Scattered Point Cloud Model Triangle Mesh Processing Algorithm And Realization

With the rapid development of hardware technologies of 3D measuring and dataset capturing, such as laser-ranging-scanning technique and so on, the acquisition of scattered data of objects with more details has become possible and this makes people very easy to gain more and more high-accuracy and high-density 3D data on the surface. Establishing the real object's digital model with 3D data on the surface has become the trend of development in international graphic field in recent years, which greatly promoted development in reverse engineering, CAD/CAM and other rapid prototyping manufacturing technology. As an important research subject, modelling 3D shapes based on scattered points is also widely applied in the field such as computer vision, medicine, aviation, virtual reality and so on.Triangulation of scattered points cloud is the first important step and the base of surface reconstruction, thus the direct study of trigulation from scattered points data has great realistic significance in constructing interpolation surface and its extended use in rather more widely area. In view of this, we take a deep research on a series of issues which ranges from the point cloud denoising and simplification, to point data triangulation and local optimization, as well as the multi-resolution terrain display.The major contributions of this paper are:1. The method for denosing scattered point cloud data with noise and outliers are proposed. Based on Kd-Tree to quickly find k-nearest points of each point, the Gaussian function was used as a weight function to estimate the current point's effect on its neighbors. Besides this, we set threshold value for noise and outlier adjustment according to average station distance and influence of point cloud, so as to improve the adaptivity of the denoising algorithm.2. Based on local neighborhood and local sampling density concepts, a adaptive point-based model simplification algorithm is presented in this paper. The algorithm can complete any point cloud simplification according to user-specified simplification ratio, and support multi-resolution representation of the model.3. A new incremental triangulation algorithm for unorganized points is proposed. Based on the method of wave front, the algorithm can reconstruct incrementally triangular meshes from points cloud data. Recurring to Kd-tree space division and other optimum vertex estimation, the algorithm starts with selected intialized triange, and uses its three edges as searching elements to reconstruct mesh gradually and symmetrically. The model with holes and gaps can be identified effectively. The expeimental results show that the algorithm has a strong adaptability, can work well for models with arbitrary topology.4. Combine point cloud simplification algorithm with triangulation algorithm for point cloud. It can not only support multi-resolution triangle meshes display, but also can greatly shorten the processing time of triangultion. A larger degree of simplification led to shorter processing time, but made the reconstructed mesh model loss some details. So without high precision requirement, we can first perform a processing of point-cloud simplification and then triangulation.