Hierarchical Matching of 3D Shape Based on Heat Kernel Signature

3Ddense matching is an important problem in computer vision.In this paper, we address the issues of robustness and efficiency in dense matching for 3Dshapes.We present a new strategy on choosing points and a hierarchical matching method upon heat kernel signature (HKS) robust to noise.The matching starts from a small subset of stable feature points which is significant to describe the topological characteristics of 3Dshapes, and performs from the coarse to fine according to the entropy on HKS features.The new hierarchical strategy renders fast matching, and meanwhile prunes those points of less discrimination by fusion method or adds a few “helpful points”by farthest point sampling for robust and accurate matching.Experimental results on the TOSCA data set demonstrate that the proposed method is more suitable for practical applications than state-of-the-art methods.