一种基于Kinect的角色骨骼动画方法  

A role skeletal animation method based on Kinect

作　　者：李红波[1] 冉光勇[1] 吴渝[1] 丁林建[1] 

机构地区：[1]重庆邮电大学计算机科学与技术学院网络智能研究所,重庆400065

出　　处：《重庆邮电大学学报（自然科学版）》2014年第4期530-535,共6页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金："核高基"重大专项(2009ZX01038-002-002-2);科技部"原创动漫软件开发技术人才"计划扶持(2009-593)~~

摘　　要：针对当前骨骼动画制作方法存在实时性差、抗光照干扰能力弱、角色模型运动误差大的问题,提出了一种新的骨骼动画方法。采用微软Kinect设备获取人体三维骨骼坐标数据,用改进的D-H(denavit-hartenberg)描述法在关节点处建立虚拟坐标系,对捕捉对象单帧姿态进行抽象空间位形描述,再根据前后帧位形数据求解骨骼矩阵,对根节点进行位移调整形成骨骼动画。实验结果表明,该方法实时性与抗光照干扰能力强,角色动作与捕捉对象动作之间的相似度高。Since the current skeletal animation production methods have problems of poor real-time and weak anti-interference ability for illumination, as well as big motion error in role model, a new skeletal animation method is proposed. Firstly, the three-dimensional coordinate data of human skeleton is obtained by using Microsoft＇ s Kinect. Then an improved D- H（denavit-hartenberg） description method is used to establish the virtual coordinate system at the joints, and the single frame posture of the captured object is described by use of the configuration in the abstract space. Later, the bone matrix is solved based on the configuration data of front and rear frames. Finally, the displacement of the root node is adjusted to form the skeletal animation. Experimental results show that the proposed method has good real-time, anti-interference ability for illumination, and high similarity between the motion of the role and the captured object.

关 键 词：骨骼动画 实时 KINECT 骨骼数据 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]