The Algorithms Of Mesh Deformation And Texture Optimization Based On Shape Matching

Due to the rapid development of data acquisition, computer hardware and software,3D digital geometry models have emerged to be the fourth generation of digital media which had experienced several representations, such as 1D sound,2D image and 2D video. Because of the advantages of high reality and more human-satisfied intuition, this kind of representation has been widely used in those, fields, like video production, network game, computer animation, industry design, scientific computing and cultural relics protection.Based on the fundamental theories of computer graphics, differential geometry, computational geometry and digital signal processing, the intensive review and comprehensive analyses of various approaches to mesh deformation and texture optimization are presented. Moreover, we propose some solutions aiming at solving problems related to mesh model deformation and texture space optimization. Main contributions of the dissertation include:1) A novel approach to as-rigid-as-possible mesh deformation based on tetrahedral cells is introduced. First, sets of tetrahedral cells are produced through sparse voxelization. Preserving the rigidity of local transformations of mesh surface and these tetrahedra during deformation is achieved by minimizing the corresponding energy formulation that prevents unnatural artifacts both in the surface and interior of the geometry model. Meantime, in order to avoid the possible collapse emerged under large-scale deformation, we present a simple adaptive tetrahedron decomposition method. In our experiments, we show that the volume and surface details of the whole mesh can be approximately preserved and the local degeneration inside the mesh under large-scale deformation could be efficiently avoided.2) A texture space optimization algorithm based on perceptual saliency feature map is proposed. First, the corresponding saliency map of textured model is efficiently computed by combining both geometry and texture properties. Then the texture image is highly relaxed according to this saliency map. In other words, the essential signals in the texture image are accordingly redistributed. At the same time, an optimization algorithm is presented to deal with the phenomenon of severe mesh distortion. By preserving the rigidity of local mesh, this algorithm could adjust the positions of control mesh during large-scale deformation and then avoid artifacts when mapped back to the model.