CN103002196A - Method for estimating prediction motion vector - Google Patents
Method for estimating prediction motion vector Download PDFInfo
- Publication number
- CN103002196A CN103002196A CN2011102681827A CN201110268182A CN103002196A CN 103002196 A CN103002196 A CN 103002196A CN 2011102681827 A CN2011102681827 A CN 2011102681827A CN 201110268182 A CN201110268182 A CN 201110268182A CN 103002196 A CN103002196 A CN 103002196A
- Authority
- CN
- China
- Prior art keywords
- pixel
- value
- pixel difference
- vector
- estimating
- 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
- Image Analysis (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention relates to a method for estimating a prediction motion vector. The method for estimating prediction motion vector is used for an image block with a plurality of pixels and includes calculating a pixel difference value between a current image and a reference image for each pixel in the image block; determining a pixel difference area according to the pixel difference value; and determining the prediction motion vector according to the pixel difference area.
Description
Technical field
The present invention relates to a kind of estimating and measuring method of predicting motion-vector, relate in particular to a kind of method that the prediction motion-vector of estimating by the different information between two pictures.
Background technology
Mobile estimating (Motion Estimation) is important technology during the video image compression is processed, the association of picture before and after its purpose is to judge, to determine a motion-vector (Motion Vector), and obtain according to this follow-up image, and can reduce the redundant information (Redundancy Information) between each image frame in the different time.Because video image is the result that a succession of image frame is play continuously, and its cardinal principle is the slight change by adjacent image frame, makes people that the reaction of the persistence of vision be arranged and produce the effect of animation.In general, adjacent picture spatially with on the time has stronger association usually, should all have to each other a part and not change fully.Therefore, when storing or during the transmission image frame, not having vicissitudinous part just not need to store, only need the information of the previous picture of record and follow-up picture can be rebuild by the information of recording in the moving process.In other words, in video image coding and decoding process, just need not process the information of all pictures, and can reach the amount of information that effectively reduces transmission, and meet the effect of image compression.
At present block comparison (Block Matching) method is to calculate one of method that motion-vector often is employed, the block comparison method sees through image frame is cut into a plurality of nonoverlapping blocks, and the most similar part between each block in the searching different time, obtain the motion-vector information of each block.In background technology, existing many search algorithms are suggested the relevant information of seeking motion-vector, for instance, utilize universe search method (Full Search), three step search methods (Three Step Search), four step search methods (Four Step Search), diamond search method (Diamond Search), 3D recurrence search method (Three Dimensional Recursive Search) or cross search method (Cross Search) etc. various search pattern obtains the relevant information of motion-vector, then, see through again comparison calculation (match estimation), for example absolute error and (sum of absolute difference, SAD) computing, obtain the prediction motion-vector, to realize optimized block comparison, yet, though the block comparison method can obtain more accurately motion-vector information, search time and the larger computational complexity long because of tool are not suitable for real-time application.
In the case, background technology proposes a kind of phase plane correlation (phase plane correlation) mode in addition, realize the prediction of fast moving vector, it mainly is that video image is changed to frequency domain by spatial domain, and directly carries out the prediction of motion-vector in the phase difference of two image frames of frequency domain comparison.Yet the program that the mode of employing phase plane correlation must be carried out Fast Fourier Transform conversion (FFT) begins to finish conversion, thus, also can expend too much system resource and translation operation time, therefore, still is unsuitable in the real-time application.
Summary of the invention
Therefore, the object of the present invention is to provide a kind of can be simply by the different information between two pictures and promptly estimate the method that the prediction motion-vector, to solve the above problems.
According to embodiments of the invention, it is to disclose a kind of estimating and measuring method of predicting motion-vector, for the image block with a plurality of pixels, this estimating and measuring method includes each pixel in this image block, calculates the pixel difference value corresponding to a present picture and a reference picture; According to this pixel difference value, determine pixel difference zone; And according to this pixel difference zone, determine a prediction motion-vector.
Description of drawings
Fig. 1 is the schematic diagram of the embodiment of the invention one flow process.
Fig. 2 A and 2B are respectively the schematic diagram in movement of objects when running in the video image.
Fig. 3 is the schematic diagram that the pixel difference value after the movement of objects in the video image distributes.
Fig. 4 is the schematic diagram of the image block among Fig. 3.
Fig. 5 is the schematic diagram that the pixel difference value after another movement of objects in the video image distributes.
Fig. 6 is the schematic diagram of the image block among Fig. 5.
Wherein, description of reference numerals is as follows:
10 flow processs
100,102,104,106,108 steps
A, A_SUB pixel difference zone
B, B_SUB white space
F (n-1), F (n) image frame
MB_O image block
Obj_A, Obj_B object
ED, ED ', P, P ', P1, P1 ' position
Embodiment
Please refer to Fig. 1, Fig. 1 is the schematic diagram of the embodiment of the invention one flow process 10.For convenience of description, suppose that a video image can be cut into a plurality of image blocks, each image block includes a plurality of pixels.Flow process 10 can be used to estimate a prediction motion-vector of each image block, and flow process 10 may further comprise the steps:
Step 100: beginning.
Step 102: to each pixel in the image block, calculate the pixel difference value corresponding to a present picture and a reference picture.
Step 104: according to the pixel difference value, determine pixel difference zone.
Step 106: according to pixel difference zone, determine a prediction motion-vector.
Step 108: finish.
According to flow process 10, the present invention can according to the information of reference picture, estimate the prediction motion-vector that corresponding to the image block of present picture.At first, in step 102, for each pixel in the image block, calculate the pixel difference value corresponding to present picture and reference picture.Specifically, for each pixel, can detect first it corresponding to current pixel value of present picture and corresponding to a reference pixel value of reference picture.Then, current pixel value and the reference pixel value that detects carried out additive operation, to calculate the difference of current pixel value and reference pixel value, in the case, the result who calculates is the pixel difference value of this corresponding pixel.Described current pixel value and reference pixel value can be respectively a brightness value or a chroma value, and relatively, the pixel difference value then is a luminance difference value or a chroma difference value.Thus, the pixel difference value of each pixel can represent the intensity of variation between its present picture and reference picture.
Preferably, described reference picture is to be a previous picture, that is to say, this reference picture can be video image in the at present front single image frame of picture or the set of a plurality of image frames, for example, the image frame when if picture is time T (n) at present, reference picture is last picture (image frame when being time T (n-1)), thus, in step 102, namely calculate each pixel corresponding to the pixel difference value of present picture and last picture, certainly, reference picture is not limited only to last picture, and for example reference picture also can be the image frame before several display cycles of present picture, the image frame when for example reference picture can be time T (n-3), those skilled in the art should understand, and do not repeat them here.In addition, reference picture also can be the set of front several pictures, the for example set of first three picture (time T (n-1) to time T (n-3) time image frame), then in step 102, for each pixel, can calculate it corresponding to the current pixel value of present picture and difference corresponding to the reference pixel mean value of front several pictures (for example first three picture), that is to say, when the current pixel value that detects respectively present picture and after corresponding to the corresponding reference pixel value of front several pictures, calculate again the mean value corresponding to the corresponding reference pixel value of front several pictures.Then, the result who the mean value of current pixel value and reference pixel is carried out additive operation is the pixel difference value of present picture and front several pictures.
Further, video image is comprised of a series of image frame, for continuous image frame, because time point approaches, so stronger spatial coherence and temporal correlation are arranged to each other.In the case, if the object in the video image presents when static, adjacent picture difference is little usually, that is to say, the current pixel value of picture can be identical or very approaching with reference pixel value at present.In like manner, when if the object in the video image has mobile situation to produce, for the edge part of object for the pixel of process, it will have obvious difference condition corresponding to the current pixel value of present picture and reference pixel value corresponding to reference picture and occur, that is during movement of objects, its edge part the corresponding pixel of process, will have larger pixel difference value.In the case, after the pixel difference value of all pixels in the image block is calculated, via execution in step 104, according to the pixel difference value that calculates, decide pixel difference zone.Specifically, in step 104, can be relatively change threshold value corresponding to the pixel difference value and of each pixel, to produce a comparative result, and according to this comparative result, choose corresponding this pixel difference value greater than the pixel of this variation threshold value, to form this pixel difference zone, thus, this pixel difference zone namely represents the zone that move through of edge part between present picture and reference picture of object.
For instance, please refer to Fig. 2 A and 2B, Fig. 2 A and 2B are respectively the schematic diagram in movement of objects when running in the video image.Suppose that it is image frame for forward and backward two in the video image that image frame F (n-1) and F (n) are respectively, that is the image frame of image frame F (n-1) when being time T (n-1), and the image frame of image frame F (n) when being time T (n).For convenience of description, only there are in the video image single mobile object (being object Obj_A) and image frame F (n-1) to illustrate as present picture as example as reference picture and image frame F (n), shown in Fig. 2 A and 2B, in image frame F (n-1), object Obj_A is positioned at position P; In image frame F (n), object Obj_A is positioned at position P '.That is to say, object Obj_A moves to position P ' from position P.In general; the edge of object Obj_A the pixel of process usually can have larger pixel difference value; and in image frame F (n-1), F (n), all be used to show the pixel of object Obj_A or all non-pixel (being background pixel) that is used to show object Obj_A in image frame F (n-1), F (n); owing to do not show to change, so its corresponding pixel difference value being generally 0 or very little.Please refer to Fig. 3 and Fig. 4, Fig. 3 is the schematic diagram that the pixel difference value after the object Obj_A in the video image moves distributes.Fig. 4 is the schematic diagram of the image block MB_O among Fig. 3.Suppose object Obj_A the edge the pixel of process have the pixel difference value that changes threshold value greater than, and about the pixel that all is used to show the pixel of object Obj_A among image frame F (n-1), the F (n) or treats as background, the pixel difference value that it has is all 0.In other words, for each pixel of video image, after the brightness value (or chroma value) during with image frame F (n) carries out additive operation to the brightness value (or chroma value) of image frame F (n-1), the distribution of the pixel difference value of each pixel will be as shown in Figure 3, wherein, greater than the formed zone of pixel that changes threshold value, white space B represents that the pixel difference value is 0 zone by the pixel difference value in pixel difference zone A (net region) expression.Therefore, as shown in Figure 4, when wish is estimated a prediction motion-vector of single image block (for example image block MB_O), can see through execution in step 102 obtain each pixel among the image block MB_O the pixel difference value after, execution in step 104 has decided pixel difference zone A_SUB and white space B_SUB again, to be provided as the foundation of follow-up estimation program.
Then, in step 106, according to the information in the selected pixel difference zone of above-mentioned steps, decide the prediction motion-vector of corresponding image block.For example, can predict according to the length of the horizontal direction in pixel difference zone or vertical direction the motion-vector of this image block.That is to say, decide the horizontal component of prediction motion-vector according to the length of the horizontal direction in pixel difference zone, or decide the vertical component of prediction motion-vector according to the length of the vertical direction in pixel difference zone.For example, please continue with reference to figure 4, if the height of pixel difference zone A_SUB is Δ X, then the horizontal component of the prediction motion-vector V of image block MB_O can be set as Δ X; If the height of pixel difference zone A_SUB is Δ Y, then the vertical component of the prediction motion-vector V of image block MB_O is set as Δ Y, or the prediction motion-vector V of image block MB_O can be set as V=(Δ X, Δ Y) or V=is (Δ X, 0) or V=(0, Δ Y).Though it should be noted that at this highly to explain, so in fact pixel difference zone A_SUB can be divided into positive region and negative region, so vertical component or horizontal component can have positive negativity.In addition, in the situation that picture complicated can't distinguishing too also may be that positive sign and negative sign all use.In addition, when determining pixel difference zone A_SUB, the also less or mixed and disorderly zone of filtering area more.
In addition; since pixel difference zone be object in the video image when mobile the pixel of process form; therefore; the shape in pixel difference zone can spatially have very large association with object usually; in the case, the motion-vector that in step 106, also can predict the image block according to shape and the correlation between object in pixel difference zone.For instance, when if the mobile range of object is little, the shape in pixel difference zone has similar place to the shape of the edge part of object, therefore, can according to the association of the edge part of pixel difference region shape and object, decide to predict motion-vector.Please refer to Fig. 5 and Fig. 6, Fig. 5 is the schematic diagram that the pixel difference value after the object Obj_B in the video image moves distributes.Fig. 6 is the schematic diagram of the image block MB_O among Fig. 5.Suppose that Fig. 5 is the distribution situation that the corresponding pixel value of two image frames carries out the result that produces behind the additive operation, and single mobile object (being object Obj_B) is only arranged in the video image.For convenience of description, the condition of choosing of pixel difference zone A (net region) and white space B, all the condition with shown in Figure 3 is similar or identical, therefore give unnecessary details no longer separately.Such as Fig. 5 and shown in Figure 6, after object Obj_B moves to position P1 ' by position P1, clearly can distinguish mobile trend by the shape of pixel difference zone A_SUB and the face shaping of object Obj_B, be that measurable object is to be moved towards the direction of position ED ' by position ED, therefore the direction that can estimate the prediction motion-vector is by the direction of position ED towards position ED ', or also position ED and the formed vector of position ED ' can be set as the prediction motion-vector V of image block MB_O.In other words, by the variation of mobile distribute (Transition distribution) between the detecting two articles, can estimate trend or the direction of movement of objects.
On the other hand, aforementionedly estimate the prediction motion-vector that and also can be applicable in the mobile estimating program, search reference when being used to provide as various motion-vector search method running, with the correct precision of searching of further raising, and the search time of reduction motion-vector.For instance, if the present invention estimated the prediction motion-vector that as the search reference of 3D recurrence search method, can reduce significantly the quantity of search and search rapidly more accurate motion-vector.
The above only is the preferred embodiments of the present invention, and all equalizations of doing according to claim of the present invention change and modify, and all should belong to covering scope of the present invention.
Claims (10)
1. estimating and measuring method of predicting motion-vector is used for an image block with a plurality of pixels, it is characterized in that, this estimating and measuring method includes:
To each pixel in this image block, calculate the pixel difference value corresponding to a present picture and a reference picture;
According to this pixel difference value, determine pixel difference zone; And
According to this pixel difference zone, determine a prediction motion-vector.
2. estimating and measuring method as claimed in claim 1 is characterized in that, to each pixel in this image block, the step that calculates corresponding to this pixel difference value of this present picture and this reference picture includes:
To each pixel in this image block, detect corresponding to a current pixel value of this present picture and a reference pixel value corresponding to this reference picture; And
Calculate the difference of this current pixel value and this reference pixel value, with as this pixel difference value.
3. estimating and measuring method as claimed in claim 2 is characterized in that, this current pixel value and this reference pixel value are respectively a brightness value, and this pixel difference value is to be a luminance difference value.
4. estimating and measuring method as claimed in claim 2 is characterized in that, this current pixel value and this reference pixel value are respectively a chroma value, and this pixel difference value is to be a chroma difference value.
5. estimating and measuring method as claimed in claim 1 is characterized in that, according to this pixel difference value, the step that determines this pixel difference zone includes:
Relatively this pixel difference value of each pixel and changes threshold value, to produce a comparative result; And
According to this comparative result, choose this pixel difference value greater than the pixel of this variation threshold value, to form this pixel difference zone.
6. estimating and measuring method as claimed in claim 1 is characterized in that, according to this pixel difference zone, the step that determines this prediction motion-vector is the length according to the horizontal direction in this pixel difference zone, determines the horizontal component of this prediction motion-vector.
7. estimating and measuring method as claimed in claim 1 is characterized in that, according to this pixel difference zone, the step that determines this prediction motion-vector is the length according to the vertical direction in this pixel difference zone, determines the vertical component of this prediction motion-vector.
8. estimating and measuring method as claimed in claim 1 is characterized in that, according to this pixel difference zone, the step that determines this prediction motion-vector is the correlation according to this pixel difference region shape and an object, determines the direction of this prediction motion-vector.
9. estimating and measuring method as claimed in claim 8, it is characterized in that, according to this pixel difference zone, the step that determines this prediction motion-vector is the association according to the edge part of this pixel difference region shape and this object, determines the direction of this prediction motion-vector.
10. estimating and measuring method as claimed in claim 1 is characterized in that, this reference picture is a previous picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102681827A CN103002196A (en) | 2011-09-09 | 2011-09-09 | Method for estimating prediction motion vector |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102681827A CN103002196A (en) | 2011-09-09 | 2011-09-09 | Method for estimating prediction motion vector |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103002196A true CN103002196A (en) | 2013-03-27 |
Family
ID=47930295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011102681827A Pending CN103002196A (en) | 2011-09-09 | 2011-09-09 | Method for estimating prediction motion vector |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103002196A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109144043A (en) * | 2017-06-27 | 2019-01-04 | 金宝电子工业股份有限公司 | The method for tracking object |
| CN109308710A (en) * | 2017-07-27 | 2019-02-05 | 南宁富桂精密工业有限公司 | Monitoring method, computing device and computer readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003098922A1 (en) * | 2002-05-15 | 2003-11-27 | The Board Of Governors For Higher Education, State Of Rhode Island And Providence Plantations | An imaging system and method for tracking the motion of an object |
| CN1688164A (en) * | 2005-06-09 | 2005-10-26 | 上海交通大学 | Fast motion assessment method based on object edge shape |
| CN101017573A (en) * | 2007-02-09 | 2007-08-15 | 南京大学 | Method for detecting and identifying moving target based on video monitoring |
| CN101640798A (en) * | 2008-07-30 | 2010-02-03 | 索尼株式会社 | Motion vector detection device, motion vector detection method, and program |
| US20100123792A1 (en) * | 2008-11-20 | 2010-05-20 | Takefumi Nagumo | Image processing device, image processing method and program |
-
2011
- 2011-09-09 CN CN2011102681827A patent/CN103002196A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003098922A1 (en) * | 2002-05-15 | 2003-11-27 | The Board Of Governors For Higher Education, State Of Rhode Island And Providence Plantations | An imaging system and method for tracking the motion of an object |
| CN1688164A (en) * | 2005-06-09 | 2005-10-26 | 上海交通大学 | Fast motion assessment method based on object edge shape |
| CN101017573A (en) * | 2007-02-09 | 2007-08-15 | 南京大学 | Method for detecting and identifying moving target based on video monitoring |
| CN101640798A (en) * | 2008-07-30 | 2010-02-03 | 索尼株式会社 | Motion vector detection device, motion vector detection method, and program |
| US20100123792A1 (en) * | 2008-11-20 | 2010-05-20 | Takefumi Nagumo | Image processing device, image processing method and program |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109144043A (en) * | 2017-06-27 | 2019-01-04 | 金宝电子工业股份有限公司 | The method for tracking object |
| CN109308710A (en) * | 2017-07-27 | 2019-02-05 | 南宁富桂精密工业有限公司 | Monitoring method, computing device and computer readable storage medium |
| CN109308710B (en) * | 2017-07-27 | 2021-03-05 | 南宁富桂精密工业有限公司 | Monitoring method, computing device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1364535B1 (en) | A method for compressing video information | |
| JP2023542397A (en) | Video motion estimation method, device, equipment, and computer program | |
| TWI685250B (en) | Method, device and encoder for selecting the best reference frame of HEVC encoding | |
| CN101236656A (en) | Movement target detection method based on block-dividing image | |
| TW201146011A (en) | Bi-directional, local and global motion estimation based frame rate conversion | |
| US20180260040A1 (en) | Tracker for cursor navigation | |
| US20180005039A1 (en) | Method and apparatus for generating an initial superpixel label map for an image | |
| CN107483953B (en) | Inter frame motion estimation method, apparatus and electronic equipment | |
| CN106534858B (en) | True motion estimation method and device | |
| US9661307B1 (en) | Depth map generation using motion cues for conversion of monoscopic visual content to stereoscopic 3D | |
| CN108615241B (en) | Rapid human body posture estimation method based on optical flow | |
| CN104123733B (en) | A kind of method of motion detection and reduction error rate based on Block- matching | |
| Philip et al. | A comparative study of block matching and optical flow motion estimation algorithms | |
| US9420306B2 (en) | Estimating method of predicted motion vector | |
| EP3596698B1 (en) | Motion estimation method and apparatus for plurality of frames | |
| CN103002196A (en) | Method for estimating prediction motion vector | |
| Laumer et al. | Moving object detection in the H. 264/AVC compressed domain | |
| TW201238359A (en) | Block matching method | |
| US20140126639A1 (en) | Motion Estimation Method | |
| CN101001380A (en) | Movable estimation method of vedio code | |
| KR101777776B1 (en) | Apparatus and method for sampled profile based motion estimation | |
| Vijverberg et al. | Two novel motion-based algorithms for surveillance video analysis on embedded platforms | |
| US20200068214A1 (en) | Motion estimation using pixel activity metrics | |
| Liu et al. | Target Detection and Extraction Based on Motion Attention Model | |
| KR101787004B1 (en) | Apparatus and method for sampled pixel based motion estimation |
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 |
Application publication date: 20130327 |