CN101098475A - Interactive time-space accordant video matting method in digital video processing - Google Patents
Interactive time-space accordant video matting method in digital video processing Download PDFInfo
- Publication number
- CN101098475A CN101098475A CNA2007100699707A CN200710069970A CN101098475A CN 101098475 A CN101098475 A CN 101098475A CN A2007100699707 A CNA2007100699707 A CN A2007100699707A CN 200710069970 A CN200710069970 A CN 200710069970A CN 101098475 A CN101098475 A CN 101098475A
- Authority
- CN
- China
- Prior art keywords
- point
- video
- value
- alpha
- partiald
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000002452 interceptive effect Effects 0.000 title claims abstract description 11
- 238000009792 diffusion process Methods 0.000 claims description 11
- 239000003086 colorant Substances 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 abstract description 5
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 7
- 230000002950 deficient Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 230000003993 interaction Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 241001515806 Stictis Species 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013476 bayesian approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000006748 scratching Methods 0.000 description 1
- 230000002393 scratching effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a video image pickup method with consistent time and space domains, of digit video process, which comprises that (1), interactive disperse that user interacts on a three-dimension video element, and obtains a shield via disperse calculation, and divides video element into foreground area, background area, and unknown area, (2), picking iterated video image that calculating out the alpha value of each point on the three-dimension video element as non-transparency, (3), foreground reconstruction with consistent time and space domains that according to the alpha value, calculating out the foreground color value with continuity at time and space domains. The inventive method can quickly, effectively, and high-quality pickup image of video, which can resolve the problems of prior video pickup method as discontinuous result and long process time, with high practicality.
Description
Technical field
The video that the present invention relates to interactive time-space unanimity in a kind of Digital Video Processing is scratched drawing method.
Background technology
It is a kind of digitized video treatment technology that the part of prospect in the image is come out from background separation that image is scratched figure, has a wide range of applications at aspects such as video display special efficacy making.According to process object, image is scratched figure can be divided into two classes usually, and promptly image is scratched figure and the stingy figure of video.The stingy figure problem of image can be defined as: given any pictures, obtain the contained foreground color F of every bit color value I on the image, and background color B and alpha (Alpha) are worth α.The alpha value is meant the opacity of pixel, and the relation between them is as shown in the formula I=α F+ (1-α) B.The set of the alpha value of former picture correspondence, the alpha value of promptly being had a few is called outline.On video, but this problem natural expansion is for to find the solution each frame in the video.
Existing video is scratched diagram technology roughly can be divided into four big classes, and wherein preceding two classes are stingy diagram technologies of image, can directly apply to finding the solution frame by frame of video:
It is the most widely used technology of film and tv industry up to now that early stage blue screen is scratched figure (blue screen matting).The background color that it requires input picture is fixing solid color, utilizes in the image background different with the color of foreground object, reaches the purpose that accurately takes prospect.It is simple that blue screen is scratched diagram technology, and amount of calculation is little, and effective.But this method has limited its range of application to the strict demand of image background color, and significant limitation is arranged.
Natural image matting (natural image matting) technology has obtained good effect for natural picture.Wherein representative have bayes method (Bayesian matting, Y.-Y.Chuang, a B.Curless, D.H.Salesin, and R.Szeliski, " A bayesian approach to digital matting; " inProceedings of Computer Vision and Pattern Recognition, vol.2,2001, pp.264-271.), Poisson method (Poisson matting, J.Sun, J.Jia, C.-K.Tang, and H.-Y.Shum, " Poisson matting; " ACM Transactions on Graphics, vol.23, no.3, pp.315-321,2004.) and Easy matting method (Y.Guan, W.Chen, X.Liang, Z.Ding, and Q.Peng, " Easy matting-a stroke based approach for continuous image matting, " ComputerGraphics Forum, vol.25, no.3, pp.567-576,2006.).They generally can be divided into following two steps:
1, makes shade (trimap).Shade is meant picture is characterized by definite foreground area, definite background area and a division of zone of ignorance.The user comes the different zone of mark with the stroke of different colours usually.Making shade is a job that takes time and effort, particularly for the picture of prospect topological structure complexity.
2, find the solution the prospect F of zone of ignorance, background B and alpha value α.Wherein bayes method is estimated α according to the statistical information of sampling, calculates suitable prospect F and background B then.A shortcoming of this processing mode is to be easy to generate the color that does not have on the former picture, thereby produces visual distortion.And the Poisson method will be put in order pictures and be regarded a gradient fields as, estimate α by finding the solution Poisson's equation, in the fore/background zone of determining, find respectively then from waiting to ask a little nearest foreground point and background dot, as the prospect value and the background value of this point, avoid generating new color with their color effectively.The defective of this method is its quality as a result depends on strongly whether the input picture satisfies prospect and background is smooth hypothesis, and effect is unsatisfactory when the scene picture is complicated.
Bayes's video is scratched diagram technology (Bayes video matting, N.Apostoloff and A.W.Fitzgibbon, " Bayesian video matting using learnt image priors; " in Proceedings ofComputer Vision and Pattern Recognition, vol.1,2004, pp.407-414.) be the expansion that Bayes image is scratched drawing method.This method is scratched on the basis of figure at image, has adopted optic flow technique (opticalflow) to carry out bi-directional tracking between the frame, the shade on the key frame is diffused into whole video, thereby has reduced the workload of making shade.The defective of this method is that to containing high-speed motion, profile is out of shape, the video of contents such as topologies change, and effect is bad.On the other hand, the nuance between the consecutive frame can cause result's fluctuation, thereby causes visual discontinuity.This also is the total defective of stingy frame by frame figure that the natural image matting technology directly applies to video.
Adopt the stingy diagram technology of video of special hardware equipment to comprise that defocusing video scratches figure (Defocus videomatting, M.McGuire, W.Matusik, H.Pfister, J.F.Hughes, and F.Durand, " Defocusvideo matting; " ACM Transactions on Graphics, vol.24, no.3, pp.567-576,2005.) and the stingy figure of natural video frequency (natural video matting, N.Joshi, W.Matusik, and S.Avidan, " Natural video matting using camera arrays; " ACM Transactions on Graphics, vol.25, no.3, pp.779-786,2006.) etc.Defocusing the stingy figure of video is that first full-automatic video at dynamic scene is scratched drawing system.Its hardware components comprises three cameras that have identical photocentre and different focal.According to the difference of focal length, calculate shade automatically, calculate zone of ignorance on this basis.The defective of this method is, poor effect when prospect is close with background color.In addition, can cause tangible motion blur for moving scene.The hardware components that natural video frequency is scratched drawing method is made of a camera array.The colouring information that each camera obtains is projected on the degree of depth of prospect, calculates shade and outline according to statistical informations such as the average of these color values and variances then.This method can be with approaching real-time velocity process foreground color and the bigger video of background color gap.Its defective is to handle the foreground color video close with background color equally.
Stingy figure except stricti jurise, an other class technology of extracting objects claims target to clip and paste from video, mainly contain interactive video and shear (Interactive video cutout, J.Wang, P.Bhat, R.A.Colbum, M.Agrawala, and M.F.Cohen, " Interactive video cutout; " ACM Transactions onGraphics, vol.24, no.3, pp.585-594,2005.) and object video clip and paste (video object cutand paste, Y.Li, J.Sun, and H.-Y.Shum, " Video object cut and paste; " ACMTransactions on Graphics, vol.24, no.3, pp.595-600,2005.).They carry out the estimation of alpha value on the border of cutting apart on the basis of binary segmentation.Binary segmentation is meant that non-prospect is cutting apart firmly of background, and just the alpha value can only get 0 or 1.Because this binary limit, there is an intrinsic defective in this method, promptly only is applicable to the video that object boundary is smooth.
Summary of the invention
The objective of the invention is to overcome prior art, to handle the result that video brings frame by frame discontinuous, and the deficiency that the processing time is long provides a kind of fast video of interactively space-time unanimity to scratch drawing method.
The fast video of interactively space-time unanimity is scratched drawing method in the Digital Video Processing of the present invention, may further comprise the steps:
1) interactively body diffusion: the user calculates shade through diffusion again after carrying out alternately on the 3 D video body, the video body is divided into foreground area, background area and three parts of zone of ignorance;
2) the video outline of iteration takes: calculate the alpha value of every bit on the 3 D video body, i.e. opacity;
3) prospect of space-time unanimity is rebuild: according to the alpha value that calculates, calculate have successional foreground color value on space-time.
Said interactively body diffusion the steps include: among the present invention
1) by the user video body is carried out preliminary area dividing, it is divided into three parts: foreground area, background area and zone of ignorance;
2) set of the point in the zone of ignorance of user's mark is designated as Ω
e, set up following propagation energy equation group thereon:
Wherein p is Ω
eIn more arbitrarily, q is that the p point six is communicated with consecutive points, α on the video body
pBe the alpha value that p is ordered, I
pBe the RGB color value that p is ordered, ‖ I
p-I
q‖ is the Euclidean distance of p point and the q color of ordering;
3) find the solution above-mentioned system of linear equations, obtain the alpha estimated value of each point, if: the prospect threshold value is A, background threshold is B, the Alpha estimated value is grouped into foreground area greater than the dot-dash of A, the alpha estimated value is grouped into the background area less than the dot-dash of B, and remaining point still belongs to not to be determined to obtain a meticulous shade in the zone.The video outline of said iteration takes among the present invention, the steps include:
1) to each the some P in the zone of ignorance, from determine foreground area, gather this nearest M foreground point of space length, from this M foreground point, get colors again and estimate as prospect near K point of this point, from determine the background area, gather this nearest M background dot of space length, from this M background dot, get colors again and estimate as a setting near K point of this point;
2) set up following equation group:
I wherein
pBe the color value that P is ordered, F is the color value of foreground point in the sampling, and B is the color value of background dot in the sampling,
Be the three-dimensional Laplacian on the video body, σ
p 2Be I
pWith α
pF
p i+ (1-α
p) B
p j(i, and j ∈ (1,2, K, the K) variance of) Euclidean distance set, λ is an energy coefficient;
3) find the solution above-mentioned equation group, obtain the alpha value of every bit on the video; Divide the alpha value into definite foreground point greater than the point of prospect threshold value A, the alpha value divides definite background dot into less than the point of background threshold B;
4) carried out iteratively for 1 to 3 step, in the 3rd step, no longer include new point and divided into definite foreground point or definite background dot, obtain final outline.
The prospect of said space-time unanimity is rebuild among the present invention, the steps include:
1) the alpha value is labeled as Ω greater than background threshold A less than the set of the point of prospect threshold value B
m, at Ω
mOn set up following prospect energy equation group:
Wherein
Be the mean value of K the some color of estimating as a setting, σ
f 2Be coefficient of variation, γ is an energy coefficient,
Be the three-dimensional Laplacian on the video body, F
pBe the dedicated tunnel value of the p prospect RGB color of ordering, ‖ I
p-I
q‖ is the Euclidean distance of p point and q point color value;
2) space-time is as one man found the solution Ω
mThe value of middle each point RGB color R passage;
3) space-time is as one man found the solution Ω
mThe value of middle each point RGB color G passage;
4) space-time is as one man found the solution Ω
mThe value of middle each point RGB color B passage.
The invention has the advantages that:
Existing video is scratched diagram technology all needs the user to carry out meticulous shade making frame by frame, and this step is extremely loaded down with trivial details, and is consuming time huge, and stingy figure effect difficult to calculate and that control is final.This method required mutual amount on single-frame images is less than existing method far away, and this method also allows the user directly to carry out further having reduced amount of user effort alternately on the 3 D video body.
Method of the present invention takes in the step at the video outline, has considered the slickness constraint between color value constraint and the neighbor first simultaneously.In the constraint of neighbor, method of the present invention in the constraint of considering on the time dimension between the neighbor, has guaranteed temporal continuity first.Method of the present invention is set up energy equation according to these two constraints, and obtains the energy equation minimal solution, has fully satisfied the accuracy of color and result's continuity.
The present invention takes in the step at the video outline, and the method for sampling is immediate K the point that get colors from M nearest point of space length.The unsteadiness that the sum of errors of so just having avoided the fixing color probability Distribution Model of employing to be brought adopts the only a few sampled point to be brought, the sampled result robustness is significantly increased.
After outline takes, simply adopt the color of sampled point can not obtain satisfied effect as foreground color, can not guarantee the continuity on the room and time.As the mean value that adopts a plurality of sampled points also may produce the color that did not have originally as foreground color, causes visual deviation.This method proposes the prospect reconstruction technique first, has considered the constraint of continuity constraint and original color, thereby has obtained high-quality prospect.
In a word, application the present invention can fast and effeciently take out the prospect part in any video.The present invention has solved existing video well, and to scratch the result who exists in the drawing method discontinuous, the deficiency that the processing time is long, and on the agility of the convenience of user interactions, calculating and stingy figure result's continuity, method of the present invention all is significantly increased.
Description of drawings
Fig. 1 is a schematic flow sheet of the present invention;
Fig. 2 is the mutual schematic diagram of video body, and wherein, (1) is the section on the video body, and (2) are in the enterprising row labels of section, and (3) are the results of body diffusion, and (4) are the shades that obtains according to body diffusion result.
Embodiment
The flow process of the stingy drawing method of the fast video of interactively space-time unanimity is seen Fig. 1 in the Digital Video Processing of the present invention, may further comprise the steps:
The first step, the diffusion of interactively body: adopt on the 3 D video body, carry out mutual, the video body alternately referring to Fig. 2, the 3 D video body is made up of by the time preface each frame of video, comprises x axle, y axle and time shaft Time.
1) user carries out on the 3 D video body alternately, and the user can rotate arbitrarily the video body, section and cutting, and divides foreground area, background area and zone of ignorance by stroke mark on the video body of different colours.Fig. 2 (1) is a section in the video body, and Fig. 2 (2) is that F is a foreground area among the figure in the enterprising row labels of section, and B is the background area, and U is a zone of ignorance.
2) the some set in the zone of not determining with user's mark is designated as Ω
e, it is thin to set up the propagation energy equation thereon, as follows:
Wherein p is Ω
eIn more arbitrarily, q is that the p point six is communicated with consecutive points, α on the video body
pBe the alpha value that p is ordered, I
pBe the RGB color value that p is ordered, ‖ I
p-I
q‖ is the Euclidean distance of p point and q point color.
At Ω
eBoundary member, adopt second boundary condition.For example the point of first frame, first row, first row is established it and is a, and last, left, the place ahead that a is ordered upwards all do not have consecutive points.Increase by three virtual consecutive points on these three directions, alpha value is all with to put a identical, and promptly to lead be 0 to the single order of energy equation.
3) find the solution above-mentioned equation group with the biconjugate gradient method, obtain the estimated value of each some alpha value, the result of body diffusion is shown in Fig. 2 (3).If: the prospect threshold value A equals 0.97, and background threshold B equals 0.03, and the Alpha estimated value divides definite foreground point into greater than 0.97 point, and the alpha estimated value divides definite background dot into less than 0.03 point, and remaining point still belongs to does not determine the zone.So just obtained a meticulous shade, shade is shown in Fig. 2 (4), and among the figure, F is a foreground area, and B is the background area, and U is a zone of ignorance.
In second step, the video outline of iteration takes:
1) each in zone of ignorance point p gathers 25 nearest points of space length from determine foreground area, 5 points of ordering near p that therefrom get colors are again estimated as prospect.Gather 25 nearest points of space length from determine the background area, 5 points of ordering near p that therefrom get colors are estimated as a setting.
2) set up following stingy figure energy equation group:
I wherein
pBe the color value that p is ordered, F is the color value of foreground point in the sampling, and B is the color value of background dot in the sampling,
Be the three-dimensional Laplacian on the video body, σ
p 2Be I
pWith α
pF
p i+ (1-α
p) B
p j(i, and j ∈ (1,2, K, the K) variance of) Euclidean distance set, λ is an energy coefficient, λ is made as 0.01 in the present embodiment.
3) this example adopts the biconjugate gradient method to find the solution above-mentioned equation group, obtains new alpha value;
4) upgrade Ω
F +And Ω
B +Be about to the alpha value and be grouped into Ω greater than 0.97 point
F +In, the alpha value is grouped into Ω less than 0.03 point
B +In; Ω wherein
F +After being meant each iteration, the alpha value is greater than the set of 0.97 point in the zone of ignorance.Similarly, Ω
B +After being meant each iteration, the alpha value is less than the set of 0.03 point in the zone of ignorance.
5) with Ω
F +In dot-dash be classified as definite prospect, with Ω
B +In dot-dash be classified as definite background, empty Ω subsequently
F +And Ω
B +
6) repeating step 1) to 5), Ω in step 4)
F +And Ω
B +No longer including new point adds; Obtain final alpha estimated value, i.e. outline.
In the 3rd step, the prospect of space-time unanimity is rebuild:
1) with the alpha value greater than 0.03, the set of the point less than 0.97 is labeled as Ω
m, at Ω
mOn set up following prospect energy equation group:
Wherein
Be the mean value of K the some color of estimating as a setting, σ
f 2Be coefficient of variation, get 0.5 in the present embodiment, γ is an energy coefficient, gets 0.01 in the present embodiment,
Be the three-dimensional Laplacian on the video body, F
pBe the dedicated tunnel value of the p prospect RGB color of ordering, ‖ I
p-I
q‖ is the Euclidean distance of p point and q point color value;
2) space-time is as one man found the solution Ω
mThe value of middle each point RGB color R passage;
3) space-time is as one man found the solution Ω
mThe value of middle each point RGB color G passage;
4) space-time is as one man found the solution Ω
mThe value of middle each point RGB color B passage.
By above step, can partly carry out quick, high-quality taking to the prospect in any video.
Table one has been listed in the present embodiment, and the inventive method and representative bayes method are under the environment of 2.8GHz CPU and 2.0G internal memory, and handling one section resolution is 640 * 480, when length is the video of 79 frames, and the processing time that each step is required.The user interactions time wherein is meant the average time that does not have trained domestic consumer required.Can see that under same environment, Bayes's video is scratched drawing method needs 150 minute processing time altogether, and method of the present invention only needs 30 minutes, and on the continuity of scratching figure result, method of the present invention there is remarkable improvement.
Table one distinct methods is the required processing time under the environment of 2.8 GHz CPU and 2.0G internal memory
Method name | User interactions | Diffusion | Take | Prospect is rebuild | Total time |
Bayes method | 30 minutes | 105 minutes | 15 minutes | Do not have | 150 minutes |
The inventive method | 10 minutes | 8 minutes | 10 |
2 minutes | 30 minutes |
What more than enumerate only is specific embodiments of the invention.Obviously, the invention is not restricted to above embodiment, many distortion can also be arranged.All distortion that those of ordinary skill in the art can directly derive or associate from content disclosed by the invention all should be thought protection scope of the present invention.
Claims (3)
1, the video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing, it is characterized in that may further comprise the steps:
1) interactively body diffusion: the user calculates shade through diffusion again after carrying out alternately on the 3 D video body, the video body is divided into foreground area, background area and three parts of zone of ignorance;
2) the video outline of iteration takes: calculate the alpha value of every bit on the 3 D video body, i.e. opacity;
3) prospect of space-time unanimity is rebuild: according to the alpha value that calculates, calculate have successional foreground color value on space-time.
2, the video of interactive time-space unanimity is scratched drawing method in the Digital Video Processing according to claim 1, it is characterized in that said interactively body diffusion, the steps include:
1) by the user video body is carried out preliminary area dividing, it is divided into three parts: foreground area, background area and zone of ignorance;
2) set of the point in the zone of ignorance of user's mark is designated as Ω e, sets up following propagation energy equation group thereon:
Wherein p is Ω
eIn more arbitrarily, q is that the p point six is communicated with consecutive points, α on the video body
pBe the alpha value that p is ordered, I
pBe the RGB color value that p is ordered, ‖ I
p-I
q‖ is the Euclidean distance of p point and the q color of ordering;
3) find the solution above-mentioned system of linear equations, obtain the alpha estimated value of each point, if: the prospect threshold value is A, background threshold is B, the Alpha estimated value is grouped into foreground area greater than the dot-dash of A, the alpha estimated value is grouped into the background area less than the dot-dash of B, and remaining point still belongs to not to be determined to obtain a meticulous shade in the zone.
3, the video of interactive time-space unanimity is scratched drawing method in the Digital Video Processing according to claim 1, it is characterized in that the video outline of said iteration takes, and the steps include:
1) to each the some p in the zone of ignorance, from determine foreground area, gather this nearest M foreground point of space length, from this M foreground point, get colors again and estimate as prospect near K point of this point, from determine the background area, gather this nearest M background dot of space length, from this M background dot, get colors again and estimate as a setting near K point of this point;
2) set up following equation group:
I wherein
pBe the color value that p is ordered, F is the color value of foreground point in the sampling, and B is the color value of background dot in the sampling,
Be the three-dimensional Laplacian on the video body, σ
p 2Be I
pWith α
pF
p i+ (1-α
p) B
p j(i, and j ∈ (1,2, K, the K) variance of) Euclidean distance set, λ is an energy coefficient;
3) find the solution above-mentioned equation group, obtain the alpha value of every bit on the video; Divide the alpha value into definite foreground point greater than the point of prospect threshold value A, the alpha value divides definite background dot into less than the point of background threshold B;
4) carried out iteratively for 1 to 3 step, in the 3rd step, no longer include new point and divided into definite foreground point or definite background dot, obtain final outline.
4, the video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing according to claim 1, it is characterized in that the prospect of said space-time unanimity is rebuild, and the steps include:
1) the alpha value is labeled as Ω greater than background threshold A less than the set of the point of prospect threshold value B
m,
Ω
mOn set up following prospect energy equation group:
Wherein
Be the mean value of K the some color of estimating as a setting, σ
f 2Be coefficient of variation, γ is an energy
Coefficient,
Be the three-dimensional Laplacian on the video body, F
pBe that p is ordered
The dedicated tunnel value of prospect RGB color, ‖ I
p-I
q‖ is the Euclidean distance of p point and q point color value;
2) space-time is as one man found the solution Ω
mThe value of middle each point RGB color R passage;
3) space-time is as one man found the solution Ω
mThe value of middle each point RGB color G passage;
4) space-time is as one man found the solution Ω
mThe value of middle each point RGB color B passage.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2007100699707A CN100539698C (en) | 2007-07-10 | 2007-07-10 | The video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2007100699707A CN100539698C (en) | 2007-07-10 | 2007-07-10 | The video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101098475A true CN101098475A (en) | 2008-01-02 |
CN100539698C CN100539698C (en) | 2009-09-09 |
Family
ID=39011938
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2007100699707A Expired - Fee Related CN100539698C (en) | 2007-07-10 | 2007-07-10 | The video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100539698C (en) |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101588459B (en) * | 2009-06-26 | 2011-01-05 | 北京交通大学 | Video keying processing method |
CN101582168B (en) * | 2009-06-16 | 2011-06-15 | 武汉大学 | Matting sample set construction method based on fuzzy connectedness |
CN103049894A (en) * | 2012-09-14 | 2013-04-17 | 深圳市万兴软件有限公司 | Image processing method and device |
CN103400386A (en) * | 2013-07-30 | 2013-11-20 | 清华大学深圳研究生院 | Interactive image processing method used for video |
CN103474009A (en) * | 2013-09-22 | 2013-12-25 | 深圳市优一城商业管理有限公司 | Wisdom media system and interacting method |
CN103475826A (en) * | 2013-09-27 | 2013-12-25 | 深圳市中视典数字科技有限公司 | Video matting and synthesis method |
CN103559509A (en) * | 2013-11-06 | 2014-02-05 | 飞歌康采(北京)软件科技有限公司 | Real-time image matting method based on scene information |
CN104200470A (en) * | 2014-08-29 | 2014-12-10 | 电子科技大学 | Blue screen image-matting method |
CN105120185A (en) * | 2015-08-27 | 2015-12-02 | 新奥特(北京)视频技术有限公司 | Image matting method and device for video images |
CN103942794B (en) * | 2014-04-16 | 2016-08-31 | 南京大学 | A kind of image based on confidence level is collaborative scratches drawing method |
CN106204567A (en) * | 2016-07-05 | 2016-12-07 | 华南理工大学 | A kind of natural background video matting method |
CN107134194A (en) * | 2017-05-18 | 2017-09-05 | 河北中科恒运软件科技股份有限公司 | Immersion vehicle simulator |
CN107610041A (en) * | 2017-08-16 | 2018-01-19 | 南京华捷艾米软件科技有限公司 | Video portrait based on 3D body-sensing cameras scratches drawing method and system |
CN105590309B (en) * | 2014-10-23 | 2018-06-15 | 株式会社理光 | Foreground image dividing method and device |
CN113253890A (en) * | 2021-04-02 | 2021-08-13 | 中南大学 | Video image matting method, system and medium |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI693576B (en) * | 2019-02-26 | 2020-05-11 | 緯創資通股份有限公司 | Method and system for image blurring processing |
-
2007
- 2007-07-10 CN CNB2007100699707A patent/CN100539698C/en not_active Expired - Fee Related
Cited By (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101582168B (en) * | 2009-06-16 | 2011-06-15 | 武汉大学 | Matting sample set construction method based on fuzzy connectedness |
CN101588459B (en) * | 2009-06-26 | 2011-01-05 | 北京交通大学 | Video keying processing method |
CN103049894A (en) * | 2012-09-14 | 2013-04-17 | 深圳市万兴软件有限公司 | Image processing method and device |
CN103400386A (en) * | 2013-07-30 | 2013-11-20 | 清华大学深圳研究生院 | Interactive image processing method used for video |
CN103400386B (en) * | 2013-07-30 | 2016-08-31 | 清华大学深圳研究生院 | A kind of Interactive Image Processing method in video |
CN103474009B (en) * | 2013-09-22 | 2015-10-28 | 深圳市优一城商业管理有限公司 | Wisdom media system and exchange method |
CN103474009A (en) * | 2013-09-22 | 2013-12-25 | 深圳市优一城商业管理有限公司 | Wisdom media system and interacting method |
CN103475826A (en) * | 2013-09-27 | 2013-12-25 | 深圳市中视典数字科技有限公司 | Video matting and synthesis method |
CN103559509A (en) * | 2013-11-06 | 2014-02-05 | 飞歌康采(北京)软件科技有限公司 | Real-time image matting method based on scene information |
CN103559509B (en) * | 2013-11-06 | 2017-02-08 | 汇隆基业科技(北京)有限责任公司 | Real-time image matting method based on scene information |
CN103942794B (en) * | 2014-04-16 | 2016-08-31 | 南京大学 | A kind of image based on confidence level is collaborative scratches drawing method |
CN104200470A (en) * | 2014-08-29 | 2014-12-10 | 电子科技大学 | Blue screen image-matting method |
CN104200470B (en) * | 2014-08-29 | 2017-02-08 | 电子科技大学 | Blue screen image-matting method |
CN105590309B (en) * | 2014-10-23 | 2018-06-15 | 株式会社理光 | Foreground image dividing method and device |
CN105120185A (en) * | 2015-08-27 | 2015-12-02 | 新奥特(北京)视频技术有限公司 | Image matting method and device for video images |
CN105120185B (en) * | 2015-08-27 | 2018-05-04 | 新奥特(北京)视频技术有限公司 | A kind of video image is scratched as method and apparatus |
CN106204567A (en) * | 2016-07-05 | 2016-12-07 | 华南理工大学 | A kind of natural background video matting method |
CN106204567B (en) * | 2016-07-05 | 2019-01-29 | 华南理工大学 | A kind of natural background video matting method |
CN107134194A (en) * | 2017-05-18 | 2017-09-05 | 河北中科恒运软件科技股份有限公司 | Immersion vehicle simulator |
CN107610041A (en) * | 2017-08-16 | 2018-01-19 | 南京华捷艾米软件科技有限公司 | Video portrait based on 3D body-sensing cameras scratches drawing method and system |
CN107610041B (en) * | 2017-08-16 | 2020-10-27 | 南京华捷艾米软件科技有限公司 | Video portrait matting method and system based on 3D somatosensory camera |
CN113253890A (en) * | 2021-04-02 | 2021-08-13 | 中南大学 | Video image matting method, system and medium |
CN113253890B (en) * | 2021-04-02 | 2022-12-30 | 中南大学 | Video image matting method, system and medium |
Also Published As
Publication number | Publication date |
---|---|
CN100539698C (en) | 2009-09-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100539698C (en) | The video of interactive time-space unanimity is scratched drawing method in a kind of Digital Video Processing | |
Märki et al. | Bilateral space video segmentation | |
Xiao et al. | Fast image dehazing using guided joint bilateral filter | |
Gerrits et al. | Local stereo matching with segmentation-based outlier rejection | |
Wang et al. | An iterative optimization approach for unified image segmentation and matting | |
CN103942794B (en) | A kind of image based on confidence level is collaborative scratches drawing method | |
CN104935832B (en) | For the video keying method with depth information | |
CN104820990A (en) | Interactive-type image-cutting system | |
CN104732506A (en) | Character picture color style converting method based on face semantic analysis | |
CN103839223A (en) | Image processing method and image processing device | |
CN103559719A (en) | Interactive graph cutting method | |
Xiao et al. | Accurate motion layer segmentation and matting | |
CN101770649A (en) | Automatic synthesis method for facial image | |
CN108230243B (en) | Background blurring method based on salient region detection model | |
CN109559328B (en) | Bayesian estimation and level set-based rapid image segmentation method and device | |
CN102420985B (en) | Multi-view video object extraction method | |
CN104915946B (en) | A kind of object segmentation methods based on conspicuousness suitable for serious degraded image | |
CN104715451A (en) | Seamless image fusion method based on consistent optimization of color and transparency | |
Xie et al. | Seamless video composition using optimized mean-value cloning | |
CN102005061B (en) | Method for reusing cartoons based on layering/hole-filling | |
Lermé et al. | Reducing graphs in graph cut segmentation | |
CN112348972A (en) | Fine semantic annotation method based on large-scale scene three-dimensional model | |
CN104159098A (en) | Time-domain consistent semi-transparent edge extraction method for video | |
CN106469440A (en) | Dark mist elimination parallel optimization method based on OpenCL | |
CN113421210A (en) | Surface point cloud reconstruction method based on binocular stereo vision |
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 | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090909 Termination date: 20100710 |