A novel feature points selection algorithm for 3D triangular mesh models

As an important technique of 3D (Three-dimensional) mesh model retrieval and partition, feature points are widely used to describe the intrinsic characteristics of 3D surfaces. In this paper, a novel feature points selection algorithm is proposed for 3D triangular mesh models. Detecting from the sudden variation regions of the 3D mesh models, feature points selected by our algorithm carry the significant information and possess the global geometry characteristic of the models. Experimental results show that the selected feature points are robust against attack operations on the mesh models such as mesh simplification and adding noise.