Rotation-based three dimensional shape descriptor

Recent developments in techniques for modeling, digitizing and visualizing three dimensional shapes have led to an explosion in the number of available three dimensional models on the Internet and in many domain-specific databases. However, how to describe three dimensional shapes is a difficult problem. A new three dimensional shape description method based on rotation is proposed in this paper. The first step is to sample points on the surface and compute normal vectors. Secondly, we do Gauss mapping and rotate model, at the same time, rotate the Gauss sphere. The third step is to analysis normal distributions on the sphere surface. Then the Euclidean distance is computed, and finally a statistic histogram is constructed. Experimental results show the effectiveness of this method.