Multi-feature 3d model retrieval method based on shape context

The selection of features, the representation of features, and the mode of fusion are keys of 3d model retrieval technology. In the paper, we propose a new 3d model retrieval method which is based on shape context. It combines the fast ORB features and the precise shape context features. ORB features describe the local information. After extracting the CANNY edge information, the shape context features are extracted to describe the global information. Then calculating the final similarity based on shape context features and ORB features. Experimental results demonstrated that our method outperforms several existing methods.