Research On Spatial Index And Visualization Of Massive 3D Point Cloud

Point cloud data processing is an important research direction in the field of information technology.It is widely used in topographic survey,cultural relics protection,three-dimensional reconstruction,urban planning,engineering survey,medical health and other fields.With the rapid development of point cloud acquisition technology and the rapid expansion of application fields,how to establish an efficient spatial index for massive three-dimensional point clouds,how to realize more rapid and accurate point cloud data visualization has become a research hotspot.At present,the spatial division of spatial index is based on spatial regularity and axis aligned bounding box,which is inconsistent with the non-uniformity and irregularity of point cloud spatial distribution,and can not accurately express the spatial structure of point cloud itself.At the same time,the massive point cloud data also poses new challenges to the performance of traditional visualization methods.Aiming at the problems existing in the organization index and visualization of point cloud,based on the research of point cloud acquisition technology,this paper focuses on the organization and management of 3D point cloud data,analyzes the index algorithms commonly used in point cloud data,and puts forward a multi-level index structure based on the spatial distribution characteristics of 3D point cloud,so as to improve the efficiency of data query.On this basis,a multi-resolution LOD model based on spatially accelerated Poisson disk sampling is established,and fast and efficient point cloud visualization is realized by using visibility judgment and elimination technology,LOD selection and judgment technology.The research work of this paper is as follows:(1)The acquisition technology of 3D point cloud data is studied,the principles and categories of 3D laser scanning technology and photogrammetry technology are sorted out,and the characteristics and application scenarios of different point cloud acquisition technologies are summarized and analyzed.(2)This paper studies the organization index algorithm of 3D point cloud data,analyzes and summarizes the common spatial index algorithms of 3D point cloud data,puts forward a spatial division method of directional octree,and further uses KD tree to construct the leaf nodes of directional octree,forming a multi-level index structure based on the spatial distribution characteristics of 3D point cloud,which improves the efficiency of point cloud query.The experimental results show that the directional octree can effectively reduce the total number of nodes and the number of redundant nodes,and improve the space utilization by 5%;Compared with other index structures,the multi-level index structure in this paper consumes less time to build the index,occupies less memory,and the average neighborhood search efficiency is improved by18%.It is fully verified that the combined nested structure of directional octree and KD tree can effectively divide massive point cloud data,realize efficient retrieval of point cloud data,and effectively manage point cloud data.(3)The multi-level of detail(LOD)technology and blue noise sampling algorithm are studied.A multi-resolution LOD model based on spatially accelerated Poisson disk sampling is proposed.The technologies of view elimination,LOD selection and judgment are comprehensively used to reduce the amount of calculation of point cloud rendering and speed up the visualization of point cloud.The experimental results show that the sampling points obtained by the spatially accelerated Poisson disk sampling method are evenly distributed and can well retain the banded characteristics of the point cloud.Compared with the original Poisson disk sampling method,the sampling efficiency is improved by 65.71%.Combined with LOD technology,the problem of slow data scheduling in point cloud data drawing and roaming is solved.