Algorithm For Point Cloud Simplification Based On Hybrid Indexing And Quality Evaluation

With the rapid development of geographic information industry,the accuracy requirements of geographic information and geographic data are increasing.3D point cloud data is important for the expression and description of geological phenomena as well as modeling.As an important means to obtain point cloud data,3D laser scanning technology is widely used in many fields such as topographic mapping and deformation monitoring because of its characteristics of non-basic measurement,rapidity,real-time,digitization and automation.However,the massive amount of point cloud data obtained by 3D laser scanning technology can seriously affect the efficiency of computer transmission,storage,processing and subsequent 3D modeling,so it is very important to perform the necessary streamlining of point cloud data.In this paper,a feature-preserving hybrid index point cloud streamlining algorithm is proposed.The main research contents are as follows:(1)In terms of point cloud K-neighborhood construction,this paper proposes an improved K-neighborhood construction method,which can realize the K-neighborhood construction of any point in the point cloud more quickly.Firstly,the point cloud is divided by the raster method,and then the KD tree is constructed in its raster and adjacent raster in order to quickly find the nearest neighbors in the point cloud,so as to realize the construction of Kneighborhood.The experimental results show that the algorithm in this paper has higher efficiency in finding the nearest neighbor points of the point cloud,thus improving the overall efficiency of constructing the K-neighbor domain of each point in the point cloud data.(2)In terms of point cloud feature point extraction,this paper proposes a feature point extraction method based on point cloud boundary preservation.Firstly,on the basis of Kneighborhood construction,the distance from each point in the point cloud data to its neighborhood center of gravity and the curvature of each point are calculated,and then the discriminant threshold is set to discriminate the boundary points and feature points of the point cloud respectively,and the redundant data are removed from the set of the two obtained point cloud data sets to finally obtain the point cloud boundary-preserving feature points.The experimental results show that the improved algorithm can identify the point cloud boundary points and feature points simultaneously.(3)In point cloud refinement,a hybrid index point cloud refinement algorithm with feature preservation is proposed in this paper.Firstly,the extracted boundary points and feature points are retained by using the feature point identification method proposed in this paper,and then the remaining non-feature points are streamlined.Considering the logical structure of the octree can be better combined with the grid method,this paper uses the hybrid index structure of KD tree and octree and combines the grid simplification algorithm to streamline the non-feature points.The upper layer of the point cloud is divided by KD tree,and after reaching the set threshold,the lower layer is divided by octree,and then the point cloud is simplified within each small cube according to the centroid simplification algorithm in the grid simplification algorithm.The experimental results show that the algorithm can retain the point cloud boundary information and feature information better,and the point cloud distribution is more uniform in the flat area without large area of hole region.(4)In order to objectively evaluate the quality of the streamlined point cloud data,this paper uses surface area assessment,volume assessment and deviation analysis to evaluate the quality of the streamlined point cloud data model by random sampling method,grid method,curvature method and the proposed algorithm respectively.The experimental results show that the evaluation index values of the streamlined point cloud data model of the algorithm in this paper are better than the other three algorithms,so the algorithm proposed in this paper has higher accuracy compared with the traditional methods in point cloud streamlining processing.