基于双层候选集的快速人体-服装冲突检测  

An Efficient Cloth-Body Collision Detection Method Based on Bi-Hierarchy Candidate Sets

作　　者：朱小龙[1] 毛天露[1] 夏时洪[1] 于勇[1] 

机构地区：[1]中国科学院计算技术研究所虚拟现实技术实验室

出　　处：《计算机辅助设计与图形学学报》2008年第7期925-932,共8页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60403042);国家自然科学基金重点项目(60533070);国家“九七三”重点基础研究发展规划项目(2002CB312104);北京市自然科学基金(4062032)

摘　　要：冲突检测是人体-服装动画研究的核心技术之一,已有的层次冲突检测方法在处理服装变形时需要对层次节点进行大量更新,计算复杂度高,难以满足实时检测的要求.针对人体和服装的特殊位置关系和运动规律,提出一种基于双层候选集的实时冲突检测方法.首先为服装模型的每个顶点建立可能发生位置冲突的人体模型面片候选集,利用候选集的母子双层结构剔除绝大多数不可能发生冲突的图元.该候选集相对于已有的层次方法具有更快的更新速度且具有较好的独立性,便于并行处理;在确定可能冲突的图元后,采用GPU片段程序来加速大规模图元间的相交测试.实验结果表明,该方法检测迅速准确,对一般人体-服装模型(10000面以下)能达到30～75帧/s的实时处理.Cloth-body collision detection is essential to study apparel animation, but the developed hierarchical methods are hard to meet the requirement of real-time detection since clothes belong to deformable objects, which require frequent hierarchy updates. According to the special geometric position of cloth and body, a fast collision detection method based on bi-hierarchy candidate sets is proposed in this paper. First of all, the candidate sets are built for each cloth vertex to eliminate most potentially non-colliding primitives. The candidate sets have the advantage of high update efficiency and computation independence, suitable for parallel processing. Then in the following step, GPU fragment program is used to accelerate the intersection tests of potentially colliding primitives. Experimental result demonstrates that the proposed method shows its advantages in algorithm＇s accuracy and efficiency for achieving real-time collision detection for general cloth models.

关 键 词：虚拟现实 服装动画 冲突检测 双层候选集 GPU 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]