)/Range
Vec refers to the motion vector of block script, and FB_Vec refers to the reliable motion vector of above-mentioned calculating, that is, replaces motion arrow Amount, alpha refer to weighted average coefficients, and Range refers to the effective range of coefficient;
During alpha=0, the vector of illustrated block script is credible preferably, without changing;Alpha is bigger, shows block script vector It is more insincere, go weighting to obtain a new preferable vector with another vector.
The present invention uses above-mentioned technical scheme, caused to have the beneficial effect that:
MEMC technologies have limitation under many scenes, and the motion vector faulty convergence of indivedual blocks can cause local break The problems such as broken, shake, influence viewing experience, technical solution of the present invention, to those bad motion vectors of effect always, find week Side restrains correct motion vector and goes to replace it, also may not be accurate although replacing vector, what smooth motion vector obtained Interpolated frame, phenomena such as obvious interpolation can be avoided to crush, improve visual experience.The technology can be classified as a kind of regional degeneration's plan Slightly, when partial result is bad, it is not necessary to which whole frame picture degenerates to primitive frame, is avoided with shake local broken, but uses and puts down Sliding vector field reduces the problem of local broken.
Brief description of the drawings
The detailed description made by reading with reference to the following drawings to non-limiting example, further feature of the invention, Objects and advantages will become more apparent upon:
Fig. 1 is the flow chart of method provided by the present invention.
Embodiment
With reference to specific embodiment, the present invention is described in detail.Following examples will be helpful to the technology of this area Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill to this area For personnel, without departing from the inventive concept of the premise, various modifications and improvements can be made.These belong to the present invention Protection domain.
As shown in figure 1, local motion vector modification method in MEMC technologies provided by the present invention, including step are as follows:
1) each block specifies some FX scope around it;
2) accuracy of all block motion vector and the vector in the range of this, such as the SAD of motion vector are obtained, The SAD of motion vector shows the matching degree of block, and SAD is bigger, and Block- matching is poorer, and it is not motion good enough to illustrate the MV Vector;
3) relatively reliable motion vector in the region is found out, for example motion vector corresponding to minimum SAD, or per a line Motion vector corresponding to minimum SAD;
4) reliable motion vector is will be considered to as replacing motion vector, if reliable motion vectors have multiple, can be calculated Go out their weighted value as replacement motion vector;
5) analysis judges whether the motion vector of the block is reliable, obtains a weighted average coefficients, weighted average coefficients It is bigger, show that the motion vector is more unreliable, moved the motion vector of block scripts and its replacement with the weighted average coefficients Vector weighting goes out final motion vector of the new motion vector as current block.
Embodiment:
1) FX is selected, centered on current block, includes the scope of 5x5 block;
2) motion vector and its SAD of this 25 blocks are obtained, weighs the accuracy of motion vector in this example with SAD;
3) this 25 motion vectors are screened, selects reliable motion vector;One threshold value, SAD can be set It is worth small motion vector and is regarded as reliable motion vector;Selection mode is not limited to method cited in the present embodiment;
4) motion vector obtained in step 3) can be weighted according to reliability, obtains final replacement vector;
5) judge whether the motion vector of current block is reliable, the judgement is with reference to the receipts in the local cell domain comprising current block Hold back situation decision;
6) it is 7x7 to assume the local cell domain, counts the convergence situation of this 49 blocks, for example counts SAD and is both less than one The number of the block of thresholding, the convergence situation in 7x7 regions is reflected by the number;The judgement of the convergence situation is not limited to recited herein Example;
7) weighted average coefficients, i.e. fallback level are mapped out by the convergence situation in 7x7 regions in 6);
8) with the weighted average coefficients fallback level in 7), the original vector of block and 4) in obtained replacement swear Amount is weighted, that is, obtains the final motion vector of block.
The specific embodiment of the present invention is described above.It is to be appreciated that the invention is not limited in above-mentioned Particular implementation, those skilled in the art can make various deformations or amendments within the scope of the claims, this not shadow Ring the substantive content of the present invention.

Claims (9)

1. local motion vector modification method in a kind of MEMC technologies, it is characterised in that as follows including step:
1)Some FX scope around it is specified to each block;
2)Obtain the accuracy of all block motion vector and the motion vector in the range of this, i.e. the block of motion vector is similar Matching degree;
3)Motion vector corresponding to minimum SAD in the region is found out, or per motion vector corresponding to a line minimum SAD;It will recognize For reliable motion vector as replace motion vector, if reliable motion vectors have it is multiple, calculate they weighted value make To replace motion vector;
4)Judge whether the motion vector of current block is reliable, the judgement is with reference to the convergence shape in the local cell domain comprising current block What condition determined, a weighted average coefficients are mapped out by the convergence situation in local cell domain;
5)The motion vector of block scripts and its replacement motion vector are weighted into out a new motion with the weighted average coefficients Final motion vector of the vector as current block.
2. local motion vector modification method in MEMC technologies according to claim 1, it is characterised in that the step 1) Middle FX scope is a regional extent for containing N × M block around current block, and the regional extent, which contains, to be worked as Preceding piece, N and M are integer.
3. local motion vector modification method in MEMC technologies according to claim 2, it is characterised in that the step 2) In, the accuracy of motion vector refers to:In scope represented by N × M block, the Similarity matching degree of block preferably, can be approximate Represent the vector of the range of motion.
4. local motion vector modification method in MEMC technologies according to claim 3, it is characterised in that described piece similar Matching degree is judged according to the index of the similarity degree for the corresponding blocks that can show that motion vector.
5. local motion vector modification method in MEMC technologies according to claim 4, it is characterised in that the index bag Include absolute difference and criterion or mean-square error criteria.
6. local motion vector modification method in MEMC technologies according to claim 5, it is characterised in that the absolute difference And criterion, i.e. sad value, sad value is smaller, represents motion vector more accurately and reliably, sad value is bigger, and Block- matching is poorer, the motion The accuracy of vector is poorer, and its calculation formula is as follows:
PaAnd P (i)b(i) the ith pixel point in the block a and block b being compared is referred to.
7. local motion vector modification method in MEMC technologies according to claim 3, it is characterised in that the step 3) In, selection block matching degree best one or several motion vectors, finally weight out one from N × M motion vector Motion vector is as replacement motion vector.
8. local motion vector modification method in MEMC technologies according to claim 3, it is characterised in that the step 4) In, whether accurately and reliably analysis judges current block motion vector, is with reference to the receipts in the local cell domain comprising current block Situation decision is held back, weighted average coefficients are the convergence situations for reacting the local cell domain, by the convergence shape in the local cell domain The index of condition maps to obtain, and the index of the convergence situation is the number for the block that matching error meets certain condition, or often goes Minimum match error is cumulative and to represent.
9. local motion vector modification method in MEMC technologies according to claim 8, it is characterised in that the step 5) In, final motion vector=(Vec × (Range-alpha)+FB_Vec × alpha)/Range,
Vec refers to the motion vector of block scripts, and FB_Vec refers to replacement motion vector, and alpha refers to weighted average coefficients, and Range refers to The effective range of coefficient.
CN201510203610.6A 2015-04-24 2015-04-24 Local motion vector modification method in MEMC technologies Active CN104811723B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510203610.6A CN104811723B (en) 2015-04-24 2015-04-24 Local motion vector modification method in MEMC technologies

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510203610.6A CN104811723B (en) 2015-04-24 2015-04-24 Local motion vector modification method in MEMC technologies

Publications (2)

Publication Number Publication Date
CN104811723A CN104811723A (en) 2015-07-29
CN104811723B true CN104811723B (en) 2018-03-16

Family

ID=53696152

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510203610.6A Active CN104811723B (en) 2015-04-24 2015-04-24 Local motion vector modification method in MEMC technologies

Country Status (1)

Country Link
CN (1) CN104811723B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106851047B (en) * 2016-12-30 2020-02-07 中国科学院自动化研究所 Method and system for detecting static pixel points in video image
CN107124572A (en) * 2017-03-08 2017-09-01 宏祐图像科技(上海)有限公司 The method and system of partial result amendment in motion estimation motion compensation
CN117834906A (en) * 2019-03-08 2024-04-05 华为技术有限公司 Motion vector refined search area

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1328405A (en) * 2000-06-13 2001-12-26 三星电子株式会社 Format converter using bidirectional motion vector and its method
CN101416523A (en) * 2006-04-03 2009-04-22 英特尔公司 Motion compensated frame rate conversion with protection against compensation artifacts
CN101883278A (en) * 2009-05-08 2010-11-10 联发科技股份有限公司 Motion vector correction apparatus and method thereof
CN102123283A (en) * 2011-03-11 2011-07-13 杭州海康威视软件有限公司 Interpolated frame acquisition method and device in video frame rate conversion
CN102131058A (en) * 2011-04-12 2011-07-20 上海理滋芯片设计有限公司 Speed conversion processing module and method of high definition digital video frame
CN102868879A (en) * 2011-07-05 2013-01-09 北京大学 Method and system for converting video frame rate

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8204104B2 (en) * 2006-03-09 2012-06-19 Sony Corporation Frame rate conversion system, method of converting frame rate, transmitter, and receiver

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1328405A (en) * 2000-06-13 2001-12-26 三星电子株式会社 Format converter using bidirectional motion vector and its method
CN101416523A (en) * 2006-04-03 2009-04-22 英特尔公司 Motion compensated frame rate conversion with protection against compensation artifacts
CN101883278A (en) * 2009-05-08 2010-11-10 联发科技股份有限公司 Motion vector correction apparatus and method thereof
CN102123283A (en) * 2011-03-11 2011-07-13 杭州海康威视软件有限公司 Interpolated frame acquisition method and device in video frame rate conversion
CN102131058A (en) * 2011-04-12 2011-07-20 上海理滋芯片设计有限公司 Speed conversion processing module and method of high definition digital video frame
CN102868879A (en) * 2011-07-05 2013-01-09 北京大学 Method and system for converting video frame rate

Also Published As

Publication number Publication date
CN104811723A (en) 2015-07-29

Similar Documents

Publication Publication Date Title
US9940719B2 (en) Alignment of an ordered stack of images from a specimen
CN102291578B (en) Apparatus and method for frame rate conversion
US8340186B2 (en) Method for interpolating a previous and subsequent image of an input image sequence
Kang et al. Motion compensated frame rate up-conversion using extended bilateral motion estimation
Yoo et al. Direction-select motion estimation for motion-compensated frame rate up-conversion
US8644387B2 (en) Motion estimation method
KR100676704B1 (en) Motion Estimator and Motion Estimating Method Thereof
US8773595B2 (en) Image processing
US20030206246A1 (en) Motion estimator for reduced halos in MC up-conversion
US8467453B2 (en) Motion vector calibration circuit, image generating apparatus and method thereof
JP2005287048A (en) Improvement of motion vector estimation at image border
TW201143431A (en) Frame rate conversion using motion estimation and compensation
US20100157162A1 (en) Image processing apparatus, image processing method and computer readable medium
CN104811723B (en) Local motion vector modification method in MEMC technologies
KR101050135B1 (en) Intermediate image generation method using optical information
US9106926B1 (en) Using double confirmation of motion vectors to determine occluded regions in images
US9894367B2 (en) Multimedia device and motion estimation method thereof
US8451900B2 (en) System and method for global motion estimation using profile matching
JP5197374B2 (en) Motion estimation
US9094561B1 (en) Frame interpolation and motion vector reconstruction
EP2237559B1 (en) Background motion estimate based halo reduction
CN109788297B (en) Video frame rate up-conversion method based on cellular automaton
Park et al. Adaptive weighting scheme for edge-based line interpolation
Bai et al. Visual-weighted motion compensation frame interpolation with motion vector refinement
Qian et al. Global motion estimation under translation-zoom ambiguity

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
EXSB Decision made by sipo to initiate 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: 20200115

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, 312, 316, 318.

Patentee before: Hongyou Image Technology (Shanghai) Co., Ltd.

TR01 Transfer of patent right