An Effective Approach to Content-Based 3D Model Retrieval and Classification

Abstract

The Development of effective content-based 3D model retrieval and classification is still an important research issue due to the growing amount of digital information, this paper present a novel 3D model retrieval and classification algorithm. In feature representation, a method combining distance histogram and moment invariants is proposed to improve the retrieval performance. A major advantage of the distance histogram is its invariance to the transforms of scaling, translation and rotation. Based on the premise that two similar images should have high mutual information, or equivalently, the querying image should convey high information about those similar to it, this paper proposed a mutual information distance measure to perform the similarity comparison. Multi-class support vector machine performs the classification for it has a very good generalization performance. This paper tested the algorithm with a 3D model retrieval and classification prototype, the experimental evaluation demonstrates the satisfactory retrieval results and good classification accuracy.