3D Model Retrieval Based on the Significant Slice

This paper presents a method that the salient points of the model are used to get the slice and a feature extraction method which is based on Zernike moments and the points' distribution. Firstly, the intercept planes were set in three directions evenly, and then the salient points based on the curvature of the surface vertexes of the 3D model were determined. Intercept planes were updated by the salient points, and the slice was obtained. The characteristics of Zernike moments and points' distribution of the slice sequence were extracted. The weight of two features were determined according to the slicing bounding box size. The experimental results show that slices got with the method can describe the model's characteristics better, and the fused features can improve the precision of the model retrieval.