Research On Point Cloud Compression Algorithm Based On Geometric Feature Constraint

As an emerging measurement technology,3D laser scanning technology plays an increasingly important role in the surveying and mapping industry.3D laser scanning technology has the advantages of high efficiency and high precision,and is widely used in practical engineering measurement,road monitoring and urban planning.3D laser scanning technology can acquire massive point cloud data in a very short time,but the huge amount of data brings a lot of inconvenience to the processing of point cloud.How to compress point cloud data while retaining its geometric feature information is an important research topic.The main research contents of this paper are as follows:(1)By analyzing the rule grid space index,quadtree index,KD tree index,R-tree index and octree index,the encoding-based octree index is selected as the spatial index structure of the point cloud data in this paper.Because it is more efficient for neighborhood queries(2)Based on the coded octree tree,the gross error elimination method of the point cloud is constructed,and the gross error in the point cloud is eliminated.(2)Using the neighborhood point and elevation threshold combined with the coded octree tree to construct the point cloud gross error elimination method,the gross error in the point cloud is eliminated,and the gross error effect is eliminated by experimental data.(3)The uniform compression method based on rule bounding box and the slope-based compression method are analyzed,and the compression method combining octree and curvature algorithm used in this paper is expounded.The uniform compression method based on the rule bounding box,the slope-based compression method and the compression method combining the octree and the curvature are used to compress and analyze the on-vehicle 3D laser point cloud data respectively,and compare the compression time rate.Experimental analysis of the retention characteristics of the representative geometrical feature information of roadside stone,traffic light,uneven road surface and street tree after compression.Through the compression experiment of the actual point cloud data,the compression algorithm based on the rule bounding box is fast but the retention effect on the geometric feature information is poor.The slope-based compression algorithm can retain the geometric features but the rate is very slow.For the curvature compression method in this paper,it also has a faster rate while retaining the characteristics of the feature information.