3D Model Retrieval Based on Latent Semantic Linear Embedding

In the field of 3D model retrieval,in order to holding the shape similarity and the semantic relationship of the 3D models during the course of feature dimension reduction,the paper explores a LSA based non-liner method to reduce 3D model features′ dimension.We can get all the 3D models′ semantic neighbors by constructing a semantic space and reduce the data dimension by improved Locally Linear Embedding method.The experiments on Princeton Shape Benchmark show that the proposed method achieves good performance not only in reducing dimension but also in retrieval.