Research And Implementation Of Content-Based 3D Model Retrieval Methods

Due to the development of multimedia and Internet techniques, the 3D Models are gradually becoming the forth media data type after sound, images and video. Aim at the problem of searching for 3D models in massive network information, we study pro-processing, interactive retrieval methods, feature extraction and other 3D model key technologies. The research work has an important theoretical and practical value. The main progress:(1) Summarize and analyze present status of content-based retrieval of 3D models, related technology and the typical 3D model retrieval system.(2) Implement the CPCA and isotropic preprocessing, and we also compare and analyze the performance. It provides a basis for feature extraction, and effectively improves the accuracy of the search algorithm.(3) Propose a new 3D model retrieval method with building blocks. Different from the existing retrieval ways, the user constructs imaginary query model through building blocks, and achieve a new human-computer interaction on-line retrieval way. The experiments show that the construction of such a model-based interactive search method is simple to use, more in line with the actual retrieval environment, and open up a 3D model of the new ideas retrieval method.(4) Improve an extraction algorithm based on the elevation descriptor. From the experimental results, compared to a simple shape distribution extraction algorithm, the method of the model has a good ability to distinguish between, and a combination of two methods of preprocessing of the analysis and comparison. (5) Design and develop a 3DModelRetrival system which is based on the P/P-style software architecture. The system combines C/S and B/S structure, and implements relevant retrieval technology of the paper. It is simple, easy-to-use, expansion and strong characteristics.