3D model features co-clustering based on heterogeneous semantic network

3D model features clustering is an effective way for 3D model classification and content-based retrieval. Traditional clustering algorithms only gathered the models that have similar features, which can not effectively express semantic information of 3D models. Therefore, we take into account using the heterogeneous features of 3D model which interrelated to each other to co-clustering. The method used the limited semantic information which comes from the multi-channel information to build a heterogeneous semantic network, and then converted the heterogeneous semantic network to the semantic features of 3D model. Through the method of semantic and shape features co-clustering that making the clustering results of shape features include some semantic information. Experimental results on Princeton Shape Benchmark had shown that the co-clustering is better than traditional feature clustering method.