Research on Shape Similarity and Shape Matching

Abstract

A triangular mesh model matching method by combination of principle component analysis (PCA) and ICP algorithm is presented. Firstly, two 3D models are initially located by using PCA method. Secondly, ICP algorithm is applied to do fine adjustment and calibration. Finally, quantitatively evaluate the similarity between two models by using difference sum of squares function. This kind of two-stage location and calibration method, not only makes all models under the same conditions of calibration, but also avoids happening the situation, which is caused by excessive displacement, that may lead to the failure of ICP calibration. In the end, we give some examples that estimate similarity of models by using the method in this paper.