Realization Of Spectrum Transform-based 3d Model Retrieval And Editing

The continuous development of computer graphics technology makes itself have been in-depth applied In the film and video game industries and so on, in which digital geometric model of geometry processing to become a very important link. Now, three dimensional model of the Internet increased dramatically, how quickly find the model and do geometry editing become a user's pressing need. The research work in 3D model retrieval and geometry editing which are intergrated with spectral transform theory has an important theoretical and practical value.The main progress of the thesis is as follows:1. Propose an improved 3D model retrieval method based on Spherical Wavelet, it overcomes drawback of Spherical Extended Functions which can't extract feature exactly, and it is verified on standard 3D test database. Implement a 3D model retrieval method based on Fourier Transform, experiments and related methods confirm various spectral transform should be rational applied in different domain, and provide a new idea for special domain 3D model retrieval.2. Implement Laplacian mesh fitering and spectral mesh filtering two kinds of mesh filtering methods, which help doing smoothing for geometry blending and so on, provide support for other mesh geometry processing as a useful tool.3. Propose an improved mesh deformation method based on local rigidity. The method intergrates with manifold harmonics transform and mesh segment method, besides, use a symmetric Laplace-Beltrami operator for computing deformed weight.The experimental result presents that deformation effects is good and interactive speed is improved, it also provide a new idea for nonlinear deformation methods.4. Design and development 3DQModelRetrieval system which base on plug-in architecture, make spectral transform theory effectively applied to mesh geometry processing, intergrate with 3D model retrieval and editing algorithms, provide users useful service including retrieval and editing.This research work is partly supported by the National High Technology Research and Development Program(863 Program. No:2008AA01Z301) of China.