欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
冲突检测是人体一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，对一般人体一服装模型（10000面以下）能达到30-75帧／s的实时处理．
英文摘要：
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.
同期刊论文项目
同项目期刊论文
期刊信息
位置：