3D symmetrical model retrieval based on local feature comparison in mechanical engineering

摘要

The radius angle histogram (RAH) is a useful method to retrieve 3D models in the engineering field. This method often requires using the absolute value of the cosine function of the angle between surface normal and radius and usually makes the inner face and outer face indistinguishable. As a result, the retrieval efficiency and precision are insufficient when this method is applied to retrieve models with symmetrical characteristics. In order to improve retrieval efficiency and precision, an improved radius angle histogram method is proposed. This method directly employs the angle between surface normal and radius as the shape descriptor to replace the cosine function of the angle between surface normal and radius. In the process of retrieval, the normalised model coordinates are first established by principal component analysis (PCA). Then local features in each quadrant are extracted and the similarity of the local features is ascertained. Models with a high similarity of local features are obtained and are considered as the retrieval results. Finally, an experiment is described. The simulation results show that the proposed method has higher retrieval efficiency and precision than RAH, especially for symmetrical models. The proposed method is suitable for mechanical component design, especially in the rapid prototyping domain.