Research On Hole Repair And Reconstruction Technology Of Point Cloud Data

The three-dimensional reconstruction technology can provide reliable analysis of environmental information for autonomous operation of engineering equipment and obstacle avoidance for robots by quickly completing the restoration of the appearance of scene objects.The three-dimensional reconstruction method of point cloud data based on lidar scanning has received more and more attention from experts and scholars at home and abroad in recent years.With the continuous improvement of the scanning accuracy of the three-dimensional point cloud acquisition system,massive data undoubtedly affects the operation efficiency of the reconstruction algorithm,and the existing reconstruction algorithms are increasingly unable to meet people's high requirements for reconstruction efficiency.In addition,due to the complex topology of the measured object,the error of the equipment system,or the external occlusion,etc.the point cloud data collected by the scan contains holes.In this paper,the density cloud pre-processing is first performed on the point cloud data collected by lidar,then the greedy projection reconstruction algorithm in 3D reconstruction is improved,and finally an RBF point cloud hole repair algorithm based on Monte Carlo method is proposed.In order to solve the problem of uneven distribution of point cloud density and holes in data reconstruction,and improve reconstruction efficiency The main work is as follows.1)A point cloud density adjustment algorithm based on fault contour point interpolation is designed.Aiming at the problem of severe uneven distribution of point cloud density of scanning lines,it is proposed to delete redundant data points according to the point cloud separation distance in the measurement direction;in the scanning direction,tomographic contour point interpolation algorithm is used to interpolate between every two adjacent scan line point clouds.By preprocessing the point cloud data,the point cloud data is basically evenly distributed.2)An improved greedy projection reconstruction algorithm is proposed.It is proposed to use voxelized grid filtering to downsample the point cloud to simplify the point cloud data.In order to solve the problem of multiple backtracking when searching for neighboring points in the k-d tree reconstruction,an octree is used instead.The calculation example shows that through data reduction and improvement of search algorithm,the reconstruction efficiency is improved under the premise of ensuring reconstruction accuracy.3)A hole repair algorithm for RBF point cloud based on Monte Carlo method is proposed.Hole repair includes three parts: hole boundary extraction,hole data point filling,and generation of 3D data to be measured.In the hole filling part,for the problem that the commonly used trigonal filling has a large calculation amount and is more complicated,an algorithm based on Monte Carlo method and generating hole filling data points according to the density of edge points is proposed.The calculation example shows that the original morphology and structure of the hole area can be effectively restored,the repaired hole is smoothly connected with its boundary point,and the transition is natural.And shorten the repair time.