CN106952276A - A kind of image matting method and device - Google Patents
A kind of image matting method and device Download PDFInfo
- Publication number
- CN106952276A CN106952276A CN201710164954.XA CN201710164954A CN106952276A CN 106952276 A CN106952276 A CN 106952276A CN 201710164954 A CN201710164954 A CN 201710164954A CN 106952276 A CN106952276 A CN 106952276A
- Authority
- CN
- China
- Prior art keywords
- image
- rgbd
- ternary diagram
- target area
- target 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of image matting method, including：The image and target frame of input are obtained, and builds the depth information of target frame correspondence image；The figure to be formed is modeled to RGBD using gauss hybrid models and cuts algorithm to target frame correspondence image progress coarse segmentation, target area is obtained；Wherein, D is depth information；Corrosion expansive working is carried out to target area, ternary diagram is obtained；The transparence value of ternary diagram is determined, acquisition prospect accurately extracts result；This method is compared to the stingy drawing method of tradition, and to color, illumination variation has more preferable robustness, and precision is higher, and it is that can obtain preferably stingy figure result only to need a small amount of user mutual；Map device is scratched the invention also discloses a kind of image, with above-mentioned beneficial effect.
Description
Technical field
The present invention relates to technical field of computer vision, more particularly to a kind of image matting method and device.
Background technology
It is a kind of technology extracted using algorithm to the foreground target in image that image, which scratches diagram technology,.Diagram technology is scratched to exist
Production of film and TV, has application widely in image synthesis.
The technology usually requires user and gives certain interaction, such as marks object institute roughly with a rectangle frame in place
Put；Or sketched the contours of in prospect, background must shape.Pass through interaction, it is possible to provide rough prospect, background information,
It is then placed in algorithm, it is possible to obtain comparatively fine foreground extraction result.At present, image scratches diagram technology in algorithm reality
There are many improvable spaces in Shi Xing, robustness, correctness.Particularly in terms of to color and illumination variation precision compared with
It is low.Therefore robustness and accuracy of the stingy figure in terms of color, illumination variation how are improved, is that those skilled in the art need solution
Technical problem certainly.
The content of the invention
It is an object of the invention to provide a kind of image matting method, device, to color, illumination variation has more preferable robust
Property, and precision is higher, it is that can obtain preferably stingy figure result only to need a small amount of user mutual.
In order to solve the above technical problems, the present invention provides a kind of image matting method, including：
The image and target frame of input are obtained, and builds the depth information of target frame correspondence image；
The figure to be formed is modeled to RGBD using gauss hybrid models and cuts algorithm to target frame correspondence image progress rough segmentation
Cut, obtain target area；Wherein, D is the depth information；
Corrosion expansive working is carried out to the target area, ternary diagram is obtained；
The transparence value of the ternary diagram is determined, acquisition prospect accurately extracts result.
Optionally, model the figure to be formed to RGBD using gauss hybrid models and cut algorithm to enter the target frame correspondence image
Row coarse segmentation, obtains target area, including：
Build the S-T figures of the target frame correspondence image；
The RGBD of each pixel in the S-T figures is modeled using gauss hybrid models and obtains corresponding RGBD
It is worth, and it is each in the difference imparting S-T figures according to the RGBD values in the S-T figures per corresponding two pixels of a line
The weight on bar side；
According to the weight in the S-T figures per a line, split using maximum-flow algorithm, and to the image after segmentation
Contour detecting is carried out, area largest contours are determined；
The area largest contours are filled with acquisition target area.
Optionally, corrosion expansive working is carried out to the target area, obtains ternary diagram, including：
To the target area, using wide 5% and 5% high pixel-parameters of the target frame, corroded
Expansive working, obtains ternary diagram.
Optionally, the transparence value of the ternary diagram is determined, including：
The transparence value of the ternary diagram is calculated using shared sample point algorithm.
Optionally, this programme also includes：
According to the amendment frame of input, the iteration execution utilization gauss hybrid models model the figure to be formed to RGBD and cut algorithm
Coarse segmentation is carried out to the target frame correspondence image, the step of obtaining target area.
Optionally, the image of input is obtained, including：
Obtain the image of dual camera image acquisition device.
The present invention also provides a kind of image and scratches map device, including：
Image collection module, image and target frame for obtaining input；
Depth information builds module, the depth information for building target frame correspondence image；
Image coarse segmentation module, algorithm is cut to the target for modeling the figure to be formed to RGBD using gauss hybrid models
Frame correspondence image carries out coarse segmentation, obtains target area；Wherein, D is the depth information；
Ternary diagram acquisition module, for carrying out corrosion expansive working to the target area, obtains ternary diagram；
The accurate extraction module of image, the transparence value for determining the ternary diagram, acquisition prospect accurately extracts result.
Optionally, the depth information builds module, including：
S-T figure member units, the S-T for building the target frame correspondence image schemes；
Weight assignment unit, for being carried out using gauss hybrid models to the RGBD of each pixel in the S-T figures
Modeling obtains corresponding RGBD values, and according to the difference of the RGBD values in the S-T figures per corresponding two pixels of a line
Assign the weight per a line in the S-T figures；
Outline specifying unit, for according to the weight in the S-T figures per a line, being divided using maximum-flow algorithm
Cut, and contour detecting is carried out to the image after segmentation, determine area largest contours；
Target area determining unit, for the area largest contours to be filled with acquisition target area.
Optionally, the accurate extraction module of described image, including：
Transparence value computing unit, the transparence value for calculating the ternary diagram using shared sample point algorithm.
Optionally, this programme also includes：
Iteration module, for the amendment frame according to input, iteration performs described image coarse segmentation module.
A kind of image matting method provided by the present invention, including：The image and target frame of input are obtained, and builds mesh
Mark the depth information of frame correspondence image；The figure to be formed is modeled to RGBD using gauss hybrid models and cuts algorithm to target frame corresponding diagram
As carrying out coarse segmentation, target area is obtained；Wherein, D is depth information；Corrosion expansive working is carried out to target area, three are obtained
Member figure；The transparence value of ternary diagram is determined, acquisition prospect accurately extracts result；
It can be seen that, this method is compared to the stingy drawing method of tradition, and to color, illumination variation has more preferable robustness, and precision
Higher, it is that can obtain preferably stingy figure result only to need a small amount of user mutual；Map device, tool are scratched present invention also offers a kind of image
There is above-mentioned beneficial effect, will not be repeated here.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are only this
The embodiment of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
The accompanying drawing of offer obtains other accompanying drawings.
The flow chart for the image matting method that Fig. 1 is provided by the embodiment of the present invention；
The schematic diagram of the S-T figures for the structure target frame correspondence image that Fig. 2 is provided by the embodiment of the present invention；
The schematic flow sheet for the image matting method that Fig. 3 is provided by the embodiment of the present invention；
The effect diagram for the image matting method that Fig. 4 is provided by the embodiment of the present invention；
The image that Fig. 5 is provided by the embodiment of the present invention scratches the structured flowchart of drawing system.
Embodiment
The core of the present invention is to provide a kind of image matting method, device, and to color, illumination variation has more preferable robust
Property, and precision is higher, it is that can obtain preferably stingy figure result only to need a small amount of user mutual.
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
With continuing to develop for development of Mobile Internet technology, the depth information of objects in images calculates comparative maturity, and
Many terminals have been provided with dual camera.The image shot using dual camera can obtain the depth information of object, with reference to depth
Degree information is obtained with more preferable foreground extraction result.Specifically it refer to Fig. 1, the figure that Fig. 1 is provided by the embodiment of the present invention
Flow chart as scratching drawing method；This method can include：
S100, the image and target frame for obtaining input, and build the depth information of target frame correspondence image.
Specifically, depth information of the step mainly for acquisition image, and the prospect institute that user outlines in the picture
Target frame in position.The present embodiment does not limit specific acquisition image, and the method for calculating the image depth information, only
The depth information for the image that can be collected.And the present embodiment does not limit the shape of target frame, for example generally
In the case of, the target frame is rectangle frame.
Wherein, the depth information of image is built for convenience, and image here can carry out image using dual camera
Collection, you can choosing, using dual camera image acquisition device image, dual camera image collecting device here can be with
It is the mobile phone with dual camera, i.e., using the dual camera collection image of mobile phone, and image is gone out according to the Image Reconstruction of collection
Depth information.The acquisition methods of depth information are more than that, naturally it is also possible to be to obtain multiple image using monocular, reconstruct
Depth information be also completely can with.
S110, the figure to be formed is modeled using gauss hybrid models to RGBD cut algorithm rough segmentation is carried out to target frame correspondence image
Cut, obtain target area；Wherein, D is depth information.
Specifically, the step mainly completes the coarse segmentation of image.Current image partition method species is various.It can substantially divide
For several classes：Dividing method based on threshold value, the dividing method based on edge, the dividing method based on region, point based on graph theory
Segmentation method, the dividing method based on energy functional, the method based on deep learning.Relatively more and ripe algorithms are applied at present
Have：Mean shift algorithm, watershed algorithm, figure cuts algorithm, deep learning algorithm.The present embodiment cuts algorithm using figure and carries out image
Coarse segmentation, because the algorithm is modeled using gauss hybrid models (i.e. GMM) to RGB.This causes the algorithm to color complex situations
Under do not possess preferable segmentation ability.But the application is improved to other factors such as illumination variations in order to further on this basis
Segmentation ability, therefore algorithm cut to traditional figure be in the present embodiment improved；It is specific that RGBD is modeled using GMM, D
Depth information is represented, after depth information, the segmentation effect of image will have greatly improved.
It is preferred that, using gauss hybrid models RGBD is modeled the figure to be formed cut algorithm target frame correspondence image is carried out it is thick
Segmentation, obtaining target area can include：
Build the S-T figures of target frame correspondence image；
Specifically, the effect of the S-T figures built may be referred to Fig. 2, S-T figures have a source node S and a terminal node
T.Each pixel in other node table diagram pictures.
The RGBD of each pixel in S-T figures is modeled using gauss hybrid models and obtains corresponding RGBD values,
And the weight in S-T figures per a line is assigned according to the difference of the RGBD values in S-T figures per corresponding two pixels of a line；
Specifically, the step is mainly realized is assigned to weight to every a line of S-T figures.I.e. using GMM to each pixel
The RGBD of point is modeled, and big weights are assigned to if the RGBD values between two pixels are close, represents to separate two pixels
Cost it is larger.If RGBD values differ greatly, less weights are assigned to, represent that the cost for separating two pixels is smaller.This
The difference and the corresponding relation of weighted value of the RGBD values of two pixels are not limited in embodiment specifically.
According to the weight in S-T figures per a line, split using maximum-flow algorithm, and the image after segmentation is carried out
Contour detecting, determines area largest contours；Area largest contours are filled with acquisition target area.
Specifically, being all assigned to every a line after weights, segmentation cost is found using maximum-flow algorithm (i.e. maxflow)
Minimum dividing method, completes segmentation.After segmentation is completed, contour detecting is carried out to image, and detects largest contours, with this
As preliminary target region, this region is filled, you can obtain target area.
S120, to target area carry out corrosion expansive working, obtain ternary diagram；
Specifically, the step is mainly for acquisition ternary diagram (i.e. trimap).I.e. image, which scratches figure, a prerequisite, that
Must exactly there is the corresponding ternary diagram of image.At present, most ternary diagram is all obtained by manual delineation.Manual delineation is obtained
Take ternary diagram cumbersome, the finer object such as hairline of correspondence then needs to waste many times.Also proposition utilizes
TOF camera obtains depth map and generates ternary diagram.But traditional utilization TOF camera obtains depth map and generates the side of ternary diagram
Formula only considered depth information, and with many defects, accurate extraction can not be realized for the much the same foreground target of depth information,
And threshold parameter sets also relatively difficult.Therefore in order to overcome drawbacks described above, the present embodiment proposes to utilize step S110 generations
Target area, to its border carry out corrosion expansion obtain ternary diagram.For example user institute gives rectangle frame to correspond to wide, high 5%
Pixel, it is possible to achieve automatic ternary diagram generation.It is i.e. preferred, to target area, utilize wide 5% of target frame and high
5% pixel-parameters, carry out corrosion expansive working, obtain ternary diagram.5% numerical value provided in the present embodiment be only one compared with
Good effect, is not defined to the numerical value.
S130, the transparence value for determining ternary diagram, acquisition prospect accurately extract result.
Specifically, the step should not realize the accurate extraction to image.Cut that algorithm obtains due to figure is the knot split firmly
Really, therefore to there is border unsmooth, some problems are not as a result waited finely.So, in order that obtained result is finer, put down
It is sliding, so using soft partitioning algorithm, to determine the transparence value of image.Specific soft segmentation is not limited in the present embodiment to calculate
Method.As long as the transparence value of ternary diagram can be determined.At present, the algorithm for calculating images transparent angle value is more, as classical
Closed Form scratch figure, and Bayes scratches nomography.
Further, the present embodiment considers practicality, Riming time of algorithm and extraction effect, it is preferred that using shared
Sample point methods carry out images transparent angle value calculating, except of course that outside shared sample point algorithm, other stingy nomographys also completely may be used
To be used as substitute.Here it is that an effect is preferable to share sample point methods, the stronger algorithm of practicality.It is i.e. optional, really
Determining the transparence value of ternary diagram can include：
The transparence value of ternary diagram is calculated using shared sample point algorithm.
Specifically, shared sample point algorithm to each pixel in zone of ignorance by finding a pair of prospects, background picture
Vegetarian refreshments, afterwards based on following formula：
Cp=αpFp+(1-αp)Bp
Wherein, CpIt is the rgb value of zone of ignorance pixel p, Fp, BpIt is prospect background pixel pair.αpIt is transparence value.According to
Above-mentioned formula and prospect background sample point are to can just calculate the transparence value of zone of ignorance pixel.
The idiographic flow of image matting method is carried out in the present embodiment can include image acquisition, image coarse segmentation, ternary
Figure generation, image is accurately extracted.Specifically it refer to Fig. 3.Whole process only needs to subscriber frame and selects a target frame, afterwards whole mistake
Journey is performed entirely automatically.Have the advantages that interaction is few.Played in addition, calculating transparence value to image also for the accurate extraction of image
Very crucial effect.Whole process has thick to thin, and the accurate extraction of foreground target is done step-by-step.Implementing effect can be with
With reference to Fig. 4.
Based on above-mentioned technical proposal, image matting method provided in an embodiment of the present invention is only needed for providing a prospect
The target frame of position, it is not necessary to which too many user interactive can just realize that display foreground is extracted.Specifically utilize computer
Vision and machine learning techniques realize the extracted with high accuracy of target.Compared to conventional method to color, illumination variation more robust,
And precision is higher, it is that can obtain preferable result only to need a small amount of user mutual.Therefore this method can be synthesized in image, video display system
It is used widely in the application of other computer visions such as work.
Based on above-described embodiment, in order to further improve precision, step can be improved according to the interaction of user, successive ignition
The result that rapid S110 is obtained, reaches the perfect extraction of target.It is i.e. optional, it can also include in the present embodiment：
According to the amendment frame of input, iteration execution models the figure to be formed to RGBD using gauss hybrid models and cuts algorithm to mesh
Mark frame correspondence image and carry out coarse segmentation, the step of obtaining target area.
Specifically, whether the interactive iteration process of the step can need to be selected according to user.Here amendment frame
I.e. user interacts what is inputted during iteration, can obtain more accurate target area according to this, and then can obtain more
Accurate ternary diagram, is finally achieved the accurate extraction that image scratches figure.
Based on above-mentioned technical proposal, image matting method provided in an embodiment of the present invention, to color, illumination variation has more
Good robustness, and precision is higher, it is that can obtain preferably stingy figure result only to need a small amount of user mutual.And can be selected according to user
Select and be iterated calculating, obtain more excellent stingy figure effect, and improve the independence of user, improve Consumer's Experience.
Below to image provided in an embodiment of the present invention scratch map device be introduced, image described below scratch map device with
Above-described image matting method can be mutually to should refer to.
Fig. 5 is refer to, the image that Fig. 5 is provided by the embodiment of the present invention scratches the structured flowchart of drawing system；The system can be with
Including：
Image collection module 100, image and target frame for obtaining input；
Depth information builds module 200, the depth information for building target frame correspondence image；
Image coarse segmentation module 300, algorithm is cut to target for modeling the figure to be formed to RGBD using gauss hybrid models
Frame correspondence image carries out coarse segmentation, obtains target area；Wherein, D is depth information；
Ternary diagram acquisition module 400, for carrying out corrosion expansive working to target area, obtains ternary diagram；
The accurate extraction module 500 of image, the transparence value for determining ternary diagram, acquisition prospect accurately extracts result.
Based on above-described embodiment, the depth information, which builds module 200, to be included：
S-T figure member units, the S-T for building target frame correspondence image schemes；
Weight assignment unit, for being modeled using gauss hybrid models to the RGBD of each pixel in S-T figures
Corresponding RGBD values are obtained, and S-T figures are assigned according to the difference of the RGBD values in S-T figures per corresponding two pixels of a line
In per a line weight；
Outline specifying unit, for according to the weight in S-T figures per a line, being split using maximum-flow algorithm, and
Contour detecting is carried out to the image after segmentation, area largest contours are determined；
Target area determining unit, for area largest contours to be filled with acquisition target area.
Based on above-mentioned any embodiment, the accurate extraction module 500 of described image can include：
Transparence value computing unit, the transparence value for calculating ternary diagram using shared sample point algorithm.
Based on above-mentioned any embodiment, the system can also include：
Iteration module, for the amendment frame according to input, iteration performs described image coarse segmentation module.
The embodiment of each in specification is described by the way of progressive, and what each embodiment was stressed is and other realities
Apply the difference of example, between each embodiment identical similar portion mutually referring to.For device disclosed in embodiment
Speech, because it is corresponded to the method disclosed in Example, so description is fairly simple, related part is referring to method part illustration
.
Professional further appreciates that, with reference to the unit of each example of the embodiments described herein description
And algorithm steps, can be realized with electronic hardware, computer software or the combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software, generally describes the composition and step of each example according to function in the above description.These
Function is performed with hardware or software mode actually, depending on the application-specific and design constraint of technical scheme.Specialty
Technical staff can realize described function to each specific application using distinct methods, but this realization should not
Think beyond the scope of this invention.
Directly it can be held with reference to the step of the method or algorithm that the embodiments described herein is described with hardware, processor
Capable software module, or the two combination are implemented.Software module can be placed in random access memory (RAM), internal memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
A kind of image matting method and device provided by the present invention are described in detail above.It is used herein
Specific case is set forth to the principle and embodiment of the present invention, and the explanation of above example is only intended to help and understands this
The method and its core concept of invention.It should be pointed out that for those skilled in the art, not departing from this hair
On the premise of bright principle, some improvement and modification can also be carried out to the present invention, these are improved and modification also falls into power of the present invention
In the protection domain that profit is required.
Claims (10)
1. a kind of image matting method, it is characterised in that including：
The image and target frame of input are obtained, and builds the depth information of target frame correspondence image；
The figure to be formed is modeled to RGBD using gauss hybrid models and cuts algorithm to target frame correspondence image progress coarse segmentation, is obtained
Take target area；Wherein, D is the depth information；
Corrosion expansive working is carried out to the target area, ternary diagram is obtained；
The transparence value of the ternary diagram is determined, acquisition prospect accurately extracts result.
2. image matting method according to claim 1, it is characterised in that shape is modeled to RGBD using gauss hybrid models
Into figure cut algorithm to the target frame correspondence image carry out coarse segmentation, obtain target area, including：
Build the S-T figures of the target frame correspondence image；
The RGBD of each pixel in the S-T figures is modeled using gauss hybrid models and obtains corresponding RGBD values,
And each is assigned in the S-T figures according to the difference of the RGBD values in the S-T figures per corresponding two pixels of a line
The weight on side；
According to the weight in the S-T figures per a line, split using maximum-flow algorithm, and the image after segmentation is carried out
Contour detecting, determines area largest contours；
The area largest contours are filled with acquisition target area.
3. image matting method according to claim 1, it is characterised in that corrosion expansion behaviour is carried out to the target area
Make, obtain ternary diagram, including：
To the target area, using wide 5% and 5% high pixel-parameters of the target frame, corrosion expansion is carried out
Operation, obtains ternary diagram.
4. image matting method according to claim 1, it is characterised in that determine the transparence value of the ternary diagram, bag
Include：
The transparence value of the ternary diagram is calculated using shared sample point algorithm.
5. the image matting method according to claim any one of 1-5, it is characterised in that also include：
According to the amendment frame of input, the iteration execution utilization gauss hybrid models model the figure to be formed to RGBD and cut algorithm to institute
State target frame correspondence image and carry out coarse segmentation, the step of obtaining target area.
6. image matting method according to claim 5, it is characterised in that obtain the image of input, including：
Obtain the image of dual camera image acquisition device.
7. a kind of image scratches map device, it is characterised in that including：
Image collection module, image and target frame for obtaining input；
Depth information builds module, the depth information for building target frame correspondence image；
Image coarse segmentation module, algorithm is cut to the target frame pair for modeling the figure to be formed to RGBD using gauss hybrid models
Answer image to carry out coarse segmentation, obtain target area；Wherein, D is the depth information；
Ternary diagram acquisition module, for carrying out corrosion expansive working to the target area, obtains ternary diagram；
The accurate extraction module of image, the transparence value for determining the ternary diagram, acquisition prospect accurately extracts result.
8. image according to claim 7 scratches map device, it is characterised in that the depth information builds module, including：
S-T figure member units, the S-T for building the target frame correspondence image schemes；
Weight assignment unit, for being modeled using gauss hybrid models to the RGBD of each pixel in the S-T figures
Corresponding RGBD values are obtained, and are assigned according to the difference of the RGBD values in the S-T figures per corresponding two pixels of a line
Weight in the S-T figures per a line；
Outline specifying unit, for according to the weight in the S-T figures per a line, being split using maximum-flow algorithm, and
Contour detecting is carried out to the image after segmentation, area largest contours are determined；
Target area determining unit, for the area largest contours to be filled with acquisition target area.
9. image according to claim 7 scratches map device, it is characterised in that the accurate extraction module of described image, including：
Transparence value computing unit, the transparence value for calculating the ternary diagram using shared sample point algorithm.
10. the image according to claim any one of 7-9 scratches map device, it is characterised in that also include：
Iteration module, for the amendment frame according to input, iteration performs described image coarse segmentation module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710164954.XA CN106952276A (en) | 2017-03-20 | 2017-03-20 | A kind of image matting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710164954.XA CN106952276A (en) | 2017-03-20 | 2017-03-20 | A kind of image matting method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106952276A true CN106952276A (en) | 2017-07-14 |
Family
ID=59472249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710164954.XA Pending CN106952276A (en) | 2017-03-20 | 2017-03-20 | A kind of image matting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952276A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171716A (en) * | 2017-12-25 | 2018-06-15 | 北京奇虎科技有限公司 | Video personage based on the segmentation of adaptive tracing frame dresss up method and device |
| WO2020062899A1 (en) * | 2018-09-26 | 2020-04-02 | 惠州学院 | Method for obtaining transparency masks by means of foreground and background pixel pairs and grayscale information |
| CN111899266A (en) * | 2020-07-17 | 2020-11-06 | 深圳奥比中光科技有限公司 | Matting method and system based on RGBD camera |
| CN112241960A (en) * | 2020-10-01 | 2021-01-19 | 深圳奥比中光科技有限公司 | Matting method and system based on depth information |
| CN112884776A (en) * | 2021-01-22 | 2021-06-01 | 浙江大学 | Deep learning cutout method based on synthesis data set augmentation |
| CN113487630A (en) * | 2021-07-14 | 2021-10-08 | 辽宁向日葵教育科技有限公司 | Image matting method based on material analysis technology |
| CN116912284A (en) * | 2023-09-15 | 2023-10-20 | 电子科技大学中山学院 | Matting method, matting device, electronic equipment and computer readable storage medium |
| CN117351118A (en) * | 2023-12-04 | 2024-01-05 | 江西师范大学 | Lightweight fixed background matting method and system combined with depth information |
| US12041374B2 (en) | 2020-09-09 | 2024-07-16 | Lemon Inc. | Segmentation-based video capturing method, apparatus, device and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102903110A (en) * | 2012-09-29 | 2013-01-30 | 宁波大学 | Segmentation method for image with deep image information |
| CN103559719A (en) * | 2013-11-20 | 2014-02-05 | 电子科技大学 | Interactive graph cutting method |
| CN104036517A (en) * | 2014-07-01 | 2014-09-10 | 成都品果科技有限公司 | Image matting method based on gradient sampling |
| CN105590309A (en) * | 2014-10-23 | 2016-05-18 | 株式会社理光 | Method and device for segmenting foreground image |
| CN105590312A (en) * | 2014-11-12 | 2016-05-18 | 株式会社理光 | Foreground image segmentation method and apparatus |
-
2017
- 2017-03-20 CN CN201710164954.XA patent/CN106952276A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102903110A (en) * | 2012-09-29 | 2013-01-30 | 宁波大学 | Segmentation method for image with deep image information |
| CN103559719A (en) * | 2013-11-20 | 2014-02-05 | 电子科技大学 | Interactive graph cutting method |
| CN104036517A (en) * | 2014-07-01 | 2014-09-10 | 成都品果科技有限公司 | Image matting method based on gradient sampling |
| CN105590309A (en) * | 2014-10-23 | 2016-05-18 | 株式会社理光 | Method and device for segmenting foreground image |
| CN105590312A (en) * | 2014-11-12 | 2016-05-18 | 株式会社理光 | Foreground image segmentation method and apparatus |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171716B (en) * | 2017-12-25 | 2021-11-26 | 北京奇虎科技有限公司 | Video character decorating method and device based on self-adaptive tracking frame segmentation |
| CN108171716A (en) * | 2017-12-25 | 2018-06-15 | 北京奇虎科技有限公司 | Video personage based on the segmentation of adaptive tracing frame dresss up method and device |
| WO2020062899A1 (en) * | 2018-09-26 | 2020-04-02 | 惠州学院 | Method for obtaining transparency masks by means of foreground and background pixel pairs and grayscale information |
| CN111899266A (en) * | 2020-07-17 | 2020-11-06 | 深圳奥比中光科技有限公司 | Matting method and system based on RGBD camera |
| US12041374B2 (en) | 2020-09-09 | 2024-07-16 | Lemon Inc. | Segmentation-based video capturing method, apparatus, device and storage medium |
| CN112241960A (en) * | 2020-10-01 | 2021-01-19 | 深圳奥比中光科技有限公司 | Matting method and system based on depth information |
| CN112241960B (en) * | 2020-10-01 | 2024-05-31 | 奥比中光科技集团股份有限公司 | Depth information-based matting method and system |
| CN112884776B (en) * | 2021-01-22 | 2022-05-31 | 浙江大学 | Deep learning matting method based on synthesis data set augmentation |
| CN112884776A (en) * | 2021-01-22 | 2021-06-01 | 浙江大学 | Deep learning cutout method based on synthesis data set augmentation |
| CN113487630A (en) * | 2021-07-14 | 2021-10-08 | 辽宁向日葵教育科技有限公司 | Image matting method based on material analysis technology |
| CN116912284A (en) * | 2023-09-15 | 2023-10-20 | 电子科技大学中山学院 | Matting method, matting device, electronic equipment and computer readable storage medium |
| CN117351118A (en) * | 2023-12-04 | 2024-01-05 | 江西师范大学 | Lightweight fixed background matting method and system combined with depth information |
| CN117351118B (en) * | 2023-12-04 | 2024-02-23 | 江西师范大学 | Lightweight fixed background matting method and system combined with depth information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106952276A (en) | A kind of image matting method and device | |
| EP2706507B1 (en) | Method and apparatus for generating morphing animation | |
| CN102999901B (en) | Based on the processing method after the Online Video segmentation of depth transducer and system | |
| CN107146201A (en) | A kind of image split-joint method based on improvement image co-registration | |
| CN103544685B (en) | A kind of image composition beautification method adjusted based on main body and system | |
| CN104616286B (en) | Quick semi-automatic multi views depth restorative procedure | |
| CN106910242A (en) | The method and system of indoor full scene three-dimensional reconstruction are carried out based on depth camera | |
| CN103839223A (en) | Image processing method and image processing device | |
| CN104732506A (en) | Character picture color style converting method based on face semantic analysis | |
| EP1899897A2 (en) | Video object cut and paste | |
| CN102034247B (en) | Motion capture method for binocular vision image based on background modeling | |
| CN104954780A (en) | DIBR (depth image-based rendering) virtual image restoration method applicable to high-definition 2D/3D (two-dimensional/three-dimensional) conversion | |
| CN105678724A (en) | Background replacing method and apparatus for images | |
| CN108463823A (en) | A kind of method for reconstructing, device and the terminal of user's Hair model | |
| CN104036481B (en) | Multi-focus image fusion method based on depth information extraction | |
| CN101459843B (en) | Method for precisely extracting broken content region in video sequence | |
| CN113658316B (en) | Rendering method and device of three-dimensional model, storage medium and computer equipment | |
| CN103413323B (en) | Based on the object tracking methods of component-level apparent model | |
| CN103578085A (en) | Image empty region inpainting method based on variable blocks | |
| CN116583878A (en) | Method and system for personalizing 3D head model deformation | |
| CN107240073A (en) | A kind of 3 d video images restorative procedure merged based on gradient with clustering | |
| CN116997933A (en) | Method and system for constructing facial position map | |
| CN104952102A (en) | Deferred shading-orientated uniform anti-aliasing method | |
| CN106023276A (en) | Pencil drawing making method and pencil drawing making device based on image processing | |
| Li et al. | Fast retargeting with adaptive grid optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170714 |