CN102467749A - Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions - Google Patents

Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions Download PDF

Info

Publication number
CN102467749A
CN102467749A CN2010105376605A CN201010537660A CN102467749A CN 102467749 A CN102467749 A CN 102467749A CN 2010105376605 A CN2010105376605 A CN 2010105376605A CN 201010537660 A CN201010537660 A CN 201010537660A CN 102467749 A CN102467749 A CN 102467749A
Authority
CN
China
Prior art keywords
human body
freedom
time
motion
degree
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
Application number
CN2010105376605A
Other languages
Chinese (zh)
Inventor
严峰
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 NEPOCH INFORMATION TECHNOLOGY Co Ltd
Original Assignee
SHANGHAI NEPOCH INFORMATION TECHNOLOGY Co Ltd
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 NEPOCH INFORMATION TECHNOLOGY Co Ltd filed Critical SHANGHAI NEPOCH INFORMATION TECHNOLOGY Co Ltd
Priority to CN2010105376605A priority Critical patent/CN102467749A/en
Publication of CN102467749A publication Critical patent/CN102467749A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Processing Or Creating Images (AREA)

Abstract

The invention relates to a three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions, belonging to the technical field of computer three-dimensional animations. The method comprises the steps of using a multi-rigid-body model with a skeleton structure as a geometrical model for three-dimensional virtual human body movement control, wherein vectors formed by all determined angle parameters indicating freedom in the geometrical model are taken as parameters for virtual human body static posture formal description; making human body static postures at two key frame moment, i.e. a minimum movement span and a maximum movement span; conducting equal or non-equal time sampling to the time sequence of movement, determining the angle offset of each time sampling point and obtaining the angle parameters of time sampling points in a movement process to be generated; and then generating virtual human body static postures at the moment, determining position parameters and generating the full movement process of a virtual human body in a three-dimensional space. By adopting the method, special video capture equipment is not required, the simplicity in making three-dimensional human body animations is improved and the method has the characteristics of convenience, high efficiency and high flexibility.

Description

Three-dimensional virtual human body movement generation method based on key frame and space-time restriction
Technical field
The invention belongs to the computer three-dimensional animation technical field, relate in particular to the 3 d human motion generation technique.
Background technology
In 3d computer animation, human body animation is one of research topic of tool difficulty and challenge.Equally, visual human's motion generation and control also are the research directions of visual human's technology.Yet because human body has more than 200 degree of freedom, the motion of Different Individual has different actions characteristics and emotional characteristics again.Simultaneously, the generation of virtual human body movement also will meet the basic law of human motion, has the mode of operation of simple controllable, needs to satisfy simultaneously the verisimilitude of motion and the validity of control.Therefore three-dimensional virtual human body movement generates, and comprises design, makes and the control each several part, and is very complicated.
At present, three-dimensional virtual human body movement generation method can be divided into two types: the one, and the movement generation method that the manikin that generates with computing machine is the basis, two are based on the movement generation method of capturing movement.
The movement generation method that computing machine generates manikin is to be the basis with the mathematical model, produces human action through methods such as key-frame animation technology, kinematicss.Key frame notion the earliest derives from the cartoon making in early stage Disney, and skilled animation teacher designs the key frame picture in the cartoon sheet, designs picture in the middle transition by general animation teacher then.
In the three-dimensional virtual human body movement of current computer generated, the key frame method was still used widely.This method may further comprise the steps:
1) confirms the geometric model of virtual human body.Can select multiple existing geometric model, as based on the multi-rigid model of skeletal structure, surface model, phantom type etc.
2) confirm the static posture describing mode of virtual human body.Promptly under this model, use which parameter just can generate different human body static postures.The parameter that the different geometric model uses is often different; Multi-rigid model like the skeleton structure; The angle parameters of expression joint freedom degrees that use (are selected the main activities joint usually; And 40~60 angle parameters of expression degree of freedom), or the location parameter in each joint is described the human body static posture.For surface model and phantom type, then richly to produce various posture effects through the dough sheet point set of control surface skin layer.
3) the human body static posture of many key frame times in the design movement process.For the multi-rigid model of skeletal structure, can adopt direct kinematics or the method for inverse kinematics generates the human body static posture.The method of direct kinematics is the degree of freedom parameter through given each joint, thereby confirms the position in each joint of human body and the orientation of end-effector.The method of inverse kinematics is then on the contrary, behind the position and direction through given end-effector, and the parameter value of Converse solved then each middle joint freedom degrees.
4), adopt interpolation arithmetic to generate the intermediate frame picture automatically by computing machine according to a plurality of key frame static postures that design.Interpolation method commonly used has two interpolation, cubic spline interpolation, Bezier spline interpolation or the like.
5) last, play each motion human body static posture constantly continuously, what watched is exactly the motion overall process of virtual human body.
This method simple, intuitive, but need a lot of key frames of design, could guarantee that the motion process after the interpolation accurately links up.The design process operation of each frame is heavy, and the motion generation for intermediate frame simultaneously lacks control again, is difficult to realize the trickle adjustment to certain action link that entire method lacks dirigibility and convenience.
In recent years, along with popularizing and popularization of motion capture system, increasing human movement capture system and commercial data base with height sense of reality appearred.This human motion generation method based on capturing movement then is that utilization sensor tracking human motion is obtained sense of reality three-dimensional motion data for basic, adopts data-driven version to generate personalized 3 d human motion.This method is used widely in fields such as motion simulation, video-game, video display stunts.Say so accurately a kind of motion of method for capturing movement reappears, but not the motion of real meaning generates.This method has higher human motion validity, but needs the capture device of special costliness, and true man's actual performance, has higher manpower and materials cost.In addition, the sense of reality exercise data that obtains according to this method is difficult to reused on the different human bodies, more is difficult to the componental movement details is edited and revised.
At present, more above-mentioned three-dimensional virtual human body movement generation methods have formed in the making that business software is used to animation product, like Maya, and 3D Max, Kaydara Motion Builder, SoftImage, Poser.Deng.Use above-mentioned software to accomplish in the process of human motion making, complex operation, inefficiency need the user to have higher experience.Simultaneously, also all there are defectives such as low, intelligent not high, a large amount of manual interventions of needs of data reusing property in these softwares.
The prior art introduction relevant with the inventive method is following:
The space-time restriction method: in virtual environment, the restriction that often applies the initial position of barrier, human motion and target location etc. to human motion is referred to as constraint.It both can be on the space, also can be on the whole motion sequence, therefore was called space-time restriction again.The method of space-time restriction is commonly used to carry out in the research fields such as motion editing, motion are redirected.
Weighted least-squares support vector method: SVMs (SVM) has complete Statistical Learning Theory basis, and foot has outstanding study and promotes performance small sample, pattern-recognition and Function Estimation non-linear, the dimensions data point especially.Least square method supporting vector machine (LS-SVM) converts SVM into find the solution system of linear equations problem to quadratic programming problem as the expansion of SVM, than the SVM method better pace of learning and simple in structure is arranged.But LS-SVM has lost loose type and the robustness of SVM, in order to overcome this problem, has proposed weighted least-squares support vector machine method (WLS-SVM), is used for solving problems such as pattern-recognition and Function Estimation more.
Summary of the invention
The present invention is the shortcoming and defect that overcomes above-mentioned prior art, has proposed a kind of three-dimensional virtual human body action generation method based on key frame and space-time restriction.This method need not special video capture device, has improved the simplification that 3 D human body animation is made, and has convenience, efficient, flexible characteristic.
Above-mentioned purpose of the present invention realizes through following technical scheme.
Three-dimensional virtual human body movement generation method based on key frame and space-time restriction of the present invention may further comprise the steps:
1) at first, uses the geometric model of the multi-rigid model of skeletal structure, and confirm the move number of needed joint and degree of freedom thereof of this geometric model of control as three-dimensional virtual human body movement control;
2) represent the vector that the angle parameter of degree of freedom constitutes by all that confirm in said visual human's geometric model, as the parameter of virtual human body static posture formalized description; The static posture of each virtual human body, the angle parameter value of corresponding one group of degree of freedom;
3) to human action to be generated, according to the human body static posture of direct kinematics or inverse kinematics making minimum movement span and these two key frame times of largest motion span;
4) time series of virtual human body movement overall process to be generated is carried out the constant duration sampling, this motion process is divided into the motor segment that the time equates; Judging whether needs to increase sampled point, if then according to sampled point number of confirming and the position on time series thereof, proceeds non-constant duration sampling, otherwise directly carries out next step;
5) employing is based on the method for space-time restriction; To each time-sampling point of confirming in the step 4); Through regulate each sampled point constantly the angle parameter of the degree of freedom of human body static posture (the concrete practice is: with the angle parameter of the degree of freedom of minimum movement span key frame time as minimum value 0 with respect to the side-play amount of minimum or largest motion span key frame time; With the angle parameter of the degree of freedom of largest motion span key frame time as maximal value 1; Then between 0~1, regulate the angle parameter of the degree of freedom of each time-sampling point :), draw the angle parameter of each time-sampling point in the motion process to be generated; To all time points that constant duration sampling in the step 4) obtains, judge whether to need to increase intermediate frame, if then change step 6), otherwise change step 7); Time point for non-constant duration sampling in the step 4) obtains changes step 6);
The angle parameter of all sampling instants that 6) step 5) generated carries out match and (can adopt the weighted least-squares support level and smooth more smooth to moving generation picture;
7) judge whether virtual human body movement to be generated relative position in three dimensions changes; If not; Then virtual human body is positioned on the initial position in three dimensions all the time; Directly change step 8): if, then provide the different location parameters of visual human in three dimensions constantly, change step 8):
8) according to step 5) and 6) angle parameter and the direct kinematics method of each degree of freedom constantly of obtaining; Generate the static posture of this moment virtual human body; And according to the location parameter of this moment visual human in three dimensions that provides in the step 7); Specified location in virtual world shows, through continuous broadcast motion sequence, finally generates the motion overall process of virtual human body in three dimensions.
Characteristics of the present invention and effect:
Method of the present invention, two extreme values selecting motion span maximum in the action to be designed are constantly as key frame, and other motions angle parameters constantly all drop within this span scope.For these two key frames, the virtual human body static posture in this moment need be set by the user.For other intermediate frames; Adopt the mode of space-time restriction; To the time point design of Interaction angle parameter of sampling offset with respect to minimum (or maximum) motion span key frame, simulate the angle change curve through the weighted least-squares SVMs, accomplish the angle interpolation.At last, through in the virtual three dimensional space position of user's appointment, playing different human body static postures constantly continuously, promptly generate the overall process that the visual human moves in virtual three dimensional space.
Method of the present invention needs the number of key frames of setting few, and the animation teacher only needs two key frames design virtual human body static postures of action to be generated are got final product.And for the motion process of centre, the user only need confirm the sampling time point, regulates the offset on each time point, and provides the visual human with respect to three-dimensional location parameter, just can in real time, automatically generate the overall process of virtual human body movement.This method need not to regulate other special parameters, also need not the puzzlement of various mathematical formulaes and experience, can be fully comes designer's body action according to user's needs.This method frees the user for the user provides simple, convenient, fast, action design platform intuitively from loaded down with trivial details design.Simultaneously, in the setting up procedure of static posture and dynamic motion, can watch 3-D effect in real time, in all directions.In addition,, can make amendment easily, edit, preserve and reuse, greatly reduce the action design cost with the static posture and the dynamic motion of this method design virtual human body.
Description of drawings
Fig. 1 is according to virtual human body movement generation method process flow diagram of the present invention:
Fig. 2 is the setting of the angle parameter side-play amount of the degree of freedom under the right hand constant duration sampled point of waving to move;
Fig. 3 be beat on the human body do in the angle parameter and the angle parameter matched curve thereof of sampled point degree of freedom constantly;
Fig. 4 be beat on the human body do in sampled point location parameter and the location parameter matched curve thereof of virtual human body in three dimensions constantly:
Embodiment
The three-dimensional virtual human body movement generation method based on key frame and space-time restriction that the present invention proposes combines accompanying drawing and embodiment to specify as follows:
The present invention is based on the three-dimensional virtual human body movement generation method of key frame and space-time restriction, as shown in Figure 1, may further comprise the steps:
1) at first, use the geometric model of the multi-rigid model of skeletal structure as three-dimensional virtual human body movement control, and confirm the move number n of needed joint and degree of freedom thereof of this geometric model of control, n is a positive integer;
2) represent the vector [θ that the angle parameter of degree of freedom constitutes by all that confirm in said visual human's geometric model 1, θ 2... θ n], as the parameter of virtual human body static posture formalized description; The static posture of each virtual human body, the angle parameter value of corresponding one group of degree of freedom;
3) to human action to be generated; The human body static posture of making minimum movement span and these two key frame times of largest motion span according to direct kinematics or inverse kinematics is (in the human motion process; The movement locus at human motion position can form a plane in the space; In the moment of two boundary positions on this space plane, promptly be chosen for minimum and largest motion span key frame; Other movement locus constantly all are included in by minimum, largest motion span key frame and form in the space plane);
4) time series of virtual human body movement overall process to be generated is carried out the constant duration sampling; This motion process is divided into the motor segment that the time equates: judge whether to need to increase sampled point; If; Then, proceed non-constant duration sampling, otherwise directly carry out next step according to sampled point number of confirming and the position on time series thereof;
5) employing is based on the method for space-time restriction; To each time-sampling point of confirming in the step 4); (concrete control method is the side-play amount of the angle parameter of the degree of freedom through regulating this moment human body static posture: with the angle parameter of the degree of freedom of minimum movement span key frame time as minimum value 0; With the angle parameter of the degree of freedom of largest motion span key frame time as maximal value 1; Then between 0~1, regulate the angle parameter of the degree of freedom of each time-sampling point), draw the angle parameter of each time-sampling point in the motion process to be generated; All time points that sampling obtains to the constant duration in the step 4) judge whether to need to increase intermediate frame, if then change step 6), otherwise change step 7); Time point for non-constant duration sampling in the step 4) obtains changes step 6);
The angle parameter of all sampling instants that 6) step 5) generated carries out match (approximating method can adopt the weighted least-squares SVMs to realize); Generate a human motion at all continuous angle parameter time changing curves constantly, make that the generation picture of human motion is level and smooth more smooth;
7) judge whether virtual human body movement to be generated relative position in three dimensions changes, if not, then virtual human body is positioned at initial position [x all the time in three dimensions 0, y 0, z 0] on, directly change step 8); If then provide the different location parameter [xs of visual human in three dimensions constantly i, y i, z i], change step 8):
8) according to step 5) and 6) angle parameter and the direct kinematics method of each degree of freedom constantly of obtaining; Generate the static posture of this moment virtual human body; And according to the location parameter of this moment visual human in three dimensions that provides in the step 7); Specified location in virtual world shows, through continuous broadcast motion sequence, finally generates the motion overall process of virtual human body in three dimensions.
Two kinds of embodiment of the design of the inventive method explain as follows:
Embodiment 1: wave to move the virtual human body movement as to be generated with the right hand.This action of waving, right hand arm is positioned at middle at first, brandishes minimum movement span place left earlier, brandishes largest motion span place more to the right, puts left more at last and is back to the centre position, accomplishes the swing overall process.
The above-mentioned three-dimensional virtual human body movement generation method of present embodiment may further comprise the steps:
1) at first; Use the geometric model of the multi-rigid model of skeletal structure as three-dimensional virtual human body movement control; And confirm the move number of needed joint and degree of freedom thereof of this geometric model of control: because human body has nearly more than 200 degree of freedom; Verisimilitude and controllability for the balance virtual human body movement; Present embodiment has been chosen 45 degree of freedom (joint of not considering trick is divided) in 18 main joints, comprises sacrum osteoarthrosis (6 degree of freedom), waist joint (3 degree of freedom), chest joint (3 degree of freedom), neck joint (3 degree of freedom), a left side (right side) breastbone clavicle joint (2 degree of freedom), a left side (right side) shoulder joint (3 degree of freedom), a left side (right side) elbow joint (2 degree of freedom), a left side (right side) wrist joint (2 degree of freedom), a left side (right side) hip joint (3 degree of freedom), a left side (right side) knee joint (2 degree of freedom), a left side (right side) ankle-joint (1 degree of freedom).The barycenter of virtual human body is arranged on the rumpbone joint, and centroid position has then represented the visual human in three-dimensional position.The six-freedom degree of sacrum osteoarthrosis has been confirmed virtual human body and the position relation that belongs in the three dimensions, three angle parameter [θ 1, θ 2, θ 3] confirmed virtual human body in three dimensions towards, three location parameter [x 0, y 0, z 0] confirmed that virtual human body is in three-dimensional position; These 45 degree of freedom comprise the angle parameter of 42 degree of freedom and the location parameter of 3 degree of freedom;
2) vector [θ that constitutes by 42 angle parameters of the expression degree of freedom of confirming in above-mentioned visual human's geometric model 1, θ 2... θ 42], as the parameter of virtual human body static posture formalized description:
3) right hand to be generated is waved to move; Make the human body static posture of minimum movement span and these two key frame times of largest motion span according to direct kinematics: to be in the setting state of high order end be minimum movement span key frame (angle parameter of this moment degree of freedom is as minimum value 0) to arm in waving to move, and the setting state that arm is in low order end is largest motion key frame (angle parameter of this moment degree of freedom is as maximal value 1):
The overall process of 4) right hand to be generated being waved to move is carried out the constant duration sampling, chooses 11 sampled points, and motion process is divided into the motor segment (present embodiment need not to increase other sampled points) that 10 times equate;
5) according to this action of waving, the original position of arm is positioned at middle, brandishes earlier minimum movement span place left; Brandish largest motion span place more to the right; Put left more at last and be back to centre position this action of end, adopt method, 11 time-sampling points confirming in the step 4) based on space-time restriction; Between 0~1, regulate each angle parameter side-play amount of the degree of freedom of human body static posture constantly, draw each sampled point angle parameter constantly in the motion process; Being provided with of the angle parameter of concrete sampling instant is as shown in Figure 2, and the pore among the figure is each sampled point angle parameter constantly; These 11 time-sampling points have enough been controlled the motion process of waving to move, and present embodiment does not need to increase intermediate frame yet again;
6) be the action that the right hand is brandished in the present embodiment, just right arm is in motion, and virtual human body does not change with respect to three-dimensional position, is in [x all the time 0, y 0, z 0] locate;
7) angle parameter to each degree of freedom constantly of obtaining in the step 5) uses the direct kinematics method, generates the static posture of this moment virtual human body, through continuous broadcast motion sequence, finally generates the right hand of virtual human body in three dimensions and brandishes action.
Embodiment 2: do as virtual human body movement to be generated to beat on the human body.This motion process is: when jumping in the preparation, and the slight semi-crouch of human body, behind the slightly downward deep-knee-bend, upwards take-off, liftoff, hang in the sky moments later, land, the buffering of squatting down slightly, last whole body uprightly stands in ground, accomplishes whole going up and jumps motion.
The above-mentioned three-dimensional virtual human body movement generation method of present embodiment may further comprise the steps:
1) at first, use the geometric model of the multi-rigid model of skeletal structure as three-dimensional virtual human body movement control, and confirm this geometric model of control move needed joint with and the number of degree of freedom; Identical with embodiment 1, chosen 18 main joints and 45 degree of freedom thereof of controlling human motion, comprise the angle parameter of 42 degree of freedom and the location parameter of 3 degree of freedom;
2) vector [θ that constitutes by 42 angle parameters of the expression degree of freedom of confirming in above-mentioned visual human's geometric model 1, θ 2... θ 42], as the parameter of virtual human body static posture formalized description;
3), make the human body static posture of minimum movement span and these two key frame times of largest motion span according to direct kinematics to the work of beating on the human body to be generated; On the human body in the jump process; The lower limb crooked at utmost state of going down on one's knees is set to minimum movement span key frame (angle parameter of this moment degree of freedom is as minimum value 0), and the state that the human body whole body is kept upright is largest motion span key frame (angle parameter of this moment degree of freedom is as maximal value 1).
4) overall process of doing beating on the virtual human body to be generated is carried out the constant duration sampling; Choose 4 sampled points; Motion process is divided the motor segment that 3 times equate; Fall the back process on ground of the take-off process of the first represents whole body from initial semi-crouch state to built on stilts, the process that the second represents whole body uprightly hangs in the sky, the 3rd represents; For finer description detailed process, present embodiment has increased by 6 time-sampling points: first sampled point is in the process of first section take-off, has added the at utmost sampled point of state of squatting down.Ensuing four sampled points are added in second section whole body and stand in the aerial process; Last point is in the process on the 3rd paragraph ground, has added the sampled point that is buffered to the semi-crouch state after is landed; So just constituted the non-constant duration sampling that 10 sampled points are arranged;
5) employing is based on the method for space-time restriction; On 10 time-sampling points confirming in the step 4), the angle parameter side-play amount of the degree of freedom through regulating this moment human body static posture draws the angle parameter in each sampled point moment in the motion process: first sampled point moment; The slight semi-crouch of human body; Prepare take-off: in second sampled point moment, human body continues to squat down downwards at utmost, promptly is minimum movement span key frame constantly; The 3rd sampled point, human body will be soon liftoff, and whole body is kept upright: ensuing four sampled points, keep whole body upright, hang in the sky; The 8th sampled point, human body just lands, and whole body is kept upright; The 9th sampled point, human body land and cushion back semi-crouch slightly: last sampled point, whole body uprightly stand on the ground, the work of beating in the end;
The angle parameter of 10 sampling instants that 6) step 5) generated; Carry out match with the weighted least-squares support vector machine method; Generate a human motion at all continuous angle parameter and change of time curves constantly, as shown in Figure 3, among the figure; Pore among the figure is each sampled point angle parameter constantly, and curve is with the curve after the angle parameter match; In carrying out match; Because angle parameter minimum, largest motion span key frame is set, and is more accurate, therefore gives higher weight; For example be set to 0.1; And, can give lower weight for the angle parameter on the other times sampled point in the motion process, for example be set to 0.02; With this weighted least-squares support vector machine method, can match obtain each level and smooth joint angles change curve, the variation of its first order derivative angular velocity also is continuous, meets the basic law of human motion:
7) judge whether the relative position of doing in three dimensions of beating on the virtual human body to be generated changes.Because human body is squatted down and gone up and jump under the unsettled state, variation has all taken place in the mass center of human body position, and promptly variation has taken place the relative position of virtual human body in three dimensions, therefore need provide the different location parameter [xs of virtual human body in three dimensions constantly i, y i, z i]; On this human body, jump in the motion, the position of mass center of human body about taken place to move on (y axle) orientation, and not about be moved on the direction of (x axle) and front and back (z axle), so have only the value of y that variation has taken place; Concrete position of human body and time-varying relationship are as shown in Figure 4, and pore is a sampled point location parameter constantly among the figure, and curve is the curve after the centroid position match.
8) angle parameter to each degree of freedom constantly of obtaining in the step 6) uses the direct kinematics method; Generate the static posture of this moment virtual human body; And according to the location parameter of this moment visual human in three dimensions that provides in the step 7); Specified location in virtual world shows, through continuous broadcast motion sequence, finally generate virtual human body in three dimensions on the work of beating.

Claims (3)

1. based on the three-dimensional virtual human body movement generation method of key frame and space-time restriction, it is characterized in that, may further comprise the steps:
1) at first, uses the geometric model of the multi-rigid model of skeletal structure, and confirm the move number of needed joint and degree of freedom thereof of this geometric model of control as three-dimensional virtual human body movement control;
2) vector that is made up of the angle parameter of all expression degree of freedom of confirming in said visual human's geometric model is as the parameter of virtual human body static posture formalized description; The static posture of each virtual human body, the angle parameter value of corresponding one group of degree of freedom;
3) to human action to be generated, according to the human body static posture of direct kinematics or inverse kinematics making minimum movement span and these two key frame times of largest motion span;
4) time series of virtual human body movement overall process to be generated is carried out the constant duration sampling, this motion process is divided into the motor segment that the time equates; Judging whether needs to increase sampled point, if then according to sampled point number of confirming and the position on time series thereof, proceeds non-constant duration sampling, otherwise directly carries out next step;
5) employing is based on the method for space-time restriction; To each time-sampling point of confirming in the step 4); The angle parameter of the degree of freedom through regulating this moment human body static posture draws the angle parameter of each time-sampling point in the motion process to be generated with respect to the side-play amount of minimum or largest motion span key frame time; To all time points that constant duration sampling in the step 4) obtains, judge whether to need to increase intermediate frame, if then change step 6), otherwise change step 7): the time point for non-constant duration sampling in the step 4) obtains, change step 6);
The angle parameter of all sampling instants that 6) step 5) generated carries out match, generates a human motion at all continuous angle parameter time changing curves constantly, the level and smooth more smoothness of generation picture that makes human motion;
7) judge whether virtual human body movement to be generated relative position in three dimensions changes, if not, then virtual human body is positioned in three dimensions on the initial position all the time, directly changes step 8); If, then provide the different location parameters of visual human in three dimensions constantly, change step 8);
8) according to the respectively angle parameter and the direct kinematics method of the degree of freedom in the moment that obtain; Generate the static posture of this moment virtual human body; And according to the location parameter of this moment visual human in three dimensions that provides in the step 7); Specified location in virtual world shows, through continuous broadcast motion sequence, finally generates the motion overall process of virtual human body in three dimensions.
2. method according to claim 1; Each sampled point of the adjusting of said step angle parameter of the degree of freedom of human body static posture constantly with respect to the method for the side-play amount of minimum or largest motion span key frame time is: with the angle parameter of the degree of freedom of minimum movement span key frame time as minimum value 0; As maximal value 1, then between 0~1, regulate the angle parameter of the degree of freedom of each time-sampling point with the angle parameter of the degree of freedom of largest motion span key frame time.
3. method according to claim 1 is characterized in that, the approximating method in the said step 6) adopts the weighted least-squares SVMs to realize.
CN2010105376605A 2010-11-10 2010-11-10 Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions Pending CN102467749A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2010105376605A CN102467749A (en) 2010-11-10 2010-11-10 Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2010105376605A CN102467749A (en) 2010-11-10 2010-11-10 Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions

Publications (1)

Publication Number Publication Date
CN102467749A true CN102467749A (en) 2012-05-23

Family

ID=46071355

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2010105376605A Pending CN102467749A (en) 2010-11-10 2010-11-10 Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions

Country Status (1)

Country Link
CN (1) CN102467749A (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103824316A (en) * 2014-03-26 2014-05-28 广州博冠信息科技有限公司 Method and equipment for generating action pictures for object
CN104112288A (en) * 2014-07-24 2014-10-22 无锡梵天信息技术股份有限公司 Three-dimensional skeleton model based action transition achieving method
CN105824432A (en) * 2016-06-14 2016-08-03 上海锡月科技有限公司 Motion capturing system
CN106125908A (en) * 2016-06-14 2016-11-16 上海锡月科技有限公司 A kind of motion capture calibration system
CN106125909A (en) * 2016-06-14 2016-11-16 上海锡月科技有限公司 A kind of motion capture system for training
CN107016715A (en) * 2017-03-06 2017-08-04 浙江大学 A kind of signal map animation method based on physical analogy
GB2574460A (en) * 2018-06-07 2019-12-11 Frontier Dev Plc Computer Animation Method and Apparatus
CN112839184A (en) * 2020-12-31 2021-05-25 深圳追一科技有限公司 Image processing method, image processing device, electronic equipment and storage medium

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103824316A (en) * 2014-03-26 2014-05-28 广州博冠信息科技有限公司 Method and equipment for generating action pictures for object
CN103824316B (en) * 2014-03-26 2017-05-10 广州博冠信息科技有限公司 Method and equipment for generating action pictures for object
CN104112288A (en) * 2014-07-24 2014-10-22 无锡梵天信息技术股份有限公司 Three-dimensional skeleton model based action transition achieving method
CN104112288B (en) * 2014-07-24 2017-05-10 无锡梵天信息技术股份有限公司 Three-dimensional skeleton model based action transition achieving method
CN105824432A (en) * 2016-06-14 2016-08-03 上海锡月科技有限公司 Motion capturing system
CN106125908A (en) * 2016-06-14 2016-11-16 上海锡月科技有限公司 A kind of motion capture calibration system
CN106125909A (en) * 2016-06-14 2016-11-16 上海锡月科技有限公司 A kind of motion capture system for training
CN107016715A (en) * 2017-03-06 2017-08-04 浙江大学 A kind of signal map animation method based on physical analogy
CN107016715B (en) * 2017-03-06 2020-05-08 浙江大学 Schematic map animation method based on physical simulation
GB2574460A (en) * 2018-06-07 2019-12-11 Frontier Dev Plc Computer Animation Method and Apparatus
GB2574460B (en) * 2018-06-07 2021-05-19 Frontier Dev Plc Computer Animation Method and Apparatus
CN112839184A (en) * 2020-12-31 2021-05-25 深圳追一科技有限公司 Image processing method, image processing device, electronic equipment and storage medium

Similar Documents

Publication Publication Date Title
CN100557639C (en) Three-dimensional virtual human body movement generation method based on key frame and space-time restriction
CN102467749A (en) Three-dimensional virtual human body movement generation method based on key frames and spatiotemporal restrictions
US11928765B2 (en) Animation implementation method and apparatus, electronic device, and storage medium
CN110930483B (en) Role control method, model training method and related device
JP7407919B2 (en) Video processing method, video processing device, computer program and electronic equipment
CN106078752B (en) A kind of anthropomorphic robot human body behavior imitation method based on Kinect
CN107330967B (en) Rider motion posture capturing and three-dimensional reconstruction system based on inertial sensing technology
CN101579238B (en) Human motion capture three dimensional playback system and method thereof
JP3579148B2 (en) Method and system for operating a plurality of motion units to generate articulated body motion
CN102708582B (en) Character movement redirecting method for heterogeneous topology
US6909431B1 (en) Position and shape control for cloth and soft body animation
CN103440037B (en) Real-time interaction virtual human body motion control method based on limited input information
CN107598897A (en) A kind of method of humanoid robot gait's planning based on human body teaching
CN107833271A (en) A kind of bone reorientation method and device based on Kinect
CN100440257C (en) 3-D visualising method for virtual crowd motion
CN102306386B (en) Method for quickly constructing third dimension tree model from single tree image
KR20110062044A (en) Apparatus and method for generating motion based on dynamics
WO2023088041A1 (en) Method and apparatus for generating walking animation of virtual character, and device and storage medium
CN107610208A (en) The movement simulating method of cartoon role under a kind of granule medium environment
CN107578461A (en) A kind of three-dimensional virtual human body physical motion generation method based on subspace screening
CN114742952A (en) Three-dimensional garment simulation method and device, terminal equipment and storage medium
Tao et al. Human modeling and real-time motion reconstruction for micro-sensor motion capture
Beacco et al. Footstep parameterized motion blending using barycentric coordinates
US20120223953A1 (en) Kinematic Engine for Adaptive Locomotive Control in Computer Simulations
Rabbani et al. PhysIK: Physically Plausible and Intuitive Keyframing.

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20120523