3D geometric modeling is the key technique in computer graphics. It is the necessary step for rendering and animation and of great value in both research and application area. As one of the most important discrete representations of the 3D object in real world, point cloud provide most of the data resource for geometric modeling. Thus much work has been done in processing and reconstruction of common point cloud. However, due to the diverse acquisition of point cloud data and specific requirement on the reconstruction result, the traditional ways of processing and reconstruction don't meet the new demand any more. In this thesis, we explore new techniques for the preprocessing, reconstruction of point cloud and model processing based on the specific feature of point cloud and reconstruction requirements. The main contributions include: 1. We propose a sampling method of huge point cloud data using spatial curve order and a surface reconstruction approach based on witness complex. The method first reorders the point cloud according the spatial curve order and then sequential samples the orders data. The method preserves the spatial characteristic well and is also suitable for out-of-core implementation. After the sampling, we use witness complex theory to reconstruct a manifold triangle surface from sampling data under the constraint of original data. Under certain conditions, the method guarantees a topological consistent reconstruction result. 2. We propose a new algorithm for quad-dominant meshing of unorganized point clouds based on global parameterization. Our meshing method is guided by principal directions so as to preserve the intrinsic geometric properties. We use local Delaunay triangulation to smooth the initial principal directions and adapt the global parameterization to point clouds. By optimizing the fairness measure we can find the two scalar functions whose gradients best align with the guided principal directions. To handle the redundant vertices in the iso-lines due to overlapped triangles, an approach is specially designed to clean the iso-lines. Our approach is fully automatic and applicable to a surface of arbitrary genus. We also show an application of our method in curve skeleton extraction from incomplete point cloud data. We further develop a meshless quadrangulation method based on global parameterization which can produce pure quadrilateral mesh. 3. We propose methods of improving the quality of reconstructed mes...
修改评论