CN102707864A - Object segmentation method and system based on mixed marks - Google Patents
Object segmentation method and system based on mixed marks Download PDFInfo
- Publication number
- CN102707864A CN102707864A CN2011100852949A CN201110085294A CN102707864A CN 102707864 A CN102707864 A CN 102707864A CN 2011100852949 A CN2011100852949 A CN 2011100852949A CN 201110085294 A CN201110085294 A CN 201110085294A CN 102707864 A CN102707864 A CN 102707864A
- Authority
- CN
- China
- Prior art keywords
- pixel
- background
- stroke
- foreground
- uncertain
- 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)
Abstract
The invention discloses an object segmentation method and an object segmentation system based on mixed marks. The method comprises the following steps of: receiving marking strokes of pixels in an image which comprises a target object, wherein the marking strokes comprise foreground strokes for the target object, background strokes for a background and undetermined strokes which are undetermined for a foreground or the background; on the basis of the foreground strokes and the background strokes, establishing a foreground model and a background model to classify the pixels in the image, which are not marked by the undetermined strokes, into foreground pixels or background pixels; and according to the foreground pixels and the background pixels, which are classified, calculating probability values with which one or more pixels corresponding to the undetermined strokes belong to the foreground strokes, determining the pixels of which the probability values are smaller than lower limit as the background pixels, and determining the pixels of which the probability values are greater than upper limit as the foreground pixels. An object in the image is interactively segmented on the basis of various marks, so complicated boundary situation can be handled, and a high-accuracy segmentation result is obtained.
Description
Technical field
The present invention relates to image processing techniques, be specifically related to a kind of Object Segmentation method and system based on mixed mark.
Background technology
The purpose of image segmentation is that interesting areas mark in the image is come out.According to the result of mark, people just can carry out Flame Image Process to specific area-of-interest.This demand of carrying out Flame Image Process to the specific region is common demand during Flame Image Process is used.But that is that all right is ripe for the method for can be efficiently the area-of-interest mark being come out.If the zone itself is regular, then utilize interactive tools (like mouse) to be easy to finish the work.But if area-of-interest is irregular, then interested area division is then relatively more difficult.Obtain area-of-interest, need with the boundary pixel of all area-of-interests one by one mark come out.Such work is very trifling and labor intensive.So people hope to extract area-of-interest with automatic or automanual mode.This interested target area is known as destination object, is called for short object or prospect.
Automanual object tag method is exactly that focus object is cut apart.The user need tell the information of some objects of calculation procedure, and computer program just can utilize these information automatically Object Segmentation to be come out.If the result does not reach final goal, the user can increase some information again.Computer program calculates once more.Through this mutual-iterative manner cut apart, the segmentation result that just possibly approach step by step.As a rule, the user has two kinds of methods that object information is provided.First kind is the border of tagged object, in case complete object bounds be labeled out, even if the Object Segmentation task is accomplished; Second kind then is the part sub pixel of tagged object, the information of utilizing sub pixel to provide, and computer program is realized automatic Object Segmentation.For example, in the second way, the user can come some subject pixels and background pixels of confirming of mark with mouse, lets calculation procedure carry out destination object then and cuts apart.
The target of focus object cutting techniques is exactly to use few accurate segmentation result of mutual acquisition as far as possible.Ideal situation, the user carries out mark one by one to object pixel.Yet this is unpractical.Because object pixels is very many, fully mark is very trifling and labor intensive.On the other hand, when object bounds was fairly simple, the focus object cutting techniques can be obtained segmentation precision preferably.If for complex situations, segmentation result is then often accurate inadequately.Such as, the object as the hair, the border is complicated unusually, and the user is difficult to the border of target-marking object.So when these boundary pixels segmentation errors or label information occur when insufficient, user's mark often obtains relatively poor result because of insufficient mark.Therefore, the border of present this complicacy of partitioning algorithm intractable.
Patent documentation 1 (US20080136820A1) has proposed a kind of Object Segmentation method based on mark, but its core technology still is to utilize prospect and two kinds of marks of background to obtain segmentation result.For border complicated such as hair, the method that this patent documentation 1 proposes still has difficulties, and segmentation precision is not high.
Summary of the invention
The objective of the invention is to propose a kind of Object Segmentation method and system based on mixed mark.
According to an aspect of the present invention; A kind of Object Segmentation method based on mixed mark has been proposed; Comprise step: receive mark stroke to the pixel in the image that comprises destination object, said mark stroke comprise prospect stroke to destination object, to the background stroke of background and uncertain be the uncertain stroke of prospect or background; Set up foreground model and background model based on said prospect stroke and background stroke, so that to not being categorized as foreground pixel or background pixel in the image by the pixel of uncertain stroke marking; Foreground pixel and background pixel according to classification; Calculate the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke; And confirm as background pixel to probable value less than the pixel of lower limit, and confirm as foreground pixel to probable value greater than the pixel of the upper limit.
According to embodiments of the invention, said method also comprises step: probable value is confirmed as uncertain pixel greater than said lower limit less than the pixel of the said upper limit.
According to embodiments of the invention, described method also comprises step: the pixel less than the said upper limit is divided into foreground pixel or background pixel greater than said lower limit with probable value with predetermined threshold value.
According to embodiments of the invention, the step that the corresponding one or more pixels of said calculating and uncertain stroke belong to the probable value of foreground pixel is based on the button nomography and carries out.
According to embodiments of the invention, described foreground model and background model are based on mixed Gauss model and set up.
According to a further aspect in the invention; A kind of Object Segmentation system based on mixed mark has been proposed; Comprise: receiving unit; Reception is to the mark stroke of the pixel in the image that comprises destination object, said mark stroke comprise prospect stroke to destination object, to the background stroke of background and uncertain be the uncertain stroke of prospect or background; Partitioning portion; Set up foreground model and background model based on said prospect stroke and background stroke; So that to not being categorized as foreground pixel or background pixel in the image,, calculate the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke according to the foreground pixel and the background pixel of classification by the pixel of uncertain stroke marking; And confirm as background pixel to probable value less than the pixel of lower limit, and confirm as foreground pixel to probable value greater than the pixel of the upper limit.
According to embodiments of the invention, described system also comprises: storage area, the mark stroke that storage acceptance division branch receives; Analysis part, analyzing the mark stroke that is received is prospect stroke, background stroke or uncertain stroke, and offers said partitioning portion.
According to embodiments of the invention, said partitioning portion is confirmed as uncertain pixel greater than said lower limit less than the pixel of the said upper limit with probable value.
According to embodiments of the invention, the pixel less than the said upper limit is divided into foreground pixel or background pixel to said partitioning portion greater than said lower limit with probable value with predetermined threshold value.
According to embodiments of the invention, said partitioning portion calculates the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke based on the button nomography.
According to embodiments of the invention, described foreground model and background model are based on mixed Gauss model and set up.
Utilize the method and system of the embodiment of the invention, come the object in the image is carried out Interactive Segmentation based on multiple mark, border condition that can dealing with complicated obtains high-precision segmentation result.
Description of drawings
From the detailed description below in conjunction with accompanying drawing, above-mentioned feature and advantage of the present invention will be more obvious, wherein:
Fig. 1 has described the whole block scheme based on the Object Segmentation system of mixed mark according to the embodiment of the invention; And
Fig. 2 shows the process flow diagram based on the Object Segmentation method of mixed mark according to the embodiment of the invention.
Embodiment
Below, specify preferred implementation of the present invention with reference to accompanying drawing.In the accompanying drawings, though be shown in the different drawings, identical Reference numeral is used to represent identical or similar assembly.For clear and simple and clear, the known function and the detailed description of structure that are included in here will be omitted, otherwise they will make theme of the present invention unclear.
Fig. 1 has described the whole block scheme based on the Object Segmentation system of mixed mark according to the embodiment of the invention.As shown in Figure 1; System according to the embodiment of the invention comprises: cutting unit 110; It is according to the mark result of user to input picture; From the stroke of dissimilar marks (the for example stroke of prospect, background and uncertain three types mark), set up corresponding prospect and background model, and utilize the stroke of these marks to carry out automated graphics to cut apart; Display unit 120, the result that it will carry out image segmentation shows.Then, the user observes segmentation result, and judges according to cutting apart target whether current segmentation result reaches requirement, if not to requiring, the user utilizes the input media such as mouse or writing pencil to carry out on the basis of segmentation result that next step is mutual.According to embodiments of the invention; Cutting unit 110 is set up foreground model and background model based on prospect stroke and background stroke; So that to not being categorized as foreground pixel or background pixel in the image by the pixel of uncertain stroke marking; Foreground pixel and background pixel according to classification; Calculate the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke, and confirm as background pixel to probable value less than the pixel of lower limit, and confirm as foreground pixel to probable value greater than the pixel of the upper limit.
According to embodiments of the invention, on a last step segmentation result, the user carries out mark to segmentation result, and the collection of pixels of mark becomes a mark stroke each time.This system comprises receiving element 150, and it receives the mark stroke of user through the input media input.Then, all once mutual strokes of storage unit 140 storage users.These strokes are corresponding different markers type respectively, like prospect, background, uncertain border.The content that storage unit 140 is stored to stroke is the location sets of this stroke institute covered pixels, a type in three types on the corresponding prospect of each stroke, background and the uncertain border.
According to embodiments of the invention, this system also comprises analytic unit 130, and it confirms that according to the once mutual some strokes of user dissimilar strokes is respectively applied for different processing procedures, uses for cutting unit 110.
The operating process of said system is described with the mode of object lesson below in conjunction with accompanying drawing 2.Fig. 2 shows the process flow diagram based on the Object Segmentation method of mixed mark according to the embodiment of the invention.
As shown in Figure 2, image is input in this system, and the digital picture that wherein contains destination object is the input of the embodiment of the invention.Destination object can be the image-region of Any user definition.This destination object is a predefined just before cutting apart, and does not have ambiguity, such as a target that its meaning is arranged: people, animal, meadow etc.It also can be the target combination of these its meaning.As shown in Figure 2, the image of input is the image that a width of cloth contains a child back.This width of cloth image contains a plurality of objects, as: child, football, meadow, trees, some footballer etc.Here the destination object of definition is child's a hair, and other object is all regarded background as.Relative background, destination object is a prospect.
At step S11, the image of 110 pairs of inputs of cutting unit is cut apart, so that extract destination object, i.e. and child's hair.
At step S12, display unit 120 shows segmentation result.In general, the result of Object Segmentation can represent with bianry image.A value representation background pixel, a value representation foreground object.Owing to before segmentation result will be presented on user plane, segmentation result is assessed to the user.Therefore, often with the segmentation result image overlay to original image.Generally can the foreground object border be shown with special color is outstanding on original image; Perhaps the object after will cutting apart on the original image shows with special color.
Then, at step S13, the assessment segmentation result.User's assessment is a kind of subjective behavior, is exactly that the user judges according to the segmentation result that appears whether the segmentation result of current acquisition reaches requirement, and tells calculation procedure with judged result.If reach, just tell that calculation procedure finishes to cut apart; Otherwise, further handle.
The purpose of user interactions is for partitioning algorithm new prospect, background and uncertain mark to be provided.According to embodiments of the invention, providing also of this mark can be provided by alternate manner.For example calculate definite prospect, background and uncertain mark automatically.To the situation that has the foreground object model on the statistical significance; Utilize this model that prospect and background pixel in the last time segmentation result are classified; The pixel of high reliability is divided into prospect; Hang down the background of confirming as of reliability, center section and locus get pixel at the prospect background intersection and confirm as uncertain pixel.
Judge whether to finish at step S14.The performance of cutting apart can come according to the definition of target to differentiate with subjective and/or objective mode.If finish, then show segmentation result, otherwise at step S16, the user through the input media input to last time segmentation result in the mark stroke of pixel.According to embodiments of the invention, the mark stroke comprises prospect stroke, background stroke and uncertain border stroke.
User interactions is the process that the user utilizes interactive tool that segmentation result is made amendment.Here, the interactive tool that refers to of the embodiment of the invention is a computer mouse.The user utilizes mouse in the computer screen marked.The mark track of mouse is called a stroke each time.Each stroke has a type, like prospect, background, uncertain mark.Each stroke is the set of a location of pixels, and it is prospect, background, uncertain type mark that the pixel that shows these positions contains.The user comes mark prospect and background according to the definition to destination object.To being not easy to make things convenient for the pixel of mark, then be labeled as uncertain type.General uncertain pixel produces main border at object.When object bounds was complicated, foreground pixel often needed a lot of mutual could completion.For instance,, a lot of tiny hairs are arranged, cause some prospect borders to have only several sometimes even a pixel wide like hair border among Fig. 2.Therefore, the user is difficult to accomplish in high quality interworking.Uncertain pixel mark then is the instrument that addresses this problem, and the user only need be directly labeled as uncertain type with these pixels that is not easily distinguishable, and then can be handled automatically by calculation procedure.
According to embodiments of the invention, the user carries out once mutual, can be made up of some strokes.Each stroke all has a unique kind mark: three kinds of prospects, background, uncertain mark.The pixel of different markers type is deposited in three set respectively and is represented.The locations of pixels set that user's mark is crossed is stored in these three set respectively.Then, flow process is got back to step S11.
At step S11, utilize the mark stroke that receives to carry out once more Object Segmentation.
Object Segmentation is meant carrying out the process of binary classification with the corresponding image pixel of object.Classification obtains the result and judges that exactly which pixel belongs to destination object (prospect), and which pixel belongs to background object.According to the embodiment of the invention, partitioning algorithm used herein is according to specific prospect and background information, sets up corresponding prospect, background model.Then, on this basis, utilize foreground model and background model,, obtain the Object Segmentation result through assorting process to pixel in the image.
In step S11, foreground model is that the foreground pixel that is used for that utilizes prior imformation to set up is differentiated mathematical model.The numerical characteristics of the foreground pixel of this model description.Here, the foreground model of this method use for example is list of references 1 (Wang, D.; Shan, S.G., Zeng; W., Zhang, H.M.; Chen, X.L.:A novel two-tier Bayesian based method for hairsegmentation.International Conference on Image Processing, (2009) 2401-2404) in mixed Gauss model.
In step S11, background model is that the background pixel that is used for that utilizes prior imformation to set up is differentiated mathematical model.The numerical characteristics of the background pixel of this model description.Here, the background model used of this method for example is the mixed Gauss model in the list of references 1.
The figure cutting method of in step S11, using is according to prospect and background model, the method that pixel is classified.This method utilization figure cuts theory, and the spatial relationship of the prospect of composite pixel and background model and pixel utilizes the minimax flow algorithm to realize the classification to the prospect background pixel, referring to list of references 1.
In step S11, use list of references 2 (Eduardo S.L.Gastal and ManuelM.Oliveira, Shared Sampling for Real-Time Alpha Matting; ComputerGraphics Forum; Volume 29 (2010), and Number 2, Proceedings ofEurographics 2010; Pp.575-584) Matting that describes in (button figure) algorithm, it is the method for the Alpha value of calculating pixel.This method is utilized current prospect and the background pixel of having confirmed, calculates the Alpha value that other does not confirm the pixel of prospect background classification.
Do not confirm in the embodiment of the invention that the pixel of prospect background classification comes from the pixel of the uncertain mark shown in the step S17 among Fig. 2.Through the Matting algorithm, confirm the Alpha value of uncertain pixel.The Alpha value is the real number between 0 to 1, this value representation this pixel belong to the probability of prospect.The prospect of 1 expression 100%, the background of 0 expression 100%.Therefore, according to the Alpha value, the Alpha value is less than the background that is judged as of the pixel of a less threshold value (lower limit); The Alpha value is greater than the prospect that is judged as of the pixel of big threshold value (upper limit).
For example, the example of a partitioning algorithm is following:
(1) starting stage, owing to have no user interactions, partitioning algorithm can adopt any among two kinds of processing policies:
1.1) adopt other automatic object dividing method to obtain preliminary segmentation result, like list of references 1;
1.2) directly all image pixels are categorized as background pixel.
(2) segmentation result is shown to the user, this result is passed judgment on by the user.
(3) if the user is satisfied to segmentation result, then algorithm finishes.
(4), then require to carry out mutual if the user is dissatisfied to segmentation result.Utilize interactive tool (like mouse), in the enterprising row labels of image.If the user thinks that some pixel is a foreground pixel, then comes out in the screen marked with interactive tool.In like manner, mark background pixel.If wrong cutting apart in last once segmentation result, occur, miss the pixel that is partitioned into background like prospect, the background mistake is divided into foreground pixel, and user's these wrong pixels of mark again is correct mark.On the other hand, for some users think can not meticulous mark pixel, the user can carry out mark with uncertain mark (embodiment of the invention specific markers).Mark is called a stroke each time, and each stroke has a type: prospect, background, uncertain type.The location of pixels set that stroke is user interaction tool mobile track on screen.
(5) according to the stroke of user's mark, the foreground pixel of these user's appointments, background pixel are used for making up prospect and background model after the renewal together with last segmentation result.Wherein, the prospect of user's mark and background pixel distribute bigger weight when new prospect and background model structure.According to the prospect background model that makes up, all pixels that are not marked as non-definite type in the segmentation procedure utilization figure cutting mode classified image are prospect or background pixel.For the uncertain pixel of user's appointment, then utilize the Matting algorithm to confirm its Alpha value.The Alpha value is the real number between 0 to 1, this value representation this pixel belong to the probability of prospect.The prospect of 1 expression 100%, the background of 0 expression 100%.Therefore, according to the Alpha value, the Alpha value is less than the background that is judged as of the pixel of a less threshold value; The Alpha value is greater than the prospect that is judged as of the pixel of big threshold value.The corresponding pixel of other value is designated as uncertain pixel, and these uncertain pixels do not participate in later front and back context update and calculate.
The result that (6) (5) steps obtained is displayed to the user, if the user is unsatisfied with current segmentation result then comes back to step (4), otherwise carries out next step.
(7) according to one near 0.5 threshold value, all uncertain pixels are re-classified as prospect and background.
According to embodiments of the invention, the judgement of the end of algorithm also can adopt following mode to carry out: (a) interaction times reaches certain interaction times algorithm and finishes; (b) segmentation result evaluation tolerance like the energy function in the Graph-cut algorithm, is measured ability function less than a threshold value when this, then representes to cut apart to reach requirement; (c) subjective assessment is carried out subjectivity by the user to segmentation result and is passed judgment on, and input instruction finishes whole process.
According to embodiments of the invention; Under the situation that has the foreground object model on the statistical significance; Then can this model just can classify to prospect and background pixel in the last time segmentation result; The pixel of high reliability is divided into prospect, hangs down the background of confirming as of reliability, center section and locus get pixel at the prospect background intersection and confirm as uncertain pixel.
Scheme according to an embodiment of the invention; Utilize three kinds of marks to carry out Object Segmentation; Particularly introduce uncertain mark and increase the performance of cutting apart; Avoided because the prospect background boundary pixel to the influence of cutting procedure, makes the processing for the border of complicacy become easily, and improved segmentation precision.
Top description only is used to realize embodiment of the present invention; It should be appreciated by those skilled in the art; In any modification that does not depart from the scope of the present invention or local replacement; All should belong to scope thereof of the present invention, therefore, protection scope of the present invention should be as the criterion with the protection domain of claims.
Claims (11)
1. Object Segmentation method based on mixed mark comprises step:
Reception is to the mark stroke of the pixel in the image that comprises destination object, said mark stroke comprise prospect stroke to destination object, to the background stroke of background and uncertain be the uncertain stroke of prospect or background;
Set up foreground model and background model based on said prospect stroke and background stroke, so that to not being categorized as foreground pixel or background pixel in the image by the pixel of uncertain stroke marking;
Foreground pixel and background pixel according to classification; Calculate the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke; And confirm as background pixel to probable value less than the pixel of lower limit, and confirm as foreground pixel to probable value greater than the pixel of the upper limit.
2. the method for claim 1 also comprises step:
Probable value is confirmed as uncertain pixel greater than said lower limit less than the pixel of the said upper limit.
3. method as claimed in claim 2 also comprises step:
Pixel less than the said upper limit is divided into foreground pixel or background pixel greater than said lower limit with probable value with predetermined threshold value.
4. the method for claim 1, the step that the corresponding one or more pixels of wherein said calculating and uncertain stroke belong to the probable value of foreground pixel are based on the button nomography and carry out.
5. the method for claim 1, wherein said foreground model and background model are based on mixed Gauss model and set up.
6. Object Segmentation system based on mixed mark comprises:
Receiving unit receives the mark stroke to the pixel in the image that comprises destination object, said mark stroke comprise prospect stroke to destination object, to the background stroke of background and uncertain be the uncertain stroke of prospect or background;
Partitioning portion; Set up foreground model and background model based on said prospect stroke and background stroke; So that to not being categorized as foreground pixel or background pixel in the image,, calculate the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke according to the foreground pixel and the background pixel of classification by the pixel of uncertain stroke marking; And confirm as background pixel to probable value less than the pixel of lower limit, and confirm as foreground pixel to probable value greater than the pixel of the upper limit.
7. system as claimed in claim 6 also comprises:
Storage area, the mark stroke that storage acceptance division branch receives;
Analysis part, analyzing the mark stroke that is received is prospect stroke, background stroke or uncertain stroke, and offers said partitioning portion.
8. system as claimed in claim 6, wherein said partitioning portion is confirmed as uncertain pixel greater than said lower limit less than the pixel of the said upper limit with probable value.
9. system as claimed in claim 8, the pixel less than the said upper limit is divided into foreground pixel or background pixel to wherein said partitioning portion greater than said lower limit with probable value with predetermined threshold value.
10. system as claimed in claim 6, wherein said partitioning portion calculates the probable value that belongs to foreground pixel with the corresponding one or more pixels of uncertain stroke based on the button nomography.
11. system as claimed in claim 6, wherein said foreground model and background model are based on mixed Gauss model and set up.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011100852949A CN102707864A (en) | 2011-03-28 | 2011-03-28 | Object segmentation method and system based on mixed marks |
| JP2011244206A JP5226119B2 (en) | 2011-03-28 | 2011-11-08 | Method and system for dividing objects based on hybrid labels |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011100852949A CN102707864A (en) | 2011-03-28 | 2011-03-28 | Object segmentation method and system based on mixed marks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102707864A true CN102707864A (en) | 2012-10-03 |
Family
ID=46900721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011100852949A Pending CN102707864A (en) | 2011-03-28 | 2011-03-28 | Object segmentation method and system based on mixed marks |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP5226119B2 (en) |
| CN (1) | CN102707864A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103473780A (en) * | 2013-09-22 | 2013-12-25 | 广州市幸福网络技术有限公司 | Portrait background cutout method |
| CN103914822A (en) * | 2012-12-31 | 2014-07-09 | 清华大学 | Interactive video foreground object extraction method based on super pixel segmentation |
| CN104899860A (en) * | 2014-03-07 | 2015-09-09 | 宏达国际电子股份有限公司 | Image segmentation device and image segmentation method |
| CN105139415A (en) * | 2015-09-29 | 2015-12-09 | 小米科技有限责任公司 | Foreground and background segmentation method and apparatus of image, and terminal |
| CN107230211A (en) * | 2017-05-05 | 2017-10-03 | 上海联影医疗科技有限公司 | A kind of image partition method and system |
| CN111105440A (en) * | 2019-11-29 | 2020-05-05 | 北京金山云网络技术有限公司 | Method, device and equipment for tracking target object in video and storage medium |
| CN111784712A (en) * | 2020-07-17 | 2020-10-16 | 北京字节跳动网络技术有限公司 | Image processing method, device, equipment and computer readable medium |
| CN113506302A (en) * | 2021-07-27 | 2021-10-15 | 四川九洲电器集团有限责任公司 | Interactive object updating method, device and processing system |
| CN113628221A (en) * | 2021-08-03 | 2021-11-09 | Oppo广东移动通信有限公司 | Image processing method, image segmentation model training method and related device |
| CN116228745A (en) * | 2023-04-24 | 2023-06-06 | 徐州医科大学 | Deep convolution network construction method for classifying benign and malignant pulmonary nodules |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013196308A (en) * | 2012-03-19 | 2013-09-30 | Ricoh Co Ltd | Image processor, image processing method, program and recording medium |
| KR101546635B1 (en) | 2014-11-28 | 2015-08-21 | 연세대학교 산학협력단 | Method and Device for Editing Image |
| WO2019176087A1 (en) | 2018-03-16 | 2019-09-19 | 三菱電機株式会社 | Learning device and learning method |
| US11514695B2 (en) * | 2020-12-10 | 2022-11-29 | Microsoft Technology Licensing, Llc | Parsing an ink document using object-level and stroke-level processing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101512598A (en) * | 2006-07-21 | 2009-08-19 | 奥多比公司 | Live coherent image selection |
| KR20090111939A (en) * | 2008-04-23 | 2009-10-28 | 주식회사 케이티 | Method and apparatus for separating foreground and background from image, Method and apparatus for substituting separated background |
| CN101567088A (en) * | 2008-04-22 | 2009-10-28 | 华为技术有限公司 | Method and device for detecting moving object |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060029275A1 (en) * | 2004-08-06 | 2006-02-09 | Microsoft Corporation | Systems and methods for image data separation |
| JP5045619B2 (en) * | 2008-09-08 | 2012-10-10 | ソニー株式会社 | Image processing apparatus and method, and program |
| EP2199981A1 (en) * | 2008-12-09 | 2010-06-23 | Koninklijke Philips Electronics N.V. | Image segmentation |
-
2011
- 2011-03-28 CN CN2011100852949A patent/CN102707864A/en active Pending
- 2011-11-08 JP JP2011244206A patent/JP5226119B2/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101512598A (en) * | 2006-07-21 | 2009-08-19 | 奥多比公司 | Live coherent image selection |
| CN101567088A (en) * | 2008-04-22 | 2009-10-28 | 华为技术有限公司 | Method and device for detecting moving object |
| KR20090111939A (en) * | 2008-04-23 | 2009-10-28 | 주식회사 케이티 | Method and apparatus for separating foreground and background from image, Method and apparatus for substituting separated background |
Non-Patent Citations (1)
| Title |
|---|
| 彭莎等: "基于数字抠图的ROI提取方法改进及其在CT图像中的应用", 《中国生物医学工程学会成立30周年纪念大会暨2010中国生物医学工程学会学术大会》, 2 December 2010 (2010-12-02), pages 82 - 84 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914822A (en) * | 2012-12-31 | 2014-07-09 | 清华大学 | Interactive video foreground object extraction method based on super pixel segmentation |
| CN103473780A (en) * | 2013-09-22 | 2013-12-25 | 广州市幸福网络技术有限公司 | Portrait background cutout method |
| CN103473780B (en) * | 2013-09-22 | 2016-05-25 | 广州市幸福网络技术有限公司 | The method of portrait background figure a kind of |
| US10073543B2 (en) | 2014-03-07 | 2018-09-11 | Htc Corporation | Image segmentation device and image segmentation method |
| CN104899860A (en) * | 2014-03-07 | 2015-09-09 | 宏达国际电子股份有限公司 | Image segmentation device and image segmentation method |
| CN105139415A (en) * | 2015-09-29 | 2015-12-09 | 小米科技有限责任公司 | Foreground and background segmentation method and apparatus of image, and terminal |
| CN107230211A (en) * | 2017-05-05 | 2017-10-03 | 上海联影医疗科技有限公司 | A kind of image partition method and system |
| CN107230211B (en) * | 2017-05-05 | 2021-07-16 | 上海联影医疗科技股份有限公司 | Image segmentation method and system |
| CN111105440A (en) * | 2019-11-29 | 2020-05-05 | 北京金山云网络技术有限公司 | Method, device and equipment for tracking target object in video and storage medium |
| CN111105440B (en) * | 2019-11-29 | 2023-07-07 | 北京金山云网络技术有限公司 | Tracking method, device, equipment and storage medium for target object in video |
| CN111784712A (en) * | 2020-07-17 | 2020-10-16 | 北京字节跳动网络技术有限公司 | Image processing method, device, equipment and computer readable medium |
| CN113506302A (en) * | 2021-07-27 | 2021-10-15 | 四川九洲电器集团有限责任公司 | Interactive object updating method, device and processing system |
| CN113506302B (en) * | 2021-07-27 | 2023-12-12 | 四川九洲电器集团有限责任公司 | Interactive object updating method, device and processing system |
| CN113628221A (en) * | 2021-08-03 | 2021-11-09 | Oppo广东移动通信有限公司 | Image processing method, image segmentation model training method and related device |
| CN116228745A (en) * | 2023-04-24 | 2023-06-06 | 徐州医科大学 | Deep convolution network construction method for classifying benign and malignant pulmonary nodules |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2012208913A (en) | 2012-10-25 |
| JP5226119B2 (en) | 2013-07-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102707864A (en) | Object segmentation method and system based on mixed marks | |
| KR101719088B1 (en) | Method for partitioning area, and inspection device | |
| CN111695488A (en) | Interest plane identification method, device, equipment and storage medium | |
| CN114219786B (en) | Chromosome karyotype analysis method and system based on deep learning | |
| KR101700648B1 (en) | System for displaying carbon emission on the region planing model | |
| Sriyakul et al. | Total quality management and logistic performance: moderating role of reserve supply chain in pharmaceutical industry of Indonesia | |
| CN109636090B (en) | Station allocation method, device, equipment and storage medium | |
| CN110851646A (en) | Working efficiency statistical method for intelligent park | |
| CN110992124A (en) | House resource recommendation method and system | |
| CN104680188A (en) | Method for constructing human body posture reference image library | |
| CN107437367B (en) | Method and device for selecting labels | |
| CN105976188A (en) | Multi-channel client information processing system and processing method | |
| CN115393560A (en) | BIM model drawing marking method, device, equipment and readable storage medium | |
| CN104572999A (en) | Heterogeneous data based detecting method and device for fault data | |
| CN113301376A (en) | Live broadcast interaction mode and system based on virtual reality technology | |
| CN108446603A (en) | A kind of headline detection method and device | |
| JP5780791B2 (en) | Cell tracking method | |
| CN104142952A (en) | Method and device for showing reports | |
| CN118366162B (en) | Image segmentation method and system based on deep learning | |
| CN113516082A (en) | Detection method and device of safety helmet, computer equipment and storage medium | |
| CN102004795A (en) | Hand language searching method | |
| CN114969135B (en) | Personalized travel route recommendation method, device and medium | |
| CN113553980B (en) | Method, system and device for generating trademark graphic element codes of pictures | |
| CN107016044B (en) | Data visualization processing method and device | |
| CN108132940A (en) | A kind of application data extracting method and device |
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: 20121003 |