Research On Algorithm Of 3D Model Retrieval Based On Orthogonal Principal Planes Analysis

The number of 3D models is increasing explosively because of the rapid development of 3D modeling technology and internet technology. In the face of so many 3D model resource, how to find the required 3D model quickly and accurately and how to reuse it are becoming a hot topic following image retrieval and video retrieval. As the space structure of 3D model is complicated, it is difficult to extract the feature of 3D models fast and effectively and to match the feature of 3D models in the domain of 3D model retrieval. After studying prevenient 3D model retrieval methods, the method of Principal Plane Analysis is studied and improved in this paper.Firstly, the method of Principal Plane Analysis was analyzed and studied deeply. Because the only one principal plane is not so adequate to describe 3D models that it has an effect on the retrieval accuracy, three orthogonal principal planes'method is proposed. 3D models are projected on three principal planes obtained by computation in order that 3D models can be described more comprehensively and adequately.Secondly, three images are gotten by projection on three principal planes. A set of rotation invariant feature vectors are abstracted from the three projection images by the method of concentric circles segmentation and image moments, and then feature distances of the two methods are calculated to get the similarity of 3D models. Weighted sum of the two methods was determinated as the result.Finally, an experimental system of 3D model retrieval was designed to testify the improved Principal Plane Analysis. Our algorithm is proved to be feasible and effective by means of comparing different retrieval methods. The performance of our algorithm is analyzed after adding noise, and the stability of our algorithm is testified. Our method is a beneficial exploration to the 3D model retrieval technology.