Font Size: a A A

Research And Application Of Fast Simulation Technology For Real People

Posted on:2020-07-13Degree:MasterType:Thesis
Country:ChinaCandidate:J W RenFull Text:PDF
GTID:2428330575476064Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
The rapid simulation technology of real people has important research value and commercial value in virtual game,3d film production,military virtual exercise and virtual community.In this paper,the quick skeleton extraction and skin algorithm of the character model are studied.The traditional Laplace contraction method and L1 measurement method have the problems of uneven skeleton shrinkage and poor effect on the spherical model.LBS algorithm has the side effect of collapse because the skin cannot be effectively supported.DQS algorithm has the defect of joint expansion.In this paper,point cloud contraction and skeleton extraction are performed simultaneously.The local Laplacian point cloud contraction method prevents uneven skeleton contraction.The contraction radius of skeleton extraction is an important criterion to determine the success of skeleton extraction.Combined with the central idea of L1 measure,this paper proposes a skeleton extraction condition discrimination method based on dynamic bounding box,and USES the relation between the centerline of bounding box and the point to determine the termination condition of skeleton extraction.In this paper,first of all,all the bounding boxes of the cloud to be fixed after contraction are calculated,and the Central Line perpendicular to the contraction direction of the bounding box is calculated.A point is selected from the bounding box,and the distance between the point and the Central Line is calculated.When the distance reaches the contraction condition,the point can be determined as the extracted skeleton point.Then,the skeleton point was removed from the pending point,and all remaining points that were not extracted as skeleton points were subjected to the next iterative contraction calculation,and the enclosing box was recalculated.The above process was repeated until all the points had been shrunk.After all skeleton extraction is completed,all skeleton branches are connected to obtain a complete skeleton system,and the location of joints in the skeleton is determined according to the geodesic distance of the skeleton and the proportion of human joints.Experiments show that the accuracy of the extraction results of the same model in this paper is higher than the traditional Laplace method and L1 measure method,and the curve skeleton obtained by other local point cloud contraction methods is smoother.This paper changes the judging condition and expression form of weight,improves the misjudgment of weight,and helps solve the disadvantage of simple linear skin.In LBS algorithm on the basis of the improved algorithm of LBS will improve the weight of the binding method combined with joint location information,according to the current frame system connection joints presented two bones in the Angle of the relationship between traditional LBS algorithm based on joint location information and the improved LBS algorithm dynamically switch between,to the greatest extent to meet the requirements of authenticity and efficiency.The experimental results show that the modified weight binding method and dynamic LBS algorithm switching strategy based on joint position information can effectively solve the problem of skin deformation and distortion in bone animation.
Keywords/Search Tags:skeletal animation, skeletal extraction, motion simulation, skin deformation
PDF Full Text Request
Related items