CN1710611A - Human body posture shape-changing method based on optimization throught - Google Patents
Human body posture shape-changing method based on optimization throught Download PDFInfo
- Publication number
- CN1710611A CN1710611A CN 200510012175 CN200510012175A CN1710611A CN 1710611 A CN1710611 A CN 1710611A CN 200510012175 CN200510012175 CN 200510012175 CN 200510012175 A CN200510012175 A CN 200510012175A CN 1710611 A CN1710611 A CN 1710611A
- Authority
- CN
- China
- Prior art keywords
- human body
- manikin
- criterion
- attitude
- parameter
- 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
- Processing Or Creating Images (AREA)
Abstract
The method includes following steps: (1) defining a common parametrized 3D human model; (2) based on video content, customizing 3D structural information of human body; (3) picking up information of human body figure from video; (4) projecting gesture of 3D human body into 2D plane so as to form profile of 2D model; (5) according to deformation rules based on optimized algorithm, carrying out deformation for gesture of parametrized 3D human model. The invention is in use for recovering each type of 3D structural parameters of moving object if original 3D information is known. Features are: favorable generality and wide application area.
Description
Technical field
The present invention relates to the Computer Applied Technology field, particularly a kind of based on the human body posture deforming method of optimizing thought.
Background technology
Obtaining three-dimensional human body attitude information from two-dimensional video is the focus and the difficult point problem in computer vision, pattern-recognition, virtual reality, Intelligent Human-Machine Interface field.In this article, by convention, we are unified to be abbreviated as " 2D " with " two dimension ", and " three-dimensional " is abbreviated as " 3D ".
Under the prerequisite of known initial human body attitude 3D information, how initial attitude to be out of shape according to video content, thereby obtaining 3D information corresponding in the video is one of them very important subproblem, it not only has important Research Significance, and at recreation, key frame 3D animation and the aspect such as obtain based on the 3D information of instance data and have broad application prospects.For example, in human motion analysis field,, just can recover corresponding human body attitude information according to video image as long as we have the 3D human body attitude storehouse of a small sample based on video.For another example, in the animation field,, realize the 3D animation effect easily as long as the 2D key frame of a given small sample and corresponding 3D attitude information thereof just can recover out with the 3D information of all 2D animation frame.
Therefore, the important in theory meaning is not only arranged, also have application fields and important practical value based on the human body posture deforming technology of video content.But, in the existing motion analysis and reconstruction software at home and abroad, all do not provide human body posture deforming technical functionality based on video content based on video.When the patent retrieval of being correlated with, do not retrieve the information of any relevant patent yet.
Summary of the invention
The purpose of this invention is to provide a kind of human body posture deforming method, be implemented under the situation of known initial 3D attitude,, the 3D attitude is out of shape, thereby recover human body 3D attitude information corresponding in the video according to video content based on optimization thought.
To achieve these goals, the invention provides a kind of based on the human body posture deforming method of optimizing thought, be used for basis at known initial human body three-dimensional attitude information, according to human body information content in the video, initial human body 3D attitude is out of shape according to video content, thereby seeks out 3D organization of human body information corresponding in the video; This method may further comprise the steps:
1) for given general parameter 3D manikin, it is customized according to video content;
2) extract human body contour outline information in the video;
3), form the 2D model silhouette with 3D human body attitude (surface geometry model description) projection on the 2D plane;
4) according to distortion criterion parametric three D manikin attitude is out of shape based on optimized Algorithm.
In the technique scheme, the general parametric three D manikin described in the step 1) is achieved in that
Parametric three D manikin comprises human synovial angle parameter x
a, internal ratio parameter (zoom factor) x
iTherefore, manikin can be expressed as: x=(x
a, x
i).
In the technique scheme, be achieved in that according to parametric three D manikin attitude being out of shape in the step 4) based on the distortion criterion of optimized Algorithm
The distortion criterion comprises:
1) model silhouette is by the criterion that comprises of visual profiles;
2) the overlapping maximization criterion of model silhouette and visual profiles;
3) joint angles constraint criterion;
More than the execution of distortion criterion realizes by the mathematical optimization objective function.
Wherein, above-mentioned model silhouette is comprised criterion by visual profiles and is meant:
If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone,, make the S of 2D projected outline of 3D manikin attitude by the scaling of model factor of automatic adjustment parametric three D manikin
aBy visual profiles S
gCover, i.e. S
a∈ S
g
The overlapping maximization criterion of above-mentioned model silhouette and visual profiles is meant:
If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone,, make the S of 2D projected outline of 3D manikin attitude by the scaling of model factor parameter and the joint angles parameter of automatic adjustment parametric three D manikin
aWith visual profiles S
gOverlapping maximum, i.e. two profile S
aWith S
gThe difference set minimum, i.e. S
a∈ S
g, S
g∈ S
aSet up simultaneously.
Above-mentioned joint angles constraint criterion is meant:
All there is specific scope the zone of action in each joint of people, in the joint angles parameter x of adjusting the parametrization manikin automatically
aThe time, observe the human synovial angle restriction that satisfies Biological Principles, simultaneously, avoid the unreasonable limbs penetration phenomenon that causes of angle parameter.
The invention has the advantages that:
1, the inventive method has realized that the human body posture deforming method based on video content is not only had important significance for theories, and is with a wide range of applications and great use value.
2, the inventive method can be used for various types of motion objects, has good versatility.
Description of drawings
Fig. 1 is based on the human body posture deforming techniqueflow chart of optimizing thought.
Embodiment
Below in conjunction with accompanying drawing, the method for the invention is described further.
As shown in Figure 1, be the process flow diagram of this method, expression operation in the frame of broken lines in the process flow diagram, the result that the expression associative operation obtains in the solid box.
A kind of human body posture deforming technology based on optimization thought of the present invention mainly is divided into following steps:
General parametric three D manikin of step 1. definition, and according to video content customization 3D model parameter;
Parametric three D manikin comprises human synovial angle parameter x
a, internal ratio parameter (zoom factor) x
iTherefore, manikin can be expressed as: x=(x
a, x
i);
According to the body information of human body in the video, the body of general 3D parametrization manikin is adjusted accordingly, make both on body, match;
The distortion criterion comprises:
A1, model silhouette are by the criterion that comprises of visual profiles.If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone.By the scaling of model factor of automatic adjustment parametric three D manikin, make the S of 2D projected outline of 3D manikin attitude
aBy visual profiles S
gCover, i.e. S
a∈ S
gThe criterion function that quantizes is:
e
a(x
i)=|P(Person(x
i,x
a))-S
g|;
Wherein, P is a projection matrix, Person (x
i) be parameterized manikin.
The overlapping maximization criterion of b1, model silhouette and visual profiles.If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone.By the scaling of model factor parameter and the joint angles parameter of automatic adjustment parametric three D manikin, make the S of 2D projected outline of 3D manikin attitude
aWith visual profiles S
gOverlapping maximum, i.e. two profile S
aWith S
gThe difference set minimum.Be S
a∈ S
g, S
g∈ S
aSet up simultaneously.The criterion function that quantizes is:
C1, joint angles constraint criterion.All there is specific scope the zone of action in each joint of people.In the joint angles parameter x of adjusting the parametrization manikin automatically
aThe time, observe the human synovial angle restriction that satisfies Biological Principles, simultaneously, avoid the unreasonable limbs penetration phenomenon that causes of angle parameter.
More than the execution of distortion criterion realizes by objective function is carried out numerical optimization.Specifically,
Given initial manikin attitude x=(x
a, x
i), in conjunction with above three distortion criterions, objective function f (x) can be expressed as: f (x)=e
a(x
i)+e
s(x
a, x
i)+e
p(x
a).
Wherein, e
aFor the model silhouette that quantizes by the criterion that comprises of visual profiles; e
sBe model silhouette and the overlapping maximization criterion of visual profiles; e
pBe the constraint of human body joint angles.It should be noted that in three above quantification criterions e
aOnly to scaling of model factor x
iOperate e
sSimultaneously zoom factor and human synovial angle parameter are changed, and e
pOnly to human synovial angle parameter x
aHandle.
Claims (7)
1, a kind of based on the human body posture deforming method of optimizing thought, be used for basis at known initial human body three-dimensional attitude information, according to human body information content in the video, initial human body 3D attitude is out of shape according to video content, thereby seeks out 3D organization of human body information corresponding in the video; This method may further comprise the steps:
1) for given general parameter 3D manikin, it is customized according to video content;
2) extract human body contour outline information in the video;
3), form the 2D model silhouette with the projection on the 2D plane of 3D human body attitude;
4) according to distortion criterion parametric three D manikin attitude is out of shape based on optimized Algorithm.
2, according to claim 1 based on the human body posture deforming method of optimizing thought, it is characterized in that the described general parameter 3D of step 1) manikin is achieved in that
Parametric three D manikin comprises human synovial angle parameter x
a, the internal ratio parameter x
iTherefore, manikin can be expressed as: x=(x
a, x
i).
3, the human body posture deforming method based on optimization thought according to claim 1 is characterized in that, is achieved in that according to based on the distortion criterion of optimized Algorithm parametric three D manikin attitude being out of shape in the step 4)
The distortion criterion comprises:
1) model silhouette is by the criterion that comprises of visual profiles;
2) the overlapping maximization criterion of model silhouette and visual profiles;
3) joint angles constraint criterion;
More than the execution of distortion criterion realizes by the mathematical optimization objective function.
4, according to claim 3ly it is characterized in that, describedly comprised criterion according to model silhouette by visual profiles and be meant based on the human body posture deforming method of optimizing thought:
If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone,, make the S of 2D projected outline of 3D manikin attitude by the scaling of model factor of automatic adjustment parametric three D manikin
aBy visual profiles S
gCover, i.e. S
a∈ S
g
5, according to claim 3 based on the human body posture deforming method of optimizing thought, it is characterized in that the overlapping maximization criterion of described model silhouette and visual profiles is meant:
If S
aBe the 2D projected outline zone of manikin, S
gBe the visual profiles zone,, make the S of 2D projected outline of 3D manikin attitude by the scaling of model factor parameter and the joint angles parameter of automatic adjustment parametric three D manikin
aWith visual profiles S
gOverlapping maximum, i.e. two profile S
aWith S
gThe difference set minimum, i.e. S
a∈ S
g, S
g∈ S
aSet up simultaneously.
6, according to claim 3 based on the human body posture deforming method of optimizing thought, it is characterized in that described joint angles constraint criterion is meant:
All there is specific scope the zone of action in each joint of people, in the joint angles parameter x of adjusting the parametrization manikin automatically
aThe time, observe the human synovial angle restriction that satisfies Biological Principles, simultaneously, avoid the unreasonable limbs penetration phenomenon that causes of angle parameter.
7, the human body posture deforming method based on video content according to claim 3 is characterized in that, the execution of described distortion criterion is the realization by the mathematical optimization objective function: specifically,
Given initial manikin attitude x=(x
a, x
i), in conjunction with above three distortion criterions, objective function f (x) can be expressed as: f (x)=e
a(x
i)+e
s(x
a, x
i)+e
p(x
a);
Wherein, e
aFor the model silhouette that quantizes by the criterion that comprises of visual profiles; e
sBe model silhouette and the overlapping maximization criterion of visual profiles; e
pBe the constraint of human body joint angles, it should be noted that in three above quantification criterions e
aOnly to scaling of model factor x
iOperate e
sSimultaneously zoom factor and human synovial angle parameter are changed, and e
pOnly to human synovial angle parameter x
aHandle;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200510012175 CN1710611A (en) | 2005-07-14 | 2005-07-14 | Human body posture shape-changing method based on optimization throught |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200510012175 CN1710611A (en) | 2005-07-14 | 2005-07-14 | Human body posture shape-changing method based on optimization throught |
Publications (1)
Publication Number | Publication Date |
---|---|
CN1710611A true CN1710611A (en) | 2005-12-21 |
Family
ID=35706850
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200510012175 Pending CN1710611A (en) | 2005-07-14 | 2005-07-14 | Human body posture shape-changing method based on optimization throught |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1710611A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104183002A (en) * | 2013-05-27 | 2014-12-03 | 索尼公司 | Three-dimensional model change method and device |
CN107105310A (en) * | 2017-05-05 | 2017-08-29 | 广州盈可视电子科技有限公司 | Figure image replacement method, device and a kind of recording and broadcasting system in a kind of net cast |
CN109770943A (en) * | 2019-01-28 | 2019-05-21 | 电子科技大学 | A kind of ultrasonic automatic optimization method positioned using computer vision |
CN110415336A (en) * | 2019-07-12 | 2019-11-05 | 清华大学 | High-precision human posture method for reconstructing and system |
-
2005
- 2005-07-14 CN CN 200510012175 patent/CN1710611A/en active Pending
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104183002A (en) * | 2013-05-27 | 2014-12-03 | 索尼公司 | Three-dimensional model change method and device |
CN104183002B (en) * | 2013-05-27 | 2020-03-27 | 索尼公司 | Three-dimensional model changing method and device |
CN107105310A (en) * | 2017-05-05 | 2017-08-29 | 广州盈可视电子科技有限公司 | Figure image replacement method, device and a kind of recording and broadcasting system in a kind of net cast |
CN107105310B (en) * | 2017-05-05 | 2020-07-10 | 广州盈可视电子科技有限公司 | Human image replacing method and device in video live broadcast and recording and broadcasting system |
CN109770943A (en) * | 2019-01-28 | 2019-05-21 | 电子科技大学 | A kind of ultrasonic automatic optimization method positioned using computer vision |
CN109770943B (en) * | 2019-01-28 | 2021-11-02 | 电子科技大学 | Ultrasonic automatic optimization method using computer vision positioning |
CN110415336A (en) * | 2019-07-12 | 2019-11-05 | 清华大学 | High-precision human posture method for reconstructing and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Li et al. | Hand gesture recognition based on convolution neural network | |
Lin et al. | Semantic decomposition and reconstruction of residential scenes from LiDAR data | |
Afif et al. | Indoor objects detection and recognition for an ICT mobility assistance of visually impaired people | |
CN1186746C (en) | Method for building three-dimensional scene by analysing sequence of images | |
CN101452582B (en) | Method and device for implementing three-dimensional video specific action | |
CN1725246A (en) | A kind of human body posture deforming method based on video content | |
CN105894045A (en) | Vehicle type recognition method with deep network model based on spatial pyramid pooling | |
CN1617174A (en) | Human limb three-dimensional model building method based on image cutline | |
CN101038671A (en) | Tracking method of three-dimensional finger motion locus based on stereo vision | |
Zhou et al. | Non-iterative denoising algorithm based on a dual threshold for a 3D point cloud | |
WO2010046101A1 (en) | Device, method and computer program for detecting a gesture in an image, and said device, method and computer program for controlling a device | |
CN110135277B (en) | Human behavior recognition method based on convolutional neural network | |
CN1932846A (en) | Visual frequency humary face tracking identification method based on appearance model | |
CN113379898A (en) | Three-dimensional indoor scene reconstruction method based on semantic segmentation | |
CN1710611A (en) | Human body posture shape-changing method based on optimization throught | |
CN104751493A (en) | Sparse tracking method on basis of gradient texture features | |
CN1569558A (en) | Moving robot's vision navigation method based on image representation feature | |
CN1710614A (en) | Method for evaluating parameter of 3-D motion of human climbs based on model | |
CN113536232A (en) | Normal distribution transformation method for laser point cloud positioning in unmanned driving | |
Moustakas et al. | SQ-Map: Efficient layered collision detection and haptic rendering | |
CN108664126B (en) | Deformable hand grabbing interaction method in virtual reality environment | |
CN1766928A (en) | A kind of motion object center of gravity track extraction method based on the dynamic background sport video | |
Wang et al. | Methods of As-is BIM reconstruction using point cloud data for existing buildings | |
CN113128624B (en) | Graph network face recovery method based on multi-scale dictionary | |
CN115761118A (en) | Real-time surface reconstruction method based on deep learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C12 | Rejection of a patent application after its publication | ||
RJ01 | Rejection of invention patent application after publication |