Research And Implementation On 3D Craniofacial Modeling And Editing Methods

Since face is the most expressive part of human being, building realistic 3D facial models is a challenging task in the field of computer graphics. People are paying more attention to the facial modeling techniques, because they improve the efficiency and quality of the traditional facial reconstruction methods. The thesis presents the basic methods and key techniques of facial reconstruction. It includes:1. Facial modeling methods based on CT data. The thesis adopts image pre-processing methods based on CT data, and then extracts image contours. The volumetric data is reconstructed by Marching Cubes algorithm.2. Parametric surfaces reconstruction based on Spherical Harmonics. The thesis analyzes the feasibility of the applications in facial reconstruction based on studying SHs. Barycentric mapping method is used to spherical parametrization, and then iterative residual fitting methods are discussed.3. 3D mesh-editing methods. With the analysis of the FFD, mesh subdivision and mesh simplification algorithms, presenting a new mesh editing algorithm which is based on calculating the weighed values of cell's vertexes, so the mesh can be edited in multi-resolution. Furthermore, mesh smoothing method is implemented to improve the rendering quality.4. System design and implementation. We have built a computer-aided facial reconstruction system, FacialReconstruction, by Visual C++ 6.0 and VTK open source 3d graphics library for the facial reconstruction. The correctness and high efficiency of the algorithms in the thesis are proven by the system.