A 3D Model Similarity Matching Algorithm Based on Spherical Harmonics Descriptor

A 3D model similarity matching algorithm based on spherical harmonics descriptor is proposed in this paper.We first use recursive bisection to partition 3D models,and then we get spherical images by mapping partitions into concentric spheres,last we calculate spherical harmonics descriptors of spherical images to get feature binary trees of 3D models.We can get the similarity among 3D models by matching the similarity among feature binary trees.The experimental results indicate that this approach is robust against rotation transform,noise,model degeneracy etc.,and achieves perfect performance.