Research On 3D Point Cloud Data Simplification And Compression

In recent years,with the development of reverse engineering and the improvement of the precision of 3D data acquisition equipment,the density of point cloud is larger than before.Moreover,storage,processing,display and reconstruction will consume a lot of time and computer resources.Therefore,how to deal with data efficiently becomes the hots-pot of experts.The compression and simplification of point cloud can alleviate the pressure of transmission,storage and reconstruction.This paper focuses on the key techniques of point cloud compression and simplification,and proposes an optimization algorithm for each circumstance.1.An octree-based point cloud compression algorithm is proposed.The algorithm improves the stop condition of segmentation to stop dividing at the right depth and to ensure appropriate voxel size.The K neighborhood is established on the basis of segmentation.Then use simple and effective statistical methods to remove the outliers from the original point cloud.In terms of data structure,each node is assigned a bitmask.By manipulating bitmask,query and operate data when traversing,and optimize the subsequent point position encoding.This algorithm effectively removes the outliers and surface noise,increases point cloud compression efficiency with range encoding.The algorithm preserves the key information of point cloud,achieves good compression ratio and shortens the compression time.2.A point cloud simplification algorithm with feature reservation is proposed for 3D point cloud data without any foreknowledge of information.Firstly,by analyzing detailed information of the point cloud data,select the axial direction X and Y to divide and calculate and then X-Y boundary is obtained.Then the voxel-grid method is used to divide the scattered point cloud data from which boundary has been extracted.Establish the K neighborhood.Key feature points of the data are extracted by computing the change of normal vector of the point cloud in different neighborhood.Finally,non-feature points are simplified by using grid index.This algorithm not only preserves the details of the point cloud and makes it close to the original point cloud,but also has good operational efficiency and simplification rate.On the basis of the study of the simplification algorithm,this paper studies the point cloud reconstruction algorithm,and displays the reconstructed result through the PCL_visualization library.The algorithm in this paper uses the point cloud file which is common in point cloud processing and the point cloud file acquired by our laboratory raster system to verify result.Experimental results show that the proposed algorithm is feasible and effective,and can do a better job on the point cloud processing.