A novel 3D model retrieval approach using combined shape distribution

Abstract

With the rapid development of 3D digital shape information, content-based 3D model retrieval has become an important research field. 3D models are likely to be as prevalent as other multimedia data types in the future. There is a pressing need for effective content-based 3D model retrieval methods. In this paper, a novel combined shape distribution (CSD) descriptor is proposed for 3D model retrieval based on principal plane analysis and group integration. Firstly, based on principal plane analysis, the second principal plane is obtained by using sequential quadratic programming. Secondly, two novel 3D shape descriptors are proposed by introducing the plane normal vectors to other shape distributions. Thirdly, since the histogram of the proposed descriptors can be classified as belonging to one of three types: positive, negative, or crossed with each principal plane, further improvements to the descriptors are presented by integrating these three types of histograms. Finally, a CSD descriptor based on the synthesis of the above descriptors is proposed. Several retrieval performance measures and visual experimental results show that the new methods achieved good retrieval performance.