3D model similarity evaluation for mechanical design reuse based on spatial correlated shape-word clique

Abstract

A novel 3D model similarity evaluation method is proposed to improve the model retrieval accuracy. The bag of words (BoW) method is widely used for 3D model evaluation, in which all the partial features are regarded as independent and disordered. However, 3D model shape diversity is characterized by the spatial differences between partial features. Based on the BoW method, the hidden spatial correlation that could be used to enrich model feature descriptors was analyzed. Given a 3D model, the method begins by building a spatial-relation graph to record the spatial relations. The feature graph is then converted and deconstructed, so that the shape-word cliques hidden in the graph can be revealed. Thereafter, the 3D model similarity evaluation is transformed into the similarity calculation with its corresponding shape-word clique histograms. The ESB benchmark and an agricultural model dataset were used to test the retrieval results. The results show that the proposed method has higher retrieval accuracy, which could satisfy the practical retrieval requirements.