基于扩散距离和MDS的非刚性模型相似性分析

在非刚性模型相似性分析方法中, 基于测地距离的等距嵌入方法对模型拓扑变化非常敏感, 对于有局部拓扑变化的完全相似的模型也无法得到正确结果。为了弥补这一不足, 提高非刚性模型相似性分析的准确性, 将扩散距离与多维尺度分析(MDS)相结合, 计算非刚性模型的等距嵌入模型, 再利用迭代最近点(ICP)算法通过计算嵌入模型的相似性来实现原始非刚性模型的相似性分析。实例证明该方法对含有拓扑变化的非刚性模型可以得到理想的相似性分析结果。

Approach of similarity analysis for non-rigid models combining diffusion distance with MDS

Among the similarity analysis methods for the non-rigid models, ones based on geodesic distances are quite sensitive to topology noise and can be problematic in cases where object's topology is changed locally. In order to solve the problem, this paper proposed a new method based on diffusion distance. It combined diffusion distance with multidimensional scale (MDS) method to compute embedding models in isometric way, and then, used iterative closest point (ICP) method to compute the similarity of embedded models, which was considered as the similarity of original non-rigid models. The method was tested on some examples. The results show that the new method is stable and robust to non-rigid models with local topological changes.