CN103139562B - Method for estimating and device - Google Patents
Method for estimating and device Download PDFInfo
- Publication number
- CN103139562B CN103139562B CN201110400975.XA CN201110400975A CN103139562B CN 103139562 B CN103139562 B CN 103139562B CN 201110400975 A CN201110400975 A CN 201110400975A CN 103139562 B CN103139562 B CN 103139562B
- Authority
- CN
- China
- Prior art keywords
- pixel
- macro block
- frame
- block
- present
- 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.)
- Expired - Fee Related
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A kind of method for estimating and device are disclosed. This method for estimating comprises: by respectively the pixel in pixel and reference frame in present frame being sampled, obtain the current picture corresponding with present frame and the reference picture corresponding with reference frame; By entirely searching in reference picture, find out the optimum movement vector of the each base pixel unit in current picture, as the optimum movement vector of many pixel precisions of the each macro block in present frame, wherein the each base pixel unit in current picture is corresponding with the each macro block in present frame respectively, and each base pixel unit in reference picture is corresponding with the each macro block in reference frame respectively.
Description
Technical field
The present invention relates to image processing field, relate more specifically to a kind of method for estimating and device.
Background technology
Estimation (MotionEstimation) is Video processing (for example, Video coding, videoDeinterleave) in widely used a kind of technology. The basic thought of estimation is by image sequenceEach frame is divided into the macro block of many non-overlapping copies, and thinks the displacement of all pixels in each macro blockIdentical, then in a certain specific hunting zone of reference frame, find out according to certain matching criterior and work asThe piece (, the match block of current macro) that front macro block is the most similar, match block is relative with current macroDisplacement is the motion vector of current macro.
In the time of video compress, only need preserve motion vector and the residual error number of current macro with respect to its match blockAccording to, just can in the time of video decompression, recover current macro completely. Conventionally in Video processing, needWithin the scope of the very large search of reference frame, carry out motion search and just can find out the optimal movement arrow of current macroAmount (, finding out the match block the most similar to current macro). In order to obtain the good speed of current macroDynamic vector need to entirely be searched in hunting zone. The complete all possible motion vector of searchCandidate, and the most accurate result is provided, but the processing time is very long.
Summary of the invention
In view of above problem, the invention provides a kind of method for estimating and device of novelty.
Comprise according to the method for estimating of the embodiment of the present invention: by respectively to the pixel in present frameSample with the pixel in reference frame, obtain the current picture corresponding with present frame and with reference frame pairThe reference picture of answering; By entirely searching for, find out the each base in current picture in reference pictureThe optimum movement vector of this pixel cell, as many pixel precisions of the each macro block in present frameGood speed dynamic vector, wherein the each base pixel unit in current picture respectively with present frame in eachMacro block is corresponding, and each base pixel unit in reference picture respectively with reference frame in eachMacro block is corresponding.
Comprise according to the movement estimation apparatus of the embodiment of the present invention: pixel sampling unit, divides for passing throughOther pixel in pixel and reference frame in present frame is sampled, obtain work as corresponding with present frameFront picture and the reference picture corresponding with reference frame; The first search unit, for passing through at reference pictureIn entirely search for, find out the optimum movement vector of the each base pixel unit in current picture, doFor the optimum movement vector of many pixel precisions of the each macro block in present frame, wherein in current pictureEach base pixel unit is corresponding with the each macro block in present frame respectively, and in reference pictureEach base pixel unit is corresponding with the each macro block in reference frame respectively.
By the present invention, can within the very short time, find out present frame in many pixels of each macro blockThe optimum movement vector of precision.
Brief description of the drawings
Below in conjunction with accompanying drawing to understanding better this in the description of the specific embodiment of the present inventionInvention, wherein:
Fig. 1 shows the example that reference frame/present frame is sampled;
Fig. 2 shows under 16 × 16 interframe encoding mode for searching for any one of present frameThe reference zone of the optimum movement vector of the whole pixel precision of individual macro block;
Fig. 3 shows according to the block diagram of the movement estimation apparatus of the embodiment of the present invention; And
Fig. 4 shows according to the flow chart of the method for estimating of the embodiment of the present invention.
Detailed description of the invention
Feature and the exemplary embodiment of various aspects of the present invention will be described in detail below. Description belowMany details are contained, to complete understanding of the present invention is provided. But, for this areaTechnical staff it is evident that, the present invention can do not need in these details someIn the situation of details, implement. Below the description of embodiment is only used to by of the present invention showing is shownExample provides the clearer understanding to the present invention. The present invention is never limited to propose any concrete belowConfiguration and algorithm, but covered under the premise of without departing from the spirit of the present invention coherent element, partsAny amendment, replacement and improvement with algorithm.
Thereby find out the best of current macro in order to carry out full search on reference frame within the limited timeMotion vector, the present invention proposes a kind of all direction search method fast.
According to one embodiment of present invention, in order to carry out full search fast, first need reference frameSample with present frame. Fig. 1 shows the example that reference frame/present frame is sampled. Wherein,Frame of pixels shown in Fig. 1 can be counted as reference frame, also can be counted as present frame.
As shown in Figure 1, the pixel in each 4 × 4 in reference frame/present frame is sampled,To obtain reference picture/current picture. Wherein, in each 4 × 4 in reference frame/present frameThe pixel of an ad-hoc location is sampled; And the picture of all 4 × 4 in reference frame/present frameElement sampling location is all identical. For example, in Fig. 1, illustrated each 4 in reference frame/present frameThe first row of × 4, the pixel of first row are sampled. But those skilled in the art should be brightIn vain, the pixel sample locations of each 4 × 4 is not limited to the position shown in Fig. 1, and can be pre-Any position of first setting. In addition, the pixel in pixel sample locations and the present frame in reference frame is adoptedSample position is identical.
The reference picture obtaining through the sampling shown in Fig. 1 and the size of current picture be reference frame and1/16 (1/4 width × 1/4 height) of present frame. So, in reference picture, carry out full search fromAnd obtain in current picture each 4 × 4 (in current picture each 4 × 4 with present frame inEach macro block corresponding) optimum movement vector only need for (this processing time in processing time of 1/16It is the best that draws the whole pixel precision of the each macro block in present frame by reference frame being carried out to full searchThe time that motion vector is required).
Obviously the reference picture, drawing by the sampling process based on shown in Fig. 1 and current picture carry outIn the current picture that full search draws, the optimum movement vector of each 4 × 4 is the fortune of 4 pixel precisionsDynamic vector. And in actual image is processed, in order to ensure the quality of estimation, should carry out wholeThe motion search of pixel precision. So, need to be to look for according to the motion vector of aforementioned 4 pixel precisionsIn larger reference zone centered by the reference zone going out, carry out full search.
For example,, for 16 × 16 interframe encoding mode, for any one in present frameMacro block B, need to be according to the fortune of 4 pixel precisions of 4 × 4 piece bs corresponding with macro block B in current pictureDynamic vector is found out the reference zone R corresponding with macro block B in reference frame, and taking reference zone R asIn 24 × 24 the region at center, entirely search for, thereby the whole pixel precision that draws macro block BGood speed dynamic vector. Fig. 2 shows under 16 × 16 interframe encoding mode for searching for present frameThe reference zone of optimum movement vector of whole pixel precision of any one macro block B. As Fig. 2 instituteShow, vowing according to the motion of 4 pixel precisions of 4 × 4 b corresponding with macro block B in current pictureAmount exists respectively find out corresponding with macro block B 16 × 16 reference zone R in reference frame after16 × 16 reference zone R is expanded 4 pixels by upper and lower, left and right four direction, get final productTo corresponding with macro block B 24 × 24 the region for full search.
For example,, for 16 × 8 interframe encoding mode, for any one from present frame16 × 8 B ' that macro block B is partitioned into, need in current picture corresponding with macro block B 4 × 4The motion vector of 4 pixel precisions of piece b is found out and 16 × 8 reference areas that B ' is corresponding in reference frameTerritory R, and entirely search in 24 × 16 region centered by reference zone R, therebyGo out the optimum movement vector of the whole pixel precision of 16 × 8 B '. Similar with the process shown in Fig. 2,Joining according to the motion vector of 4 pixel precisions of 4 × 4 b corresponding with macro block B in current pictureAfter examining the reference zone R that finds out corresponding with 16 × 8 B ' 16 × 8 in frame, respectively upper,Under, left and right four direction is 4 pixels of 16 × 8 reference zone R expansion, can obtain and this24 × 16 the region for full search that 16 × 8 B ' are corresponding.
For other interframe encoding modes, the rest may be inferred. That is to say, for certain interframe encode mouldFormula, vowing according to the motion of 4 pixel precisions of 4 × 4 b corresponding with macro block B in current pictureAmount is found out certain the big or small sub-block being partitioned into from macro block B according to this interframe encoding mode in reference frameAfter the reference zone R of B ', expand with reference to region R at upper and lower, left and right four direction respectively4 pixels, can obtain the region for full search corresponding with sub-block B '.
At the optimum movement vector that draws the whole pixel precision of each seed block under various interframe encoding modesAfter, calculate the coding cost for macro block B under various interframe encoding modes, thereby find out coding generationA kind of interframe encoding mode of valency minimum is as the best interframe encoding mode of macro block B.
Fig. 3 shows according to the block diagram of the movement estimation apparatus of the embodiment of the present invention. Fig. 4 shows rootAccording to the flow chart of the method for estimating of the embodiment of the present invention. Below in conjunction with Fig. 3 and Fig. 4, retouch in detailState according to the movement estimation apparatus of the embodiment of the present invention and method.
As shown in Figure 4, this movement estimation apparatus comprises pixel sampling unit 302 and the first search unit304. Wherein, pixel sampling unit 302 is by respectively in the pixel in present frame and reference framePixel is sampled, and obtains the current picture corresponding with present frame and the reference picture corresponding with reference frame(S402); The first search unit, by entirely searching in reference picture, is found out in current pictureThe optimum movement vector of each base pixel unit, as many pixels of the each macro block in present frameThe optimum movement vector (S404) of precision. Wherein, the each base pixel unit in current picture dividesNot corresponding with the each macro block in present frame, and each base pixel unit in reference frame is respectivelyCorresponding with the each macro block in reference frame.
In one embodiment, pixel sampling unit 302 can be by the sampling of describing in conjunction with Fig. 1Journey is sampled to the pixel in present frame and reference frame. , pixel sampling unit 302 can be to ginsengExamine the picture of an ad-hoc location in each 4 × 4 that each macro block in frame and present frame comprisesElement is sampled, with draw the reference picture corresponding with reference frame and with present frame in corresponding current pictureFace. Wherein, the size of reference picture is reference frame 1/16, and the size of current picture is present frame1/16, each 4 × 4 (, base pixel unit) in reference picture respectively with reference frame inEach macro block correspondence, each 4 × 4 (, base pixel unit) in current picture respectively withEach macro block correspondence in present frame.
Then, the first search unit 304 can be by entirely searching in reference picture, finds out and work asThe optimum movement vector of each 4 × 4 in front picture. Here, the first search unit 304 canRegard the optimum movement vector of each 4 × 4 in current picture as each macro block in current pictureThe optimum movement vector of 4 pixel precisions.
Then, in specific interframe encoding mode, in order to obtain according to this interframe encoding mode from working asThe optimal movement of the whole pixel precision of any one sub-block that any one macroblock partition in front frame goes out is vowedAmount, further comprises regional search unit 306 and the according to the movement estimation apparatus of the embodiment of the present inventionTwo search units 308.
Wherein, for any one the macro block B in present frame, regional search unit 306 is according to currentThe optimum movement vector of the base pixel unit b corresponding with macro block B in picture is looked in reference frameGo out the reference corresponding with any one the sub-block B ' being partitioned into from macro block B according to specific interframe encoding modeRegion; The second search unit 308 is expanded the reference zone corresponding with sub-block B ', and in expansionAfter reference zone in carry out full search to find out the optimum movement vector of whole pixel precision of sub-block B '.Wherein, regional search unit 306 can utilize the process of describing in conjunction with Fig. 2 to process.In sum, the present invention finds out current picture by the reference picture and the current picture that draw based on samplingIn the optimum movement vector of each base pixel unit, then utilize each base pixel in current pictureThe optimum movement vector of unit finds out that in reference frame, to carry out full search each grand to draw in present frameThe reference zone of the optimum movement vector of the whole pixel precision of piece, has accelerated processing speed, has reduced and has locatedThe reason time, thus hardware resource saved.
Below with reference to specific embodiments of the invention, the present invention has been described, but art technologyPersonnel all understand, and can carry out various amendments, combination and change to these specific embodiments, and can notDepart from the spirit and scope of the present invention that limited by claims or its equivalent.
Can perform step with hardware or software as required. Note, do not departing from the scope of the inventionPrerequisite under, can add step, therefrom remove step or amendment to the flow chart providing in this descriptionStep wherein. In general, flow chart is just used to refer to for the basic operation of practical functionA kind of possible sequence.
Embodiments of the invention can utilize programming general purpose digital computer, utilize special IC,PLD, field programmable gate array, light, chemistry, biological, quantum orSystem, assembly and the mechanism of nanometer engineering realize. In general, function of the present invention can be by abilityThe known any means in territory realize. Can use distributed or networked system, assembly and circuit. NumberAccording to communication or to transmit can be wired, wireless or by any other means.
Also will recognize, according to the needs of application-specific, in the key element shown in accompanying drawing one or manyIndividual can by more separate or more integrated mode realize, or be removed even in some cases orBe deactivated. The journey logic bomb that realization can be stored in machine readable media is carried out with permission computerState any method, also within the spirit and scope of the present invention.
In addition, it is only exemplary that any signal arrows in accompanying drawing should be considered to, instead of restrictionProperty, unless separately there is concrete instruction. When being also contemplated as, term make the ability of separation or combination unclearTime, the combination of assembly or step also will be considered to record.
Claims (8)
1. a method for estimating, comprising:
By respectively the pixel in pixel and reference frame in present frame being sampled, obtain with currentThe current picture that frame is corresponding and the reference picture corresponding with reference frame;
By entirely searching for, find out each basic in described current picture in described reference pictureThe optimum movement vector of pixel cell, as many pixel precisions of the each macro block in described present frameOptimum movement vector, wherein
Each base pixel unit in described current picture respectively with described present frame in each macro blockCorresponding, and each base pixel unit in described reference picture respectively with described reference frame inEach macro block is corresponding:
For any one macro block in described present frame, according in described current picture with described oneThe optimum movement vector of the base pixel unit that individual macro block is corresponding is found out and basis in described reference frameThe corresponding reference zone of each sub-block that specific interframe encoding mode goes out from a described macroblock partition;And
For any one sub-block going out from a described macroblock partition according to described specific coding pattern,The reference zone corresponding with a described sub-block expanded, and in reference zone after expansionCarry out the complete optimum movement vector of searching for the whole pixel precision to find out a described sub-block.
2. method for estimating according to claim 1, is characterized in that, to described present frameIn each macro block in the pixel of fixed position sample, and each grand in described reference frameThe pixel of the fixed position in piece is sampled.
3. method for estimating according to claim 2, is characterized in that, to described present frameIn each macro block comprise each 4 × 4 in the pixel of a fixed position sample,And a fixed position in the each macro block in described reference frame is comprised each 4 × 4Pixel is sampled.
4. according to the method for estimating described in any one in claims 1 to 3, it is characterized in that,The pixel sample locations of described present frame is identical with the pixel sample locations of described reference frame.
5. a movement estimation apparatus, comprising:
Pixel sampling unit, for entering by the pixel in pixel and the reference frame to present frame respectivelyRow sampling, obtains the current picture corresponding with present frame and the reference picture corresponding with reference frame;
The first search unit, for by entirely searching at described reference picture, finds out described working asThe optimum movement vector of the each base pixel unit in front picture, each as in described present frameThe optimum movement vector of many pixel precisions of macro block, wherein
Each base pixel unit in described current picture respectively with described present frame in each macro blockCorresponding, and each base pixel unit in described reference picture respectively with described reference frame inEach macro block is corresponding:
Regional search unit, for any one macro block for described present frame, works as according to describedThe optimum movement vector of the base pixel unit corresponding with a described macro block in front picture, describedIn reference frame, find out and the each sub-block going out from a described macroblock partition according to specific interframe encoding modeCorresponding reference zone; And
The second search unit, for what go out from a described macroblock partition according to described specific coding patternAny one sub-block, expands the reference zone corresponding with a described sub-block, and after expansionReference zone in carry out full search and vow with the optimal movement of the whole pixel precision of finding out a described sub-blockAmount.
6. movement estimation apparatus according to claim 5, is characterized in that, described pixel samplingSample to the pixel of the fixed position in the each macro block in described present frame in unit, and to describedThe pixel of the fixed position in the each macro block in reference frame is sampled.
7. movement estimation apparatus according to claim 6, is characterized in that, described pixel samplingA fixed position in each 4 × 4 that unit comprises the each macro block in described present framePixel sample, and in the each macro block in described reference frame is comprised each 4 × 4The pixel of a fixed position sample.
8. according to the movement estimation apparatus described in any one in claim 5 to 7, it is characterized in that,The pixel sample locations of described present frame is identical with the pixel sample locations of described reference frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110400975.XA CN103139562B (en) | 2011-11-30 | 2011-11-30 | Method for estimating and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110400975.XA CN103139562B (en) | 2011-11-30 | 2011-11-30 | Method for estimating and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103139562A CN103139562A (en) | 2013-06-05 |
| CN103139562B true CN103139562B (en) | 2016-05-04 |
Family
ID=48498797
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110400975.XA Expired - Fee Related CN103139562B (en) | 2011-11-30 | 2011-11-30 | Method for estimating and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103139562B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20170026334A (en) | 2014-07-06 | 2017-03-08 | 엘지전자 주식회사 | Method for processing video signal, and apparatus therefor |
| CN110545428B (en) * | 2018-05-28 | 2024-02-23 | 深信服科技股份有限公司 | Motion estimation method and device, server and computer readable storage medium |
| CN112738517B (en) * | 2019-10-14 | 2022-03-01 | 珠海格力电器股份有限公司 | Motion estimation search method, device, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1756355A (en) * | 2004-09-29 | 2006-04-05 | 腾讯科技（深圳）有限公司 | Motion estimating method in video data compression |
| CN1852442A (en) * | 2005-08-19 | 2006-10-25 | 深圳市海思半导体有限公司 | Layering motion estimation method and super farge scale integrated circuit |
| CN101720039A (en) * | 2009-09-08 | 2010-06-02 | 广东工业大学 | Diamond search-based multi-resolution quick motion estimation method |
| CN101945284A (en) * | 2010-09-29 | 2011-01-12 | 无锡中星微电子有限公司 | Motion estimation device and method |
| WO2011142644A2 (en) * | 2010-05-14 | 2011-11-17 | 삼성전자 주식회사 | Method for encoding and decoding video and apparatus for encoding and decoding video using expanded block filtering |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100644629B1 (en) * | 2004-09-18 | 2006-11-10 | 삼성전자주식회사 | Method for estimating motion based on hybrid search block matching algorithm and frame-rate converter using thereof |
| CN100341334C (en) * | 2005-01-14 | 2007-10-03 | 北京航空航天大学 | Multi-reference frame rapid movement estimation method based on effective coverage |
| TW200838321A (en) * | 2007-03-02 | 2008-09-16 | X8 Technology Inc | Motion estimation method and system with dual search windows for high resolution video coding |
| KR100926752B1 (en) * | 2007-12-17 | 2009-11-16 | 한국전자통신연구원 | Fine Motion Estimation Method and Apparatus for Video Coding |
| CN101227614B (en) * | 2008-01-22 | 2010-09-08 | 炬力集成电路设计有限公司 | Motion estimation device and method of video coding system |
-
2011
- 2011-11-30 CN CN201110400975.XA patent/CN103139562B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1756355A (en) * | 2004-09-29 | 2006-04-05 | 腾讯科技（深圳）有限公司 | Motion estimating method in video data compression |
| CN1852442A (en) * | 2005-08-19 | 2006-10-25 | 深圳市海思半导体有限公司 | Layering motion estimation method and super farge scale integrated circuit |
| CN101720039A (en) * | 2009-09-08 | 2010-06-02 | 广东工业大学 | Diamond search-based multi-resolution quick motion estimation method |
| WO2011142644A2 (en) * | 2010-05-14 | 2011-11-17 | 삼성전자 주식회사 | Method for encoding and decoding video and apparatus for encoding and decoding video using expanded block filtering |
| CN101945284A (en) * | 2010-09-29 | 2011-01-12 | 无锡中星微电子有限公司 | Motion estimation device and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103139562A (en) | 2013-06-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10728573B2 (en) | Motion compensated boundary pixel padding | |
| US11057639B2 (en) | Derivation method and apparatuses with candidate motion vectors | |
| US20180278951A1 (en) | Motion vector difference (mvd) prediction | |
| CN110121883B (en) | Method and apparatus for decoding image in image coding system | |
| CN113196749B (en) | Method and apparatus for decoding video data | |
| GB2589241A (en) | Update of look-up tables | |
| CN104584558B (en) | Motion vector for the inter-view prediction of 3D videos | |
| US20160337662A1 (en) | Storage and signaling resolutions of motion vectors | |
| JP2022507683A (en) | Decoder side motion vector improvement | |
| CN110383839A (en) | Affine motion information export | |
| CN105917648A (en) | Intra block copy prediction with asymmetric partitions and encoder-side search patterns, search ranges and approaches to partitioning | |
| TW201813396A (en) | Pattern-based motion vector derivation for video coding | |
| CN106576171A (en) | Unify intra block copy and inter prediction | |
| RU2017133236A (en) | DISPLAYING THE MOTION VECTOR FOR VIDEO CODING | |
| CN103609125A (en) | Method and apparatus for encoding and decoding motion vector of multi-view video | |
| CN106797476A (en) | Frame in BC and interframe are unified | |
| CN103139562B (en) | Method for estimating and device | |
| US20200260109A1 (en) | Image processing apparatus and image processing method | |
| CN103493492A (en) | Method and apparatus for encoding and decoding multi-view video | |
| US20150288983A1 (en) | Method and apparatus for processing video signal | |
| CN107360433A (en) | A kind of interframe prediction encoding method and device | |
| CN104380744A (en) | Method and device for processing video signal | |
| US10368087B2 (en) | Dynamic reload of video encoder motion estimation search window under performance/power constraints | |
| CN104980737B (en) | Inter-frame mode selecting method | |
| CN100366092C (en) | Search method for video frequency encoding based on motion vector prediction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160504 Termination date: 20181130 |