3D Model Retrieval Based on Vision Feature Fusion

By applying multi-view features of a 3D model as the whole feature descriptors to match the 3D model feature, this paper presents a modified 3D model retrieval algorithm which is based on the fusion of contour features and texture features of the model. After two-dimensional depth images of the 3D model are obtained under the different views of a spherical bounding box, the contour feature and the texture features of the model images are fused for realizing the constitute of the 3D model. The experiment results show that the proposed method gains great improvement in retrieval speed and effective rate in comparison with other view-based 3D model retrieval methods.