CN1271574C - Human limb three-dimensional model building method based on labelling point - Google Patents

Human limb three-dimensional model building method based on labelling point Download PDF

Info

Publication number
CN1271574C
CN1271574C CN 200410089300 CN200410089300A CN1271574C CN 1271574 C CN1271574 C CN 1271574C CN 200410089300 CN200410089300 CN 200410089300 CN 200410089300 A CN200410089300 A CN 200410089300A CN 1271574 C CN1271574 C CN 1271574C
Authority
CN
China
Prior art keywords
limbs
dimensional
points
skeleton
gauge point
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.)
Expired - Fee Related
Application number
CN 200410089300
Other languages
Chinese (zh)
Other versions
CN1617175A (en
Inventor
潘海朗
刘允才
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Jiaotong University
Original Assignee
Shanghai Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shanghai Jiaotong University filed Critical Shanghai Jiaotong University
Priority to CN 200410089300 priority Critical patent/CN1271574C/en
Publication of CN1617175A publication Critical patent/CN1617175A/en
Application granted granted Critical
Publication of CN1271574C publication Critical patent/CN1271574C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Processing Or Creating Images (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

The present invention relates to a three-dimensional model building method for human limbs on the basis of marking points. At first, the present invention uses the image sequence for posing postures of the human limbs, which is shot by a binocular stereo vision system, and extracts and matches the marking points from the image sequence so as to determine the three dimensional coordinates of corresponding space of the marking points of a left image and a right image at any time; thus, the present invention finds out the spatial three-dimensional position information of each limb skeleton at any time; the three dimensional coordinate value of the marking points of each limb is carried into a belonging rotary cone curved surface equation set so as to find out deformation parameters; finally, the present invention uses spherosome and rotating cone curved surfaces to draw the model of the human limbs. The method of the present invention has the advantages of simplicity and easy realization. The eastablished three-dimensional model for the human limbs is divided into a skeleton layer and a skin layer, wherein the skeleton layer represents the skeleton structure of the limbs, and is composed of articulation points and line sections connecting the articulation points; the skin layer is represented by a rotary conical curved surface. Each limb can reflect the skin deformation of the human limbs only through adjusting a deformation parameter. Thus, the present invention solves the problem of excessive skin deformation parameters representing the three-dimensional model of a human body.

Description

Human limb three-dimensional model building based on gauge point
Technical field
The present invention relates to a kind of human limb three-dimensional model building based on gauge point, be used for the perception interface, application such as motion analysis and virtual reality all has actual application value at aspects such as medical image, biomedicine, gesture identification, video conference, video-game, news broadcast automatically, film making, material deformation, image compression.Belong to computer vision and field of Computer Graphics.
Background technology
The research object in motion analysis field mainly is divided into rigid objects, connects rigid body and non-rigid object etc.Study on Kinematic Analysis in the computer vision all mainly concentrates on the research to rigid motion, and scholars have obtained a series of achievement, and has set up comparatively perfect theoretical frame, still, is non-rigid motion mostly in real world.Human body belongs to non-rigid body.
In more than ten years in the past, deformable people's volume modeling has been obtained some achievements.Nahas etc. have used B-spline surface to represent the motion of the softness of the person and face.Petland etc. have introduced a kind of based on the finite element analysis method model.Terzopoulos etc. have introduced deformable hyperquadric model that the isolated degree of freedom of the global shape parameter that is expressed as a super ellipsoid form and batten form combines.Three kinds of basic softwares of uses such as Min come human upper limb and the shoulder muscle modeling to different shape, and a some primitive produces an ellipsoid, and one group of line segment that connects produces a right cylinder, and a triangle gridding produces the shape of a complexity.Sminchisescu etc. have proposed a kind of people's phantom shape, and it is made of motion skeleton and muscle, and the skeleton that the node of connection constitutes is by the angle parameter control of node, and have covered and comprise the muscle that the auxiliary super cylindrical elliptic surface that is tapered with bending parameters constitutes.Model comprises 30 node parameters and 8 internal ratio parameters, and each limbs comprises 9 deformation parameters again.Plankers etc. are attached to a kind of implicit surface that calls first ball or software on the connection skeleton of human body and according to anatomical degree of approximation and arrange these yuan ball.This manikin has used 230 first balls altogether, and uses the B-spline surface sheet to carry out covering.Can also carry out people's volume modeling with various implicit surfaces such as blobby molecule, software, convolution curved surfaces.But these methods all adopt comparatively complicated non-rigid motion model, and controlled variable is more, can not well satisfy the evaluation criterion of contour representation: i.e. Biao Shi terseness and be suitable for the calculating of post-processing stages.So these models are lacking practicality aspect the expression human body contour outline.
Summary of the invention
The objective of the invention is to represent the too much deficiency of parameter of skin deformation at existing human 3d model, a kind of human limb three-dimensional model building based on gauge point is proposed, simply realize easily, the model of setting up uses the distortion that considerably less parameter just can representative's limb skin, reduce the calculated amount of motion analysis, be suitable for the modeling that various elasticity connect rigid body.
For realizing such purpose, the human limb three-dimensional model that the present invention sets up is divided into two-layer: casing play and skin layer.Casing play has been represented the skeleton structure of people's limbs, is made up of with the line segment that is connected these articulation points articulation point.Skin layer represents that with a kind of new revolving conic surface every section limbs only need be adjusted the distortion that a deformation parameter just can reflect skin.At first use the binocular tri-dimensional vision system to take the image sequence that people's limbs are posed, extraction, matched indicia point from the binocular image sequence, determine the 3 d space coordinate of image tagged point by the binocular stereo vision method, obtain the spatial three-dimensional position information of skeleton again according to gauge point three-dimensional coordinate and priori.Revolving conic surface system of equations under the D coordinates value substitution of gauge point on each section limbs is obtained the deformation parameter of each section limbs.Use spherosome and revolving conic surface to draw people's limbs model at last.
Modeling method of the present invention mainly comprises following step:
1. from image, extract, the matched indicia point, obtain the 3 d space coordinate of gauge point correspondence.
On upper and lower two sections of people's limbs, three gauge points are set respectively respectively, adopt the binocular tri-dimensional vision system to take and obtain the image sequence that people's limbs are posed, extract the plane of delineation two-dimensional coordinate of these gauge points in each left and right image of moment, and draw the corresponding relation between the gauge point on left side camera image and the right side camera review.Obtain the corresponding relation of planar coordinate and world coordinate system 3 d space coordinate on the left and right camera review by the camera calibration technology, by obtain each constantly on left and right image the gauge point two-dimensional coordinate obtain corresponding 3 d space coordinate.
Just can obtain its deformation parameter because of every section limbs only need 3 unique points, the inventive method has been pasted six paper gauge points on everyone limb skin, wherein three limbs epimere, three at the limbs hypomere.
2. extract people's limbs frame position information.
According to the three-dimensional coordinate of two gauge point A, the B at two ends in three gauge points of each section limbs and the gauge point A that measures in advance, B place limbs radius, obtain each 3 d space coordinate of skeleton end points constantly, obtain the spatial three-dimensional position information of skeleton thus;
3. obtain deformation parameter
Two gauge point A at limbs, 1 C of straight line top setting space that B forms, through 3 A, B, set 1 D outside the plane of C, 4 A, B, C, D forms three plane ACD, BCD, ABD, determine through some A according to these three planes, B and with straight line AC, the conical surface equation that BC is tangent, this conical surface and process A, B, the intersection on the plane that C is 3 forms conic section, conic section is formed revolving conic surface around the rotation of skeleton straight line, set up the revolving conic surface system of equations, revolving conic surface system of equations under the 3 d space coordinate value substitution of gauge point on each section limbs that obtain is found the solution, obtained each deformation parameter ρ of the revolving conic surface of each section limbs constantly.
4. modeling rendering
The present invention uses the OpenGL program to adopt parallel projection methods drawing three-dimensional people limbs model, use three spherosomes to represent three articulation points of limbs, the curved surface that two revolving conic surface equations that employing is tried to achieve are drawn out is represented upper and lower two sections limb skins, connects these three spherosomes and two curved surfaces formation people limbs models.
The inventive method is simple, realizes easily.The model of setting up uses the distortion that considerably less parameter just can representative's limb skin, and every section limbs are deformation parameter of need only, has the ability that high-level description and prediction are blocked.The inventive method is suitable for the modelling that many elasticity connect rigid body.Experiment is adopted cheaply, the easy-on video frequency pick-up head comes image data, does not need optional equipment, represents the revolving conic surface system of equations and the derivation algorithm of skin deformation simple, greatly reduces the computational complexity of motion analysis.
Description of drawings
The conic section that Fig. 1 adopts for the present invention.
Among Fig. 1, A, B, C are three summits, determine equation of conic section by these three apex coordinates.
The behave skeleton structure of limbs of Fig. 2.
Among Fig. 2, j1, j2, j3 are articulation points, and an arm has the two sections limbs, big arm and forearm.
Fig. 3 is a revolving conic surface structural representation of the present invention.
Among Fig. 3, L is the line segment of representative's limbs skeleton, and S is a conic section, and S is around the L formation revolving conic surface that rotates a circle.P 0Be on the line segment L a bit, e is the direction vector of line segment L, P be on the revolving conic surface more arbitrarily, cross the P point and do the plane vertical with line segment L, the intersection point of this plane and conic section S is P s, the intersection point of this plane and line segment L is O, A, B are two end points of conic section S, A L, B LBe two end points of line segment L.
Fig. 4 is the experimental result that the present invention is based on the modeling of image tagged point.
Among Fig. 4, first row is the image sequence that left side camera is taken, and second row is to express corresponding to the three-dimensional model of the reflection people limb motion of the first row image and skin deformation.
Embodiment
In order to understand technical scheme of the present invention better, be described in further detail below in conjunction with drawings and Examples.Embodiment specifically carries out the description of people's limbs modeling process at accompanying drawing 4 left sides first width of cloth image (first moment left side camera is taken).
1. from image, extract, matched indicia point and obtain the 3 d space coordinate of gauge point correspondence.As Fig. 4 left side first width of cloth image, just can obtain its deformation parameter because of every section limbs only need 3 unique points, on the arm skin of human body, pasted six paper gauge points, wherein three gauge points on the big arm of arm, three at the arm forearm.Use the binocular tri-dimensional vision system to take the image sequence that people's limbs are posed.Extract the plane of delineation two-dimensional pixel coordinate of the big arm gauge point of arm A, B, P, the pixel coordinate of gauge point is respectively A on the left side camera image 1(365,266), B 1(519,190), P 1(444,218); The pixel coordinate of gauge point is respectively A on the camera review of right side 2(552,289), B 2(700,209), P 2(627,241).Obtain the relation of left and right sides image pixel coordinate and world coordinate system 3 d space coordinate by the camera calibration technology:
Z C 1 u 1 v 1 1 = 997.9595 0 372.7276 0 0 994.4506 301.4770 0 0 0 1 0 x y z 1
Z C 2 u 2 v 2 1 = 990.5091 0 361.3816 0 0 986 . 1246 310.2779 0 0 0 1 0 0.9097 - 0.0053 0.4150 - 190.4981 0.0005 0.9999 0.0117 1.9943 - 0.4150 - 0.0104 0.9097 41.1533 0 0 0 1 x y z 1
(u wherein 1, v 1) be two-dimensional pixel coordinate on the left side camera image, (u 2, v 2) be two-dimensional pixel coordinate on the camera review of right side, (x, y z) are 3 d space coordinate.
Gauge point A, B, P are brought into above-mentioned equation at the pixel coordinate of left and right image obtain the 3 d space coordinate A of these gauge point correspondences (6.5983,30.516 ,-853.52), B (115.99,86.684 ,-791.21), P (58.9288,67.4501 ,-824.7681).
2. extract people's limbs frame position information.People's limbs skeleton pattern comprises articulation point j1, j2, j3 and is connected the line segment of these articulation points (epimere limbs, hypomere limbs), as shown in Figure 2.Because gauge point A, the B place limbs radius at two ends change hardly in three gauge points of the big arm of arm shown in Fig. 4, by measuring gauge point A place limbs radius in advance is 40, gauge point B place limbs radius is 49, the 3 d space coordinate A that can obtain the skeleton end points by coordinate and gauge point A, the B place limbs radius of gauge point A, B L(x LA, y LA, z LA)=(6.3824 ,-6.0572 ,-847.1456), B L(x LB, y LB, z LB)=(132.3959,41.6061 ,-782.6021).Just can get the spatial three-dimensional position information of skeleton.The direction vector of skeleton be e (m, n, p), m=x wherein LB-x LA=126.0135, n=y LB-y LA=47.6633, p=z LB-z LA=64.5435.
3. obtain deformation parameter.1 C of setting space (6.5983,30.516 ,-853.52) above the straight line that gauge point A, the B at limbs two ends form, as shown in Figure 1.Plane equation through 3 A, B, C is: f ABC(x, y, z)=-0.5937x-0.0087y+1.1759z+1000=0.At plane f ABC(x, y, z) outer 1 D (10,10,10) that sets, 4 A, B, C, D form three plane ACD, BCD, ABD, and the equation on these three planes is respectively:
Plane ACD:a 0+ a 1X+a 2Y+a 3Z=1000-0.2333x-91.024y-8.7428z=0
Plane BCD:b 0+ b 1X+b 2Y+b 3Z=1000-9365.3x+8874.4y+390.9z=0
Plane ABD:u 0+ u 1X+u 2Y+u 3Z=1000+85.1095x-179.2157y-5.8939z=0
Determine according to these three planes to be: (a through some A, B and with the tangent conical surface equation of straight line AC, BC 0+ a 1X+a 2Y+a 3Z) (b 0+ b 1X+b 2Y+b 3Z)-ρ (u 0+ u 1X+u 2Y+u 3Z)=(1000-0.2333x-91.024y-8.7428z) (1000-9365.3x+8874.4y+390.9z)-ρ (1000+85.1095x-179.2157y-5.8939z)=0
Set up the revolving conic surface system of equations:
m ( x - x S ) + n ( y - y S ) + p ( z - z S ) = 0 ( x - x L ) 2 + ( y - y L ) 2 + ( z - z L ) 2 = ( x S - x L ) 2 + ( y S - y L ) 2 + ( z S - z L ) 2 ( a 0 + a 1 x S + a 2 y S + a 3 z S ) ( b 0 + b 1 x S + b 2 y S + b 3 z S ) = ρ ( u 0 + u 1 x S + u 2 y S + u 3 z S ) 2 f ABC ( x S , y S , z S ) = 0
Wherein
x L = x LA + m · ( - m ( x LA - x S ) + n ( y LA - y S ) + p ( z LA - z S ) m 2 + n 2 + p 2 ) y L = y LA + n · ( - m ( x LA - x S ) + n ( y LA - y S ) + p ( z LA - z S ) m 2 + n 2 + p 2 ) z L = z LA + p · ( - m ( x LA - x S ) + n ( y LA - y S ) + p ( z LA - z S ) m 2 + n 2 + p 2 )
The A that tries to achieve previously L(x LA, y LA, z LA), e (m, n, p), (a 0, a 1, a 2, a 3, b 0, b 1, b 2, b 3, u 0, u 1, u 2, u 3), f ABC(the above-mentioned Nonlinear System of Equations of value substitution z) is P (x for x, y P, y P, z P) in the value substitution above-mentioned system of equations (x, y z), adopt nonlinear least square method to obtain unknown parameter P s(x S, y S, z S) and ρ=-412.2517, ρ is this deformation parameter of big arm revolving conic surface of arm constantly.
Adopting uses the same method can try to achieve the revolving conic surface equation of representing the little skin of arm of arm.
4. modeling rendering.As Fig. 4 the 2nd row, use the OpenGL program to adopt parallel projection methods to draw the three-dimensional right arm model that obtains, use three radiuses to be respectively 50,40,30 spherosome and represent j1, j2, three articulation points of j3, adopt and represent the revolving conic surface equation of the big arm of arm, little skin of arm to draw out curved surface, connect these three articulation points and two curved surfaces formation people limbs models.

Claims (1)

1, a kind of human limb three-dimensional model building based on gauge point is characterized in that comprising following concrete steps:
1) respectively on people's limbs, three gauge points respectively are set on following two sections, adopt the binocular tri-dimensional vision system to take and obtain the image sequence that people's limbs are posed, extract each left side constantly, the plane of delineation two-dimensional coordinate of these gauge points in the right image, and draw the corresponding relation between the gauge point on left side camera image and the right side camera review, obtain a left side by the camera calibration technology, the corresponding relation of planar coordinate and world coordinate system 3 d space coordinate on the right camera review is by each moment left side that obtains, the gauge point two-dimensional coordinate is obtained corresponding 3 d space coordinate on the right image;
2) according to the three-dimensional coordinate of two gauge point A, the B at two ends in three gauge points of each section limbs and the gauge point A that measures in advance, B place limbs radius, obtain each 3 d space coordinate of skeleton end points constantly, obtain the spatial three-dimensional position information of skeleton thus;
3) at two gauge point A of limbs, 1 C of straight line top setting space that B forms, through 3 A, B, set 1 D outside the plane of C, 4 A, B, C, D forms three plane ACD, BCD, ABD, determine through some A according to these three planes, B and with straight line AC, the conical surface equation that BC is tangent, this conical surface and process A, B, the intersection on the plane that C is 3 forms conic section, conic section is formed revolving conic surface around the rotation of skeleton straight line, and set up the revolving conic surface system of equations, revolving conic surface system of equations under the 3 d space coordinate value substitution of gauge point on each section limbs that obtain is found the solution, obtained each deformation parameter of the revolving conic surface of each section limbs constantly;
4) use the OpenGL program to adopt parallel projection methods drawing three-dimensional people limbs model, use three spherosomes to represent three articulation points of limbs, the curved surface that two revolving conic surface equations that employing is tried to achieve are drawn out is represented upper and lower two sections limb skins, connects these three spherosomes and two curved surfaces formation people limbs models.
CN 200410089300 2004-12-09 2004-12-09 Human limb three-dimensional model building method based on labelling point Expired - Fee Related CN1271574C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200410089300 CN1271574C (en) 2004-12-09 2004-12-09 Human limb three-dimensional model building method based on labelling point

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200410089300 CN1271574C (en) 2004-12-09 2004-12-09 Human limb three-dimensional model building method based on labelling point

Publications (2)

Publication Number Publication Date
CN1617175A CN1617175A (en) 2005-05-18
CN1271574C true CN1271574C (en) 2006-08-23

Family

ID=34766190

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200410089300 Expired - Fee Related CN1271574C (en) 2004-12-09 2004-12-09 Human limb three-dimensional model building method based on labelling point

Country Status (1)

Country Link
CN (1) CN1271574C (en)

Families Citing this family (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100530244C (en) * 2005-06-21 2009-08-19 中国科学院计算技术研究所 Randomly topologically structured virtual role driving method based on skeleton
CN100348154C (en) * 2005-10-13 2007-11-14 上海交通大学 Method for non-intrusion measuring human hand and arm joint
CN100346357C (en) * 2006-01-19 2007-10-31 上海交通大学 Method for directly performing three-dimensional model transformation with three-dimensional bench marks
JP4148281B2 (en) * 2006-06-19 2008-09-10 ソニー株式会社 Motion capture device, motion capture method, and motion capture program
CN101490716B (en) * 2006-07-17 2012-07-04 皇家飞利浦电子股份有限公司 Efficient user interaction with polygonal meshes for medical image segmentation
CN100429677C (en) * 2006-10-19 2008-10-29 浙江大学 Forming and editing method for three dimension martial art actions based on draft driven by data
CN101271589B (en) * 2007-03-22 2011-07-06 中国科学院计算技术研究所 Three-dimensional mannequin joint center extraction method
CN100523727C (en) * 2007-04-10 2009-08-05 南京航空航天大学 Finger ring type video measuring finger location system and location method
CN101515374B (en) * 2008-02-20 2010-12-01 中国科学院自动化研究所 Individualized realistic virtual character modeling method based on images
CN101866497A (en) * 2010-06-18 2010-10-20 北京交通大学 Binocular stereo vision based intelligent three-dimensional human face rebuilding method and system
CN101894389A (en) * 2010-07-02 2010-11-24 北京理工大学 Rotation angle interpolation-based three-dimensional model torsional deformation method
CN101872490A (en) * 2010-07-02 2010-10-27 北京理工大学 Improved three-dimensional model deformation method under abrupt change in coordinate axis direction
CN102609987B (en) * 2012-01-09 2014-12-17 北京电子科技学院 Method and system for drawing curved surface by calculating all real roots and multiple numbers of zero-dimensional trigonometric polynomial system
CN105336000A (en) * 2015-12-09 2016-02-17 新疆华德软件科技有限公司 Virtual human limb modeling method based on hyperboloids of revolution
CN106600626B (en) * 2016-11-01 2020-07-31 中国科学院计算技术研究所 Three-dimensional human motion capture method and system
CN108583115B (en) * 2018-05-02 2021-06-18 阜阳师范学院 Manufacturing system of mouse early embryo model
CN109191593A (en) * 2018-08-27 2019-01-11 百度在线网络技术(北京)有限公司 Motion control method, device and the equipment of virtual three-dimensional model
CN112781521A (en) * 2020-12-11 2021-05-11 北京信息科技大学 Software operator shape recognition method based on visual markers
CN113925497B (en) * 2021-10-22 2023-09-15 吉林大学 Binocular vision measurement system-based automobile passenger riding posture extraction method

Also Published As

Publication number Publication date
CN1617175A (en) 2005-05-18

Similar Documents

Publication Publication Date Title
CN1266655C (en) Human limb three-dimensional model building method based on image cutline
CN1271574C (en) Human limb three-dimensional model building method based on labelling point
CN106826833B (en) Autonomous navigation robot system based on 3D (three-dimensional) stereoscopic perception technology
Lee et al. Constructing physics-based facial models of individuals
CN103679797A (en) Human limb modeling method adopting deformable models and virtual human model control platform
CN102592136B (en) Three-dimensional human face recognition method based on intermediate frequency information in geometry image
CN108805977A (en) A kind of face three-dimensional rebuilding method based on end-to-end convolutional neural networks
CN1753028A (en) Human limb three dimensional motion parameter estimation method based on skeleton
CN104616345B (en) Octree forest compression based three-dimensional voxel access method
CN1275205C (en) Human limbs three-dimensional model construction based on muscle volume invariance
CN101916454A (en) Method for reconstructing high-resolution human face based on grid deformation and continuous optimization
CN1326096C (en) Method for evaluating parameter of 3-D motion of human climbs based on model
Tonnesen Dynamically coupled particle systems for geometric modeling, reconstruction, and animation.
CN113012271B (en) Finger three-dimensional model texture mapping method based on UV (ultraviolet) mapping
CN114998515A (en) 3D human body self-supervision reconstruction method based on multi-view images
CN1828671A (en) Gridding texture mapping method in garment virtual display system based on image
CN101996415B (en) Three-dimensional modeling method for eyeball
CN113284249A (en) Multi-view three-dimensional human body reconstruction method and system based on graph neural network
CN105303611A (en) Virtual human limb modeling method based on revolution paraboloids
Wu et al. Reconstruction and physically-based animation of trees from static images
CN107895398A (en) With reference to the embossment edit methods of vision significance
CN105336000A (en) Virtual human limb modeling method based on hyperboloids of revolution
CN113554745B (en) Three-dimensional face reconstruction method based on image
CN110752004A (en) Voxel model-based respiratory characteristic characterization method
CN117237529A (en) 4D face data acquisition 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
C14 Grant of patent or utility model
GR01 Patent grant
C19 Lapse of patent right due to non-payment of the annual fee
CF01 Termination of patent right due to non-payment of annual fee