A segmentation-based method for 3D model retrieval

Abstract

A novel 3D model segmentation and retrieval method was introduced in this paper that is based on the topological information and partial geometry features of 3D mode. The proposed algorithm extracts feature for every triangular piece using flatness of a triangular, and partitions the 3D model into a set of triangular pieces with different flatness. Then, a watershed-based algorithm is developed and followed by an efficient region merging scheme to get a meaningful segmentation of the 3D model. And also a characteristics topology graph is constructed for the model in accordance with the topological relations between the blocks. To simplify the computational complexity of the similarity of two models, spanning trees of the topography graph is constructed firstly and then the similarity between the spanning trees is calculated. Experimental results show that this segmentation and retrieval algorithm is efficient and robust.