CN101924933A - Method for tracing interested area in video frame sequence - Google Patents
Method for tracing interested area in video frame sequence Download PDFInfo
- Publication number
- CN101924933A CN101924933A CN2009102039433A CN200910203943A CN101924933A CN 101924933 A CN101924933 A CN 101924933A CN 2009102039433 A CN2009102039433 A CN 2009102039433A CN 200910203943 A CN200910203943 A CN 200910203943A CN 101924933 A CN101924933 A CN 101924933A
- Authority
- CN
- China
- Prior art keywords
- frame
- search
- area
- interest
- subsequent frame
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a method for tracing an interested area in a video frame sequence, comprising the following steps of: positioning an interested area (52) in a first frame (50) of the video frame sequence by executing a search; and positioning a corresponding interested area in a subsequent frame (60) by executing a search. The method is characterized in that the search in the subsequent frame (60) starts in the subsequent frame (60) and the first frame (50) to recognize related areas of the interested area (52).
Description
Technical field
The present invention relates to a kind of method of following the tracks of area-of-interest in the sequence of frames of video.
Background technology
In field of video processing, there are many application need processing video frames or sequence of frames of video to determine or estimate the character of this frame or frame sequence.Such as, frame of video can be categorized as interleaved or line by line scan on the basis of this processing operation.In many such operations, be unnecessary to the processing of entire frame, because just can obtain result's (for example interlaced/progressive classification) by the part of only handling this frame that comprises relevant information.This is called area-of-interest (ROI) handles, and is used to reduce to the handled frame acquisition required processing time of result.Area-of-interest may be relevant with the content of frame of video, perhaps may be relevant with the pseudo-shadow (artefact) in the frame, and for example motion between the field.
Typically, be regarded as meeting certain criterion by the block of pixels of processed frame successively until one or more described and discern area-of-interest,, described part that comprises area-of-interest or area-of-interest of this criterion indication.For instance, described can be the square that comprises 4096 pixels (i.e. 64 pixels * 64 pixels).The spiral search method is generally used for discerning area-of-interest, and initial whereby the processing is positioned at frame center or towards the piece of frame center, then is other pieces in the spiral that increases, as shown in Figure 1.
If follow the tracks of area-of-interest in the subsequent frame of sequence of frames of video, the spiral search method is often used in locating area-of-interest in each subsequent frame so, and this can cause the unnecessary long process time of cost when the identification area-of-interest.
Summary of the invention
According to a first aspect of the invention, a kind of method of following the tracks of area-of-interest in the sequence of frames of video is provided, this method comprises that carrying out search locatees the area-of-interest in first frame of this sequence and carry out corresponding area-of-interest in the subsequent frame of this sequence of search location, it is characterized in that search in this subsequent frame start from this subsequent frame with first frame in identify the relevant district (area), district of area-of-interest by the search of described first frame.
The search of described subsequent frame can start from and the corresponding position, position of identifying area-of-interest by the search of described first frame.
When area-of-interest was identified, the search of described first frame can finish.
The search of described first frame and the search of described subsequent frame can comprise spiral search.
The search of described subsequent frame can start from the corresponding direction of the direction of search with described first frame when the search of described first frame finishes.
Preferably with described first frame and subsequent frame in each block of pixels carry out the search of described first frame and subsequent frame relevantly.
According to a second aspect of the invention, provide a kind of computer program that is used to carry out the method for first aspect.
Description of drawings
With reference now to accompanying drawing,, embodiments of the invention are only described as an example, wherein:
Fig. 1 is the schematic diagram of known spiral search method;
Fig. 2 illustrates the schematic diagram that is used at the spiral search of first frame of video location of video sequence area-of-interest;
Fig. 3 illustrates to be used for using the schematic diagram of method of the present invention at the spiral search of the area-of-interest of subsequent frame network for location 2; And
Fig. 4 illustrates to be used for using the schematic diagram of the optional embodiment of the inventive method at the spiral search of the area-of-interest of subsequent frame network for location 2.
Embodiment
At first with reference to figure 1, show the schematic diagram of known spiral search method, it can be used to discern area-of-interest, for example motion between the field in the frame of video.In Fig. 1, frame of video is shown as 10, and can think and be made of piece 12, and each piece comprises some pixels.In known spiral search method because area-of-interest more likely is positioned at the core rather than the deep part of frame of video 10, search generally start from the frame of video center or towards the piece 14 at (towards) frame of video center.Like this, use suitable algorithm that center pixel block 14 is handled to determine whether it comprises area-of-interest.If in central block, do not find area-of-interest, then handle neighborhood pixels piece 16 to determine whether it comprises area-of-interest.In example shown in Figure 1, piece 16 is positioned at the right of central block 14, still it should be understood that and can select any contiguous block to handle.
If do not find area-of-interest in piece 16, search changes direction and handles other piece 18, and it is positioned at the top of piece 16 in this example.If do not find area-of-interest in block of pixels 18, search moves to contiguous block 20, determines by processing block 20 whether it contains area-of-interest.Piece in the frame of video 10 is processed successively, and wherein this searching method is identified until area-of-interest by these pieces along outward extending spiral path, can stop at this point search.Alternatively, for more area-of-interest in the identification video frame 10, the position that before search restarts, can write down area-of-interest.
If search institute along the edge of spiral path arrival frame of video 10, as in piece 22,24,26,28 take place like that, search will restart at the piece that the next one is not searched for, it is the same to just look like that this spiral path is not interrupted, as the dotted line among Fig. 1 40,42, shown in 44.Like this, processed when piece 2, search moves to piece 30, and processing block 36 is followed in processing block 24 backs.
This spiral search method is to handle single frame of video to discern the effective ways of one or more area-of-interests.But, must the processing video frames sequence time, in order to discern the area-of-interest in first frame and in one or more subsequent frames, to follow the tracks of this area-of-interest, begin to carry out spiral search from the central block of this subsequent frame or each subsequent frame and discern area-of-interest and can spend unnecessary long-time.
Appear in the sequence of frames of video in most cases at area-of-interest, the area-of-interest between each frame spatially is correlated with, and that is to say that the area-of-interest in the subsequent frame is arranged in this frame and the same or analogous district of the first frame area-of-interest.It is very big that successive frame in the sequence of frames of video generally can difference, and the displacement of area-of-interest between the successive frame of sequence is general also very little.The present invention utilizes these characteristics to reduce time that area-of-interest spent in the subsequent frame of identification video sequence, thereby has reduced the power consumption of the equipment of handling the required processing resource of this sequence and this method of operation.
In the method for the invention, the search of subsequent frame start from this subsequent frame zone with previous frame in identify position in the relevant district, the district of area-of-interest, as following referring to figs. 2 and 3 described.
In Fig. 2, the spiral search of carrying out on first frame of video 50 in sequence has identified the area-of-interest 52 in the piece 54 of frame of video 50.The position of writing down this area-of-interest 52 is to be used for determining beginning the position of the search of this sequence subsequent frame.
Fig. 3 shows such subsequent frame 60, and wherein area-of-interest 52 has moved to piece 62.The spiral search of this subsequent frame 60 starts from a little 64, and it is corresponding to the position of area-of-interest 52 in first frame 50.
Because the spiral search path the during area-of-interest 52 in identifying first frame 50 along upward to, so the spiral search of subsequent frame 60 is preferably along upward to beginning, because this possible moving direction that is area-of-interest 52.If the spiral search path the during area-of-interest 52 in identifying first frame 50 is along different directions, the spiral search of subsequent frame 60 preferably begins along this different direction.Identification area-of-interest 52 in piece 62, and its position is recorded, the position that is used for determining beginning the search of this sequence subsequent frame.
Alternatively, the spiral search of subsequent frame 60 needn't along the spiral search of first frame 50 when identifying area-of-interest 52 along direction begin.Such as in order to simplify the execution of this method, the spiral search of subsequent frame 60 can always begin along same direction, as the direction that makes progress.
Fig. 4 shows the optional embodiment of this method, wherein the spiral search of subsequent frame 60 is not to begin corresponding to the point that identifies the accurate position of area-of-interest 52 in first frame 50 from subsequent frame 60, but identifies relevant district, the district of area-of-interest 52 and begin from subsequent frame 60 with in first frame 50.In this example, select on the spiral search path with first frame 50 that identifies area-of-interest 52 the corresponding point in a little backward the position point 66 of the search of subsequent frame 60 to start with.Choice point 66 reason of the point of the search of subsequent frame 60 to start with is, this has reduced the search of passing through first frame 50 that appears in first frame 50 and has missed the pseudo-shadow " misleading " that the is identified as area-of-interest possibility to the search of second frame 60, because the search of subsequent frame must be along longer spiral path among the embodiment than Fig. 3, handle more piece in the subsequent frame 60, thereby improve before the pseudo-shadow of mistake identification in identification first frame 50, correctly discern the possibility of area-of-interest.Certainly, improving the correct cost of discerning the possibility of area-of-interest in the subsequent frame 60 is the increase a little in processing time.But, and to carry out from the central block of subsequent frame 60 begins the required processing time of the new spiral search of subsequent frame 60 is compared, this processing time has still shortened.
It will be appreciated that the point 66 of the point that the spiral search that is selected as subsequent frame 60 begins, can be to identify any point of the subsequent frame 60 in the relevant district, the district of area-of-interest 52 in subsequent frame 60 with in first frame 50.For instance, except the position that the spiral search of first frame 50 begins, point 66 can be corresponding to any position on the spiral search path of identifying area-of-interest in first frame 50.
In arbitrary embodiment of this method, if be higher than the position that the search of first frame 50 begins by searching for the position of area-of-interest 52 in first frame that first frame 50 identifies, if promptly the area-of-interest 52 in first frame 50 is positioned on the central block 14 of first frame 50, area-of-interest 52 might be along upward to moving in subsequent frame 60 so, therefore the search to subsequent frame 60 can begin from the point that makes progress a little than the point corresponding to the position of area-of-interest 52 first frame 50, thereby further reduce in the subsequent frame 60 the piece number that must handle, therefore reduced the required time of area-of-interest in the identification subsequent frame 60 52.Similarly, if the position of area-of-interest 52 in first frame of identifying by the search to first frame 50 is lower than the position that the search of first frame 50 begins, if promptly the area-of-interest 52 in first frame 50 is positioned under the central block 14 of first frame 50, area-of-interest 52 might move along downward direction in subsequent frame so, therefore can be from beginning to the search of subsequent frame 60 than the downward a little point of point corresponding to the position of area-of-interest 52 first frame 50, thus the piece number that must handle in the subsequent frame 60 further reduced.
Be used for handling the algorithm of first frame 50 and subsequent frame 60 pieces or the type that method depends on the area-of-interest that will discern, this will be apparent to those skilled in the art.For example, if motion between identification can be correlated with between the front court of frame and back court so.
Although described method of the present invention, it will be appreciated that also and can use method of the present invention to reduce identification and to follow the tracks of the required processing time of area-of-interest in the sequence of frames of video in conjunction with other searching methods about spiral search.
Claims (7)
1. method of following the tracks of area-of-interest in the sequence of frames of video, this method comprises that carrying out search searches for area-of-interest corresponding in the subsequent frame of locating described sequence with area-of-interest in first frame of locating described sequence and execution, it is characterized in that, the search in the described subsequent frame start from the described subsequent frame with described first frame in identify the relevant district, district of area-of-interest by the search of described first frame.
2. method according to claim 1, the search of described subsequent frame starts from identifying corresponding to the search by described first frame position of the position of area-of-interest.
3. method according to claim 1 and 2, wherein when identifying area-of-interest, the search of described first frame finishes.
4. according to any described method in the above claim, the search of wherein said first frame and the search of described subsequent frame comprise spiral search.
5. according to any described method in the above claim, the direction of the direction of search of described first frame began when finished corresponding to the search when described first frame on the search of wherein said subsequent frame edge.
6. according to any described method in the above claim, wherein with first frame and subsequent frame in each block of pixels carry out the search of described first frame and subsequent frame relatively.
7. one kind is used for the computer program that enforcement of rights requires any described method of 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009102039433A CN101924933A (en) | 2009-04-10 | 2009-04-10 | Method for tracing interested area in video frame sequence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2009102039433A CN101924933A (en) | 2009-04-10 | 2009-04-10 | Method for tracing interested area in video frame sequence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101924933A true CN101924933A (en) | 2010-12-22 |
Family
ID=43339533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009102039433A Pending CN101924933A (en) | 2009-04-10 | 2009-04-10 | Method for tracing interested area in video frame sequence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101924933A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103974142A (en) * | 2013-01-31 | 2014-08-06 | 深圳市快播科技有限公司 | Video playing method and system |
| CN109740590A (en) * | 2017-11-06 | 2019-05-10 | 凝眸智能科技集团公司 | The accurate extracting method of ROI and system based on target following auxiliary |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1097632A (en) * | 1996-09-20 | 1998-04-14 | Fujitsu General Ltd | Method for tracking movement of object |
| JP2000348180A (en) * | 1999-06-01 | 2000-12-15 | Fujitsu Ltd | Method and device for tracing moving area and recording medium |
| US20060104517A1 (en) * | 2004-11-17 | 2006-05-18 | Byoung-Chul Ko | Template-based face detection method |
| CN1809839A (en) * | 2003-06-23 | 2006-07-26 | 维切普公司 | Method and apparatus for adaptive multiple-dimentional signal sequences encoding/decoding |
| JP2006228061A (en) * | 2005-02-18 | 2006-08-31 | Fujitsu Ltd | Face tracing program and face tracing method |
-
2009
- 2009-04-10 CN CN2009102039433A patent/CN101924933A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH1097632A (en) * | 1996-09-20 | 1998-04-14 | Fujitsu General Ltd | Method for tracking movement of object |
| JP2000348180A (en) * | 1999-06-01 | 2000-12-15 | Fujitsu Ltd | Method and device for tracing moving area and recording medium |
| JP3934279B2 (en) * | 1999-06-01 | 2007-06-20 | 富士通株式会社 | Moving area tracking method and apparatus, and recording medium |
| CN1809839A (en) * | 2003-06-23 | 2006-07-26 | 维切普公司 | Method and apparatus for adaptive multiple-dimentional signal sequences encoding/decoding |
| US20060104517A1 (en) * | 2004-11-17 | 2006-05-18 | Byoung-Chul Ko | Template-based face detection method |
| JP2006146922A (en) * | 2004-11-17 | 2006-06-08 | Samsung Electronics Co Ltd | Template-based face detection method |
| JP2006228061A (en) * | 2005-02-18 | 2006-08-31 | Fujitsu Ltd | Face tracing program and face tracing method |
Non-Patent Citations (1)
| Title |
|---|
| TATIANA V EVREINOVA ET AL: "Video as Input: Spiral Search with the Sparse Angular Sampling", <<COMPUTER AND INFORMATION SCIENCES-ISCIS 2006 LECTURE NOTES IN COMPUTER SCIENCE;LNCS,SPRINGER,BERLIN,DE>> * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103974142A (en) * | 2013-01-31 | 2014-08-06 | 深圳市快播科技有限公司 | Video playing method and system |
| CN103974142B (en) * | 2013-01-31 | 2017-08-15 | 深圳市快播科技有限公司 | A kind of video broadcasting method and system |
| CN109740590A (en) * | 2017-11-06 | 2019-05-10 | 凝眸智能科技集团公司 | The accurate extracting method of ROI and system based on target following auxiliary |
| CN109740590B (en) * | 2017-11-06 | 2023-04-07 | 凝眸智能科技集团公司 | ROI accurate extraction method and system based on target tracking assistance |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8379987B2 (en) | Method, apparatus and computer program product for providing hand segmentation for gesture analysis | |
| US8917902B2 (en) | Image overlaying and comparison for inventory display auditing | |
| US9424668B1 (en) | Session-based character recognition for document reconstruction | |
| CN106843737B (en) | Text input method and device and terminal equipment | |
| CN105159913B (en) | Method and device for determining file to be cleaned | |
| JP5445467B2 (en) | Credit information section detection method, credit information section detection device, and credit information section detection program | |
| JP4364266B2 (en) | Image processing apparatus and program | |
| WO2017037493A1 (en) | Product auditing in point-of-sale images | |
| US8660302B2 (en) | Apparatus and method for tracking target | |
| CN103106388B (en) | Method and system of image recognition | |
| CN105183827A (en) | Search method and apparatus based on handwriting input | |
| CN105022585A (en) | Message display method and terminal | |
| US20080155465A1 (en) | Identifying items that have experienced recent interest bursts | |
| CN104822009A (en) | Method for identifying video scene transformation and device thereof | |
| CN109086725A (en) | Hand tracking and machine readable storage medium | |
| CN103152633A (en) | Method and device for identifying key word | |
| CN101924933A (en) | Method for tracing interested area in video frame sequence | |
| CN107076980A (en) | System and method for embedded images in the micro- scanning in big visual field | |
| CN102737383A (en) | Camera movement analyzing method and device in video | |
| CN104268504B (en) | Image identification method and device | |
| CN107484013B (en) | A method of television program interaction is carried out using mobile device | |
| CN111079527B (en) | Shot boundary detection method based on 3D residual error network | |
| CN103984593A (en) | Method and device for processing interrupt during game by mobile terminal | |
| CN103309993B (en) | The extracting method of a kind of key word and device | |
| JP4997179B2 (en) | Image processing apparatus, method, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20101222 |