Research On Rasterizing Algorithm Of Point Cloud Based On The Poisson Distribution

With the deepening of the field of computer graphics research,the virtual reality(AR)and the enhancement of reality(AR)have attracted a large number of research scholars,and in recent years the film industry has also led to the hot computer the development and extension of graphics,in which the game industry,virtual reality and the growing needs of the real estate content of the development of computer graphics put forward a greater demand for related industries involved in the rapid modeling has also become a research scholar Keen research direction.Point cloud data acquisition mainly through the scanning device or from the picture to extract the feature points,point cloud data acquisition methods are different point cloud density is not the same,different researchers for different density point cloud data based on sparse or dense In this paper,we propose a new method to solve the problem that the cloud density of the cloud data is not stable.Adaptive 3D Reconstruction Method Based on Point-Cloud Poisson Distribution for the Simplicity of 3D Reconstruction.This study is based on the improved approach proposed in the above issues to accomplish the following tasks:(1)To obtain the point cloud data set is discrete distribution,the first point cloud data preprocessing operation and data analysis,for the next step raster to do the argument.First,the grid threshold range is determined according to the minimum three-dimensional points and the Poisson distribution conditions required for the surface fitting in the grid,and then the threshold size is dynamically adjusted within the threshold range and the grid division step is repeated until the threshold value allows the grid division unit The midpoint cloud distribution is consistent with the Poisson distribution to ensure the stability of the algorithm's processing data and the robustness of the adaptive properties.(2)Rasterize the point cloud data with the threshold size and establish the corresponding neighborhood relation and topological relation.The spatial distribution of the three-dimensional data points is made by the dynamically adjusted thresholds in the iterative process without considering the point cloud density,so that the distribution of the point clouds in the grid cells after the partitioning tends to be stable.Then,The three-dimensional data points in the unit construct the neighborhood relation of the k neighborhood,and construct the 26 neighborhood relation of the grid cell to ensure the range of the extended dotted line in the process of generating the grid and improve the speed of the extended search.(3)Select the best candidate points to expand the generation of triangular meshes.Based on the raster division and the established topological relation,the appropriate initial triangle is chosen as the seed triangulation traversal point cloud data set.The neighborhood search of the raster midpoint and the grid neighborhood of the candidate point,Until all the three-dimensional points and the generated lines and triangles are added to the grid to verify the effectiveness of the analysis of adaptive grid division and the robustness of the algorithm.Experiment results show that compared with the two parameters of the number of triangular patches and the running time,the number of triangulation patches increased by 2.5% ~ 27.9%,and the operation time was reduced by 5.0% ~ 31.6%.Experiments show that the algorithm is robust to the efficiency of operation and the cross and mismatch of triangular patches in most cases.