三维可变形物体的三点匹配策略  

Three-point matching strategy for 3D deformable shapes

作　　者：潘翔[1] 章国栋[1] 周春燕[2] 陈启华[3] 

机构地区：[1]浙江工业大学计算机科学与技术学院,浙江杭州310023 [2]杭州市公安局信息中心,浙江杭州310009 [3]浙江工业大学机械工程学院,浙江杭州310032

出　　处：《浙江工业大学学报》2013年第5期539-544,共6页Journal of Zhejiang University of Technology

基　　金：浙江省自然科学基金资助项目(Y1110780);浙江省科技厅优先主题资助项目(2009C11038)

摘　　要：针对不同姿态下的三维可变形物体特征点匹配问题,提出了一种基于热核信号特征和测地距离的三点匹配策略,从而可以为后续的紧密匹配奠定基础.该方法主要由以下三步完成,首先提取出三维可变形物体的外部特征点;其次,对外部特征点采用热核信号定义局部特征;最后,结合局部特征描述符和特征点之间的测地距离进行最优匹配搜索,得到正确的匹配结果.实验结果表明:该方法能很好地实现对三维可变形物体的稀疏匹配.This paper addresses the problem about the feature points matching of 3D deformable objects in different poses. It proposes a new algorithm called three-point matching strategy, which based on heat kernel signature （HKS） and geodesic distance. It includes three steps. Firstly, the algorithm extracts the external feature points from input 3D meshes. Secondly, the algorithm uses HKS to define the local feature for external feature points. Finally, the algorithm searches the optimal Geodesic distance matching between local feature descriptor and geodesic distance of external points. The experiment shows that the proposed algorithm can well realize the sparse matching of 3D deformable objects.

关 键 词：特征点提取 热核信号 测地距离 稀疏匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]