CN101494757A - Motion estimating method based on time-space domain mixing information - Google Patents
Motion estimating method based on time-space domain mixing information Download PDFInfo
- Publication number
- CN101494757A CN101494757A CNA2009100458485A CN200910045848A CN101494757A CN 101494757 A CN101494757 A CN 101494757A CN A2009100458485 A CNA2009100458485 A CN A2009100458485A CN 200910045848 A CN200910045848 A CN 200910045848A CN 101494757 A CN101494757 A CN 101494757A
- Authority
- CN
- China
- Prior art keywords
- search
- piece
- match point
- current block
- optimal match
- 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.)
- Pending
Links
Images
Abstract
The present invention provides a motion estimating method based on time-space domain mixed message, which uses the time-space domain content information to judge the motion speed and motion direction of the current coding block, and selects different search templates based on different movement intensity, so that the performance of the searching algorithm is further enhanced. The invention not only avoid a local optimum, but also effectively reduces the computing complexity.
Description
Technical field
The present invention relates to the digital video coding technology, relate in particular to a kind of method for estimating based on time-space domain mixing information.
Background technology
In existing video coding, most of method for video coding all are to reach compression effects by the redundancy of seeking on the room and time.In time-domain, the video of continuous several frames has very strong correlation usually, particularly when the time domain sample rate especially true when being very high.In spatial domain, be to be mutually related between the pixel sampling point usually, such as, very close between the adjacent image point.The time domain redundancy is mainly eliminated by estimation and motion compensation, and estimation also is a whole encoder part the most consuming time.Estimation is exactly to find one optimal to predict current block in the search window of reference frame.At present, that search precision is the highest in the block matching motion algorithm for estimating is full search method FS (Full Search Method), but its amount of calculation is big, is not suitable for real-time application.In order to improve search speed, some fast search process have been proposed again, as: three step search methods (TSS), two dimensional logarithmic method (LOGS), cross search method (CS), new three step search methods (NTSS), four step search methods (FSS), gradient descent method (BBGDS), diamond search (ds) (DS) and hexagonal search method (HEXBS) etc.Though these methods are greatly improved than FS aspect search speed, the following problem of ubiquity: be absorbed in local optimum during (1) search easily; (2) can not satisfy the video sequence of different motion situation simultaneously.Had by the motion estimation algorithm of H.264/AVC being adopted at present: EPZS algorithm, UMHexagonS algorithm and improvement algorithm Simplified UMHexagonS thereof; And the multi-level hexagonal point search of asymmetric cross algorithm (UMHexagonS), these algorithms have good distortion performance and higher estimation speed, picture quality is very close with full search method, but for the real-time encoder of D1 form, its implementation complexity is too high.In addition in actual applications, the image motion amount varies, and is very little or when being zero, most of zone of video image does not change or changes very little when motion vector, when using above single search strategy, may cause great search redundancy, reduction search efficiency.
Summary of the invention
The technical problem that the present invention solves provides a kind of method for estimating based on time-space domain mixing information, described method utilizes the time-space domain content information to judge the motion speed and the direction of motion of present encoding piece, and according to different motion intensities, select different search patterns, the performance of searching algorithm is further enhanced.
In order to solve the problems of the technologies described above, the present invention has adopted following technological means: a kind of method for estimating based on time-space domain mixing information, described method is earlier current encoded frame to be divided into non-overlapping copies, the piece of fixed size, then at each current block that will be encoded, in a region of search of reference frame, search out piece with current block matching error minimum, it is best matching blocks, at last according to the best matching blocks loca, it is the position of optimal match point, the relative displacement of output best matching blocks and current block, be motion vector, described search procedure is achieved in that
Step 1, with the left piece C1 in the adjacent block of current block C, go up piece C2, upper left C3 and upper right C4, and corresponding blocks E and the adjacent right piece E1, down piece E2, lower-left piece E3 and the bottom right piece E4 totally 9 piece pieces of forming 3*3s of current block in reference frame, be the time-space domain block of information of current block, wherein piece C1, piece C2, piece C3 and piece C4 represent the spatial information (si) of current block, and piece E, piece E1, piece E2, piece E3 and piece E4 represent the time-domain information of current block; Calculate the minimum average B configuration absolute difference MAD value of current block C and corresponding blocks E;
Step 2, setting threshold T1, the size of MAD value and T1 relatively when MAD＜T1, judges that current block is to be similar to static block, execution in step 3, otherwise be moving mass, execution in step 4;
Step 3, employing diamond search (ds) carry out the optimum Match point search, output optimal match point and coordinate thereof;
Step 4, in 9 pieces of time-space domain block of information, find out the best matching blocks of current block C, the motion vector mv=(m of current block and best matching blocks
x, m
y); The mould value of calculating kinematical vector mv | mv|;
Step 6, adopt the approximate octagon search method of tri-layer to carry out the optimum Match point search, and judge, when optimal match point is in the approximate octagon of tri-layer first ring from inside to outside or on the ring, export optimal match point and coordinate thereof according to Search Results; Otherwise, when optimal match point is on approximate octagonal second ring of tri-layer, adopt the melee search method to search for, output optimal match point and coordinate thereof; Otherwise, when optimal match point is on approximate octagonal the 3rd ring of tri-layer, adopt the melee search method to search for again, output optimal match point and coordinate thereof;
Further, described T1 span is [2,3]; Described T2 is an integer, and span is [70,80].
Owing to adopted above method for estimating, compared with prior art have the following advantages: (one) finds out in the block of information of time-space domain and the most similar piece of current block motion, and judge the motion conditions of present encoding piece in view of the above, adopt different searching methods, can significantly reduce the estimation time; (2) used 9 pieces of time-space domain block of information to search for the starting point prediction, can find out more accurately and the most similar piece of present encoding piece motion; (3) these two rules of motion that are far longer than other directions up and down with the probability of side-to-side movement according to the off-centring characteristic and the motion vector of motion vector, designed new search model, not only can avoid being absorbed in local optimum, and can find the BMP point with less point, reduced search point, particularly, show more outstanding in the mild zone of motion.
Description of drawings
Method for estimating of the present invention is provided in detail by following embodiment and accompanying drawing.
Fig. 1 is the search procedure schematic flow sheet of the embodiment of the invention;
Fig. 2 is the time-space domain block of information schematic diagram of the embodiment of the invention;
Fig. 3 is the diamond search (ds) model schematic diagram of the embodiment of the invention;
Fig. 4 is a multi-level approximate octagon schematic diagram in the embodiment of the invention.
Embodiment
Below will be described in further detail the method for estimating based on time-space domain mixing information of the present invention.
A kind of method for estimating of present embodiment based on time-space domain mixing information, described method is divided into non-overlapping copies with current encoded frame, size is the piece of 16*8, then at each piece that will be encoded, (region of search) searches out the piece with its matching error minimum in the certain limit of reference frame, it is best matching blocks, at last according to the position of best matching blocks loca (being optimal match point), the relative displacement of the current block in output best matching blocks and the current encoded frame, it is motion vector, described search procedure is achieved in that its realization flow as shown in Figure 1:
S1, with the left piece C1 in the adjacent block of current block C, go up piece C2, upper left C3 and upper right C4, and the corresponding blocks E of current block in reference frame (promptly in reference frame with the identical piece of present encoding piece coordinate) and adjacent right piece E1 thereof, piece E2, lower-left piece E3 and bottom right piece E4 totally 9 pieces of forming 3*3 down, be the time-space domain block of information of current block, as shown in Figure 2, wherein piece C1, piece C2, piece C3 and piece C4 represent the spatial information (si) of current block, and piece E, piece E1, piece E2, piece E3 and piece E4 represent the time-domain information of current block; Calculate the minimum average B configuration absolute difference MAD value of current block C and corresponding blocks E;
S2, utilize above MAD value to come the motion severe degree of every width of cloth image is identified, just can to think that current block is similar to static if MAD is enough little; Threshold value by setting MAD is T1, compares the size of MAD value and T1 then; Setting T1 in this example is 2, promptly when MAD＜2, can think that then current block is approximate static block, carries out S3, otherwise be moving mass, execution S4;
S3, employing diamond search (ds) carry out the optimum Match point search, and diamond search (ds) is exported optimal match point and coordinate thereof as shown in Figure 3;
S4, in the block of information of time-space domain, find out the best matching blocks of current block C in 9 pieces, the motion vector mv=(m that current block and optimum Match are fast
x, m
y); The mould value of calculating kinematical vector mv | mv|;
S5, setting threshold T2 are 70, relatively | the size of mv| and T2, when | mv|＜T2, judge that current block is the microinching piece, execution S6; Otherwise be the rapid movement piece, carry out S7;
S6, adopt the approximate octagon search method of tri-layer to carry out the optimum Match point search, and judge, when optimal match point is in multi-level approximate octagon first ring from inside to outside or on the ring, export optimal match point and coordinate thereof according to Search Results; Otherwise, when optimal match point is on multi-level approximate octagonal second ring, adopt the melee search method to search for, output optimal match point and coordinate thereof; Otherwise, when optimal match point is on multi-level approximate octagonal the 3rd ring, adopt the melee search method to search for again, output optimal match point and coordinate thereof;
S7, earlier adopt asymmetric cross search method to carry out the optimum Match point search, then with resulting optimal match point as starting point, adopt the approximate octagon search method search of four levels; And judge according to Search Results, when optimal match point is in multi-level approximate octagon first ring from inside to outside or on the ring, output optimal match point and coordinate thereof; Otherwise, when optimal match point is on multi-level approximate octagonal second ring, adopt the melee search method to search for, output optimal match point and coordinate thereof; Otherwise, carry out the search of hexagon search method earlier, and then adopt the melee search method to search for, output optimal match point and coordinate thereof.
Approximate octagon, hexagon and diamond ring and on search point as shown in Figure 4, wherein filled circles is represented the optimal match point that approximate octagon is searched for, black triangle is represented the optimal match point that hexagon search arrives.
In other embodiment, require under the low slightly situation as picture quality, T1 also can get greater than 2 less than 3 value; Described T2 is an integer, and span is [70,80].
Claims (3)
1, a kind of method for estimating based on time-space domain mixing information, described method is the piece that earlier current encoded frame is divided into non-overlapping copies, fixed size, then at each current block that will be encoded, in a region of search of reference frame, search out piece with current block matching error minimum, it is best matching blocks, at last according to the best matching blocks loca, it is the position of optimal match point, the relative displacement of output best matching blocks and current block, it is motion vector, it is characterized in that described search procedure is achieved in that
Step 1, with the left piece C1 in the adjacent block of current block C, go up piece C2, upper left C3 and upper right C4, and corresponding blocks E and the adjacent right piece E1, down piece E2, lower-left piece E3 and the bottom right piece E4 totally 9 piece pieces of forming 3*3s of current block in reference frame, be the time-space domain block of information of current block, wherein piece C1, piece C2, piece C3 and piece C4 represent the spatial information (si) of current block, and piece E, piece E1, piece E2, piece E3 and piece E4 represent the time-domain information of current block; Calculate the minimum average B configuration absolute difference MAD value of current block C and corresponding blocks E;
Step 2, setting threshold T1, the size of MAD value and T1 relatively when MAD＜T1, judges that current block is to be similar to static block, execution in step 3, otherwise be moving mass, execution in step 4;
Step 3, employing diamond search (ds) carry out the optimum Match point search, output optimal match point and coordinate thereof;
Step 4, in 9 pieces of time-space domain block of information, find out the best matching blocks of current block C, the motion vector mv=(m of current block and best matching blocks
x, m
y); The mould value of calculating kinematical vector mv | mv|;
Step 5, setting threshold T2, relatively | the size of mv| and T2, when | mv|＜T2, the judgement current block is the microinching piece, execution in step 6; Otherwise be the rapid movement piece, execution in step 7;
Step 6, adopt the approximate octagon search method of tri-layer to carry out the optimum Match point search, and judge, when optimal match point is in the approximate octagon of tri-layer first ring from inside to outside or on the ring, export optimal match point and coordinate thereof according to Search Results; Otherwise, when optimal match point is on approximate octagonal second ring of tri-layer, adopt the melee search method to search for, output optimal match point and coordinate thereof; Otherwise, when optimal match point is on approximate octagonal the 3rd ring of tri-layer, adopt the melee search method to search for again, output optimal match point and coordinate thereof;
Step 7, earlier adopt asymmetric cross search method to carry out the optimum Match point search, then with resulting optimal match point as starting point, adopt the approximate octagon search method search of four levels; And judge according to Search Results, when optimal match point is in the approximate octagon of four levels first ring from inside to outside or on the ring, output optimal match point and coordinate thereof; Otherwise, when optimal match point is on approximate octagonal second ring of four levels, adopt the melee search method to search for, output optimal match point and coordinate thereof; Otherwise, carry out the search of hexagon search method earlier, and then adopt the melee search method to search for, output optimal match point and coordinate thereof.
2, the method for estimating based on time-space domain mixing information as claimed in claim 1 is characterized in that, the span of described T1 is [2,3].
3, the method for estimating based on time-space domain mixing information as claimed in claim 1 is characterized in that, described T2 is an integer, and span is [70,80].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2009100458485A CN101494757A (en) | 2009-01-23 | 2009-01-23 | Motion estimating method based on time-space domain mixing information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2009100458485A CN101494757A (en) | 2009-01-23 | 2009-01-23 | Motion estimating method based on time-space domain mixing information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101494757A true CN101494757A (en) | 2009-07-29 |
Family
ID=40925121
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2009100458485A Pending CN101494757A (en) | 2009-01-23 | 2009-01-23 | Motion estimating method based on time-space domain mixing information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101494757A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101815218A (en) * | 2010-04-02 | 2010-08-25 | 北京工业大学 | Method for coding quick movement estimation video based on macro block characteristics |
| CN102025989A (en) * | 2010-11-05 | 2011-04-20 | 北京理工大学 | Quick searching method for digital video image movement estimation |
| CN106156747A (en) * | 2016-07-21 | 2016-11-23 | 四川师范大学 | The method of the monitor video extracting semantic objects of Behavior-based control feature |
| CN106340031A (en) * | 2015-07-07 | 2017-01-18 | 香港生产力促进局 | Method and device for detecting moving object |
| CN109660800A (en) * | 2017-10-12 | 2019-04-19 | 北京金山云网络技术有限公司 | Method for estimating, device, electronic equipment and computer readable storage medium |
| CN109831672A (en) * | 2019-01-09 | 2019-05-31 | 北京大学 | A kind of method for estimating of Space-time Pulse array |
| WO2019169997A1 (en) * | 2018-03-07 | 2019-09-12 | 腾讯科技（深圳）有限公司 | Motion estimation method and device for video, terminal and storage medium |
-
2009
- 2009-01-23 CN CNA2009100458485A patent/CN101494757A/en active Pending
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101815218A (en) * | 2010-04-02 | 2010-08-25 | 北京工业大学 | Method for coding quick movement estimation video based on macro block characteristics |
| CN102025989A (en) * | 2010-11-05 | 2011-04-20 | 北京理工大学 | Quick searching method for digital video image movement estimation |
| CN106340031A (en) * | 2015-07-07 | 2017-01-18 | 香港生产力促进局 | Method and device for detecting moving object |
| CN106156747A (en) * | 2016-07-21 | 2016-11-23 | 四川师范大学 | The method of the monitor video extracting semantic objects of Behavior-based control feature |
| CN106156747B (en) * | 2016-07-21 | 2019-06-28 | 四川师范大学 | The method of the monitor video extracting semantic objects of Behavior-based control feature |
| CN109660800A (en) * | 2017-10-12 | 2019-04-19 | 北京金山云网络技术有限公司 | Method for estimating, device, electronic equipment and computer readable storage medium |
| WO2019169997A1 (en) * | 2018-03-07 | 2019-09-12 | 腾讯科技（深圳）有限公司 | Motion estimation method and device for video, terminal and storage medium |
| JP2021513771A (en) * | 2018-03-07 | 2021-05-27 | テンセント・テクノロジー・（シェンジェン）・カンパニー・リミテッド | Video motion estimation methods, devices, terminals, and storage media |
| JP7024099B2 (en) | 2018-03-07 | 2022-02-22 | テンセント・テクノロジー・（シェンジェン）・カンパニー・リミテッド | Video motion estimation methods, devices, terminals, and storage media |
| US11375227B2 (en) | 2018-03-07 | 2022-06-28 | Tencent Technology (Shenzhen) Company Limited | Video motion estimation method and apparatus, and storage medium |
| CN109831672A (en) * | 2019-01-09 | 2019-05-31 | 北京大学 | A kind of method for estimating of Space-time Pulse array |
| CN109831672B (en) * | 2019-01-09 | 2020-10-27 | 北京大学 | Motion estimation method of space-time pulse array, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101494757A (en) | Motion estimating method based on time-space domain mixing information | |
| CN102984521B (en) | High-efficiency video coding inter-frame mode judging method based on temporal relativity | |
| CN103260031B (en) | Method and apparatus for encoding/decoding to motion vector | |
| CN102648631B (en) | For the method and apparatus of coding/decoding high-definition picture | |
| CN106604035B (en) | A method of the estimation for Video coding and compression | |
| CN110087087A (en) | VVC interframe encode unit prediction mode shifts to an earlier date decision and block divides and shifts to an earlier date terminating method | |
| CN106454378B (en) | Converting video coding method and system in a kind of frame per second based on amoeboid movement model | |
| CN108781284A (en) | The method and device of coding and decoding video with affine motion compensation | |
| CN106797477A (en) | Derived for frame in BC and inter prediction unified AMVP and merging candidate list | |
| CN104883565A (en) | Decision-making method and device for intra-frame prediction mode of high efficiency video coding | |
| CN102811346B (en) | coding mode selection method and system | |
| CN105791826A (en) | Data mining-based HEVC inter-frame fast mode selection method | |
| JP2006304102A (en) | Image coding unit and image coding method | |
| CN103188496A (en) | Fast motion estimation video encoding method based on motion vector distribution forecast | |
| CN107318016A (en) | A kind of HEVC inter-frame forecast mode method for rapidly judging based on zero piece of distribution | |
| CN102752588A (en) | Video encoding and decoding method using space zoom prediction | |
| CN102291581A (en) | Realizing method of self-adaptive motion estimation supporting frame field | |
| CN102647598A (en) | H.264 inter-frame mode optimization method based on maximin MV (Music Video) difference value | |
| CN101304529A (en) | Method and device for selecting macro block pattern | |
| CN101895761B (en) | Quick intraframe prediction algorithm | |
| CN101883275B (en) | Video coding method | |
| CN106878754B (en) | A kind of 3D video depth image method for choosing frame inner forecast mode | |
| CN1322758C (en) | Fast motion assessment method based on object texture | |
| CN103384332A (en) | Improved algorithm for HFPS algorithm of AVS video coding | |
| CN104469380B (en) | Video image forecasting search method based on H.264/AVC standard |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090729 |