CN101833781A - Method for automatically completing hidden parts of similar objects based on geometric information - Google Patents
Method for automatically completing hidden parts of similar objects based on geometric information Download PDFInfo
- Publication number
- CN101833781A CN101833781A CN 201010158440 CN201010158440A CN101833781A CN 101833781 A CN101833781 A CN 101833781A CN 201010158440 CN201010158440 CN 201010158440 CN 201010158440 A CN201010158440 A CN 201010158440A CN 101833781 A CN101833781 A CN 101833781A
- Authority
- CN
- China
- Prior art keywords
- completion
- source object
- pixel value
- blocked
- source
- 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
Images
Abstract
The invention discloses a method for automatically completing hidden parts of similar objects based on geometric information, comprising the following steps: S1, dividing an object to be completed into two parts (part a: a shielded part; and part b: an unshielded part) by using the contour geometric information of the object to be completed in an image, the contour geometric information of a source object in the image and the established one-to-one correspondence of the object to be completed and all pixels on the contour of the source object; and S2, as for the part a, completing the part a by using the pixel value of each point in the deformed source object, wherein the shape of the deformed source object and the shape of the object to be completed are completely same, and the object to be completed and the source object are two different entities of the same object in one image. In the invention, the completed similar object has double completeness both in geometric information and in semantic information, and the unshielded part can be completed to achieve the effect of seamless splicing.
Description
Technical field
The invention belongs to technical field of image processing, particularly a kind of analogical object method for automatically completing hidden parts based on geological information.
Background technology
Along with popularizing and the continuous development of multimedia technology of multimedia equipment, equipment such as digital camera are widely used, and a large amount of photo processing demands of Chan Shenging therefrom makes picture editting's technology become one of focus that present computer section studies.At present, image processing software on the market as Adobe Photoshop etc., can only provide some elementary Pixel-level operations, the user need complicated alternately, stronger artistic accomplishment and skilled operative skill, could realize that some meet picture editting's effect of user's final demand.Therefore, the intellectuality of picture editting's technology, robotization become the main direction of nearest technical development.Wherein, picture editting's intelligence is propagated and is had a wide range of applications and user's request, and editor is propagated the image manipulation that is applied to object level, is the work that has more challenge and frontier nature.The editor of object level propagates, needing at first has the image of a plurality of repetition/analogical objects to carry out the extraction of repetition/analogical object to one, the more representational technology of this respect has 2.1D method (being a kind of method that is used to extract similar texture in computer vision field)---and utilize the form of tree to express image, to mate the method for similar node---and some are based on the similar pel detection method of texture.
Though it is said method can be found out similar texture part preferably, powerless for the situation of geological information disappearances such as object is blocked.So the image manipulation of object level must keep the geological information of analogical object, and under the situation of the geological informations such as shape of known object, the part that is blocked then needs further processing, thus the integrality of analogical object in the picture that guarantees to extract.
Summary of the invention
(1) technical matters that will solve
The technical problem to be solved in the present invention is how in known image under the situation of the geological information of analogical object, is blocked at object and partly automatically finishes the completion of disappearance part.
(2) technical scheme
For addressing the above problem, the invention provides a kind of analogical object method for automatically completing hidden parts based on geological information, may further comprise the steps:
S1, utilize the profile geological information for the treatment of source object in completion contours of objects geological information, this image in the image, and the one-to-one relationship of being set up for the treatment of all pixels on completion object and the source object profile, to treat that the completion object is divided into two parts: part is be blocked part and part b a)) part is not blocked;
S2 for part a), adopts the pixel value of the each point of the source object after the distortion to carry out completion, wherein, the source object after the distortion with treat that the shape of completion object is identical;
Wherein, described completion object and the source object treated is analogical object, that is, in the same width of cloth image with the Different Individual of a kind of object (similar object).
Wherein, for part b), completion in such a way: the pixel value that calculates the part each point that is not blocked is poor with the pixel value of described source object corresponding point, if difference, then adopts the described corresponding point partly that are not blocked of pixel value completion of the source object each point after the described distortion less than preset threshold value; Otherwise the method for cutting apart (Graph-Cut) with maximum figure finds an optimal path, makes energy
Minimum, c in the formula
I, originAnd c
I, referenceRepresent the pixel value for the treatment of completion object each point of a pixel position correspondence on the current path and the pixel value of source object corresponding point respectively; I, n are integer, and n represents the sum of the pixel on the current path, and the target of optimization is the absolute difference and the minimum of n point on the whole piece path, therefore calculates n from 0.Adopt the pixel value of the described each point partly that is not blocked and the pixel value of the source object corresponding point after the described distortion to come pixel of this part that is not blocked of completion respectively in the both sides of this optimal path.
Wherein, in step S1, utilize the in shape hereinafter method foundation of (Shape-Context) to treat the one-to-one relationship of each pixel on completion object and the two profile of source object, utilize the method for thin-plate spline interpolation (Thin Plate Interpolate) to set up the one-to-one relationship for the treatment of completion object and all pixels of source object then.
Between step S1 and step S2, comprise step with described source object distortion.
(3) beneficial effect
Technical scheme of the present invention is by the be blocked part and the part that is not blocked in the while completion analogical object, make the analogical object after the completion in the image have the dual integrality on geological information and the semantic information, handle so that carry out the intelligent image of object level truly; The method of employing Graph-Cut does not find the path of an optimum when completion is blocked part,, make the completion of shield portions has not been reached seamless spliced effect as adopting the source object content respectively and treating the two-part separatrix that completion object self content is filled with this paths.
Description of drawings
Fig. 1 is a method flow diagram of the invention process;
The synoptic diagram of Fig. 2 for carrying out completion at the different piece for the treatment of the completion object in the method for the embodiment of the invention.
Embodiment
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.Following examples are used to illustrate the present invention, but are not used for limiting the scope of the invention.
Fig. 1 is a method flow diagram of the present invention.With reference to Fig. 1, the present invention at first utilizes and treats in the image in completion contours of objects geological information, this image that (this source object does not need the accurately identical of shape size etc. with treating to belong to source object with a kind of object by the completion object, it just need be the similar object on the same width of cloth image, but profile geological information complete being presented on the image), and the one-to-one relationship of being set up for the treatment of all pixels on completion object and the source object profile, be blocked partly and the part scope of remainder thereby judge.Wherein, utilize the contextual method of shape to set up the one-to-one relationship for the treatment of each pixel on completion object and the two profile of source object, utilize the method for thin-plate spline interpolation to set up the one-to-one relationship for the treatment of completion object and all pixels of source object then.
Specifically, these parts are by realizing storing and using with the equal-sized different illiteracy plate (mask) of image.Simultaneously, under the situation of known profile geological information, the present invention need be with source object, by the mode of distortion (to reach and treat the identical purpose of completion object shapes), finishes corresponding one by one on the semantic content for the treatment of completion object and source object.Like this, in the completion for the treatment of completion object optional position, the content that fill just all has two selections: a) self content itself; B) content of the source object correspondence position of institute's reference.
What Fig. 2 represented is for the method for filling content choice.At the hiding content for the treatment of that the completion object is blocked, because the content of this part lacks fully, so directly adopt the content of the source object after the distortion to fill.And for the continuity that guarantees content after the present invention is to the object completion and integrality semantically, the part that is not blocked can not directly keep self content for the treatment of the completion object, and should keep the continuity of boundary in order to make two parts result after the completion, simultaneously, again in order to guarantee the similarity of adjusted object completion result and self raw content, as shown in Figure 2, the present invention takes following principle to carry out the filling of content:
Difference between self content of the part that at first is not blocked and the content of described source object, difference then directly adopt self content filling less than a certain threshold value; Greater than this threshold value,,, find the path of an optimum at difference, make energy with the method for Graph-Cut then at this part:
Reach minimum, c in the formula
I, originAnd c
I, referenceRepresent the pixel value for the treatment of completion object each point of a pixel position correspondence on the current path and the pixel value of source object corresponding point respectively; I, n are integer, and n represents the sum of the pixel on the paths.The both sides in path adopt the content of self content and source object to fill respectively, thereby can think that this path is by two seamless spliced different modes completions.
The present invention in this step, in order to realize completion result's optimization, can also choose a plurality of different source objects and carry out completion respectively, and get the E reckling that wherein calculates, as net result in the above.
As can be seen from the above embodiments, this technical scheme is by the be blocked part and the part that is not blocked in the while completion analogical object, make the analogical object that extracts in the image have the dual integrality on geological information and the semantic information, handle so that carry out the intelligent image of object level truly; The method of employing Graph-Cut does not find the path of an optimum when completion is blocked part,, make the completion of shield portions has not been reached seamless spliced effect as adopting the source object content respectively and treating the two-part separatrix that completion object self content is filled with this paths.
The above only is a preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the technology of the present invention principle; can also make some improvement and modification, these improve and modification also should be considered as protection scope of the present invention.
Claims (4)
1. the analogical object method for automatically completing hidden parts based on geological information is characterized in that, may further comprise the steps:
S1, utilize the profile geological information for the treatment of source object in completion contours of objects geological information, this image in the image, and the one-to-one relationship of being set up for the treatment of all pixels on completion object and the source object profile, to treat that the completion object is divided into two parts: part is be blocked part and part b a)) part is not blocked;
S2 for part a), adopts the pixel value of the each point of the source object after the distortion to carry out completion, wherein, the source object after the distortion with treat that the shape of completion object is identical;
Wherein, described completion object and the source object treated is with a kind of Different Individual of object in the same width of cloth image.
2. the analogical object method for automatically completing hidden parts based on geological information as claimed in claim 1, it is characterized in that, for part b), completion in such a way: pixel value poor of calculating the pixel value of the part each point that is not blocked and described source object corresponding point, if difference, then adopts the corresponding point of the described part that is not blocked of the pixel value completion of the source object each point after the described distortion less than preset threshold value; Otherwise the method for cutting apart with maximum figure finds an optimal path, makes energy
Minimum, c in the formula
I, originAnd c
I, referenceRepresent the pixel value for the treatment of completion object each point of a pixel position correspondence on the current path and the pixel value of source object corresponding point respectively; I, n are integer, and n represents the sum of the pixel on the current path; Adopt the pixel value of the described each point partly that is not blocked and the pixel value of the source object corresponding point after the described distortion to come pixel of this part that is not blocked of completion respectively in the both sides of this optimal path.
3. the analogical object method for automatically completing hidden parts based on geological information as claimed in claim 1, it is characterized in that, in step S1, utilize the contextual method of shape to set up the one-to-one relationship for the treatment of each pixel on completion object and the two profile of source object, utilize the method for thin-plate spline interpolation to set up the one-to-one relationship for the treatment of completion object and all pixels of source object then.
4. as claim 1 or 2 or 3 described analogical object method for automatically completing hidden parts, it is characterized in that, between step S1 and step S2, comprise step described source object distortion based on geological information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010101584401A CN101833781B (en) | 2010-04-22 | 2010-04-22 | Method for automatically completing hidden parts of similar objects based on geometric information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010101584401A CN101833781B (en) | 2010-04-22 | 2010-04-22 | Method for automatically completing hidden parts of similar objects based on geometric information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101833781A true CN101833781A (en) | 2010-09-15 |
| CN101833781B CN101833781B (en) | 2012-09-05 |
Family
ID=42717842
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010101584401A Active CN101833781B (en) | 2010-04-22 | 2010-04-22 | Method for automatically completing hidden parts of similar objects based on geometric information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101833781B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107957750A (en) * | 2017-12-15 | 2018-04-24 | 广东欧珀移动通信有限公司 | Electronic device, screenshot method and Related product |
| CN111079494A (en) * | 2019-06-09 | 2020-04-28 | 广东小天才科技有限公司 | Learning content pushing method and electronic equipment |
| CN113012126A (en) * | 2021-03-17 | 2021-06-22 | 武汉联影智融医疗科技有限公司 | Mark point reconstruction method and device, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005101324A1 (en) * | 2004-04-14 | 2005-10-27 | Koninklijke Philips Electronics N.V. | Ghost artifact reduction for rendering 2.5d graphics |
| CN101141633A (en) * | 2007-08-28 | 2008-03-12 | 湖南大学 | Moving object detecting and tracing method in complex scene |
| CN101571950A (en) * | 2009-03-25 | 2009-11-04 | 湖南大学 | Image restoring method based on isotropic diffusion and sparse representation |
| CN101635047A (en) * | 2009-03-25 | 2010-01-27 | 湖南大学 | Texture synthesis and image repair method based on wavelet transformation |
-
2010
- 2010-04-22 CN CN2010101584401A patent/CN101833781B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005101324A1 (en) * | 2004-04-14 | 2005-10-27 | Koninklijke Philips Electronics N.V. | Ghost artifact reduction for rendering 2.5d graphics |
| CN101141633A (en) * | 2007-08-28 | 2008-03-12 | 湖南大学 | Moving object detecting and tracing method in complex scene |
| CN101571950A (en) * | 2009-03-25 | 2009-11-04 | 湖南大学 | Image restoring method based on isotropic diffusion and sparse representation |
| CN101635047A (en) * | 2009-03-25 | 2010-01-27 | 湖南大学 | Texture synthesis and image repair method based on wavelet transformation |
Non-Patent Citations (1)
| Title |
|---|
| 《中国图象图形学报》 20070131 张红英等 数字图像修复技术综述 1-10 1-4 第12卷, 第1期 2 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107957750A (en) * | 2017-12-15 | 2018-04-24 | 广东欧珀移动通信有限公司 | Electronic device, screenshot method and Related product |
| CN107957750B (en) * | 2017-12-15 | 2020-05-22 | Oppo广东移动通信有限公司 | Electronic device, screenshot method and related product |
| CN111079494A (en) * | 2019-06-09 | 2020-04-28 | 广东小天才科技有限公司 | Learning content pushing method and electronic equipment |
| CN111079494B (en) * | 2019-06-09 | 2023-08-25 | 广东小天才科技有限公司 | Learning content pushing method and electronic equipment |
| CN113012126A (en) * | 2021-03-17 | 2021-06-22 | 武汉联影智融医疗科技有限公司 | Mark point reconstruction method and device, computer equipment and storage medium |
| CN113012126B (en) * | 2021-03-17 | 2024-03-22 | 武汉联影智融医疗科技有限公司 | Method, device, computer equipment and storage medium for reconstructing marking point |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101833781B (en) | 2012-09-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105488758B (en) | A kind of image-scaling method based on perception of content | |
| US9508126B2 (en) | Image haze removal using fast constrained transmission estimation | |
| CN101833781B (en) | Method for automatically completing hidden parts of similar objects based on geometric information | |
| CN102184533A (en) | Non-local-restriction-based total variation image deblurring method | |
| CN105741327A (en) | Method and apparatus for extracting dominant color and assertive color of picture | |
| CN103914819B (en) | A kind of based on the infrared image joining method improving RANSAC | |
| CN103106038B (en) | Based on video interactive method and the system of sketch gesture and arc abstract timer shaft | |
| Matsui et al. | Interactive manga retargeting | |
| CN104952089B (en) | A kind of image processing method and system | |
| CN103413331B (en) | A kind of support edits the high resolution video image content sparse expression method propagated | |
| CN104200451B (en) | Image fusion method based on non-local sparse K-SVD algorithm | |
| CN113392727B (en) | RGB-D salient object detection method based on dynamic feature selection | |
| Zheng et al. | Self-supervised monocular depth estimation based on combining convolution and multilayer perceptron | |
| CN102855025B (en) | Optical multi-touch contact detection method based on visual attention model | |
| Shi et al. | Context‐guided ground truth sampling for multi‐modality data augmentation in autonomous driving | |
| Bare et al. | Pixel fusion based stereo image retargeting | |
| CN105046696A (en) | Image matching method based on deep planar constraint graph cut optimization | |
| Wang et al. | Avoiding bleeding in image blending | |
| EP3062288A1 (en) | Method, apparatus and computer program product for reducing chromatic aberrations in deconvolved images | |
| CN113034547A (en) | Target tracking method, digital integrated circuit chip, electronic device, and storage medium | |
| Lan et al. | Physical-model guided self-distillation network for single image dehazing | |
| CN104700364A (en) | Video image enhancement method applied to players | |
| CN102695069B (en) | Depth propagation method in video conversion from two dimension to three dimension | |
| CN103258317B (en) | The method realizing image color correction conversion based on sample image in computer system | |
| Choi et al. | An efficient fundamental matrix estimation for moving object detection |
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 |