Research On Surface Reconstruction Technology Based On Point Model

Along with the great progress in 3D data sampling technique and devices and increasing requirement of reality, real-time in computer graphics, the reconstruction of curve surface based on point-cloud data has become one of the hot subject of research in CAD, CG and CV.The intersection between parametric surfaces of point model is one of the fundamental issues in computer aided geometric design. As one of the important point-cloud data interpolation tools, RBF offers a unified framework for several problems, such as surface reconstructing, smoothing and blending. This thesis proposes a novel parametric surfaces intersection algorithm based on discrete surfels by studying various algorithms for parametric surfaces intersection, integrating the point-based modeling technique developing rapidly in recent years, and taking full advantage of the simplicity in topology and multi resolution of the point-based modeling technique. Because of the huge computation load and storage complexity of RBF fitting, the speed of reconstruction and the scale of point-cloud data are restricted strictly. By studying the interpolation theory for RBF and comparing the existing reconstruction algorithms, this thesis presents an efficient algorithm for point-cloud data reconstruction. The major works involved are as follows:First, adaptively discretizing and sampling curve surfaces into the point sampled surfaces, so the parametric surfaces intersection issue of point model changes into dynamic re-sampling intersection issue. The sampling point is called surfel, with geometric information attached and an octree is built to store surfels in order to speed up the intersection test process. the surfels near to the intersection curves are re-sampled adaptively according to the error analysis. The intersection points are found out based on the intersection between the surfels in the overlapped leaf-nodes of the octrees. Finally, the intersection segments are acquired by sorting these discrete intersection points.Second, based on octree space partition, a fast surface reconstruction method for point-cloud data using RBF is proposed. It partitions the point-cloud data space firstly and creates octree topologic structure accordingly. Then, at each cell of the octree partitions, a multi-order radial basis function that interpolates or approximates the data points which belong to the current cell is created. Due to the effectiveness of octree algorithm, the size of resulting matrix needed to be processed each time is reduced sharply, which allows to reconstruct large-scale data sets in a reasonable time.Lastly, we perform tests on the intersection algorithm of parametric surface, using two groups of data form intersection of Buddha-NURBS and dragon-monkey-saddle. The test results show that the more stratums in octree, the high accuracy we get. We also reconstruct the two groups of point data processed by intersection algorithm for parametric surfaces surfel, and the ability of reconstruction has a great improvement that the algorithm can process lots of point cloud within reasonable time.