Research On Ridge-valley Feature Line Extraction Method Based On Ground Point Cloud Data

Through the development of Earth Observation Technology and modern industrial manufacturing technology,many significant space information acquisition technologies have emerged in the field of traditional surveying and mapping,such as synthetic aperture radar interferometry,aerial photogrammetry,airborne LiDAR technology and so on.It overcomes the disadvantages of slow data collection speed,small data collection amount and high comprehensive cost in traditional measurement,and can quickly acquire large area and large amount of ground point cloud data.To deal with the point cloud data which lacks topological information and has a large scale,high requirements are put forward for spatial data storage and computer processing efficiency.In this,this paper focuses on the extraction method of ridge-valley features in ground point cloud data,which is helpful to preserve feature points in data compression,and can embed ridge-valley feature lines in DEM construction to improve data compression rate and DEM accuracy.Based on the ground point cloud data,this paper studies the extraction of ridge-valley feature points and the generation of ridge-valley feature lines.The main work is as follows:(1)The main acquisition methods of ground point cloud data and the surface fitting method of local point cloud are summarized,and the surface differential geometry characteristics,the relationship between ridge(valley)points and the extreme points of curvature and the specific classification of local surface types are studied.(2)The uniform grid method was improved by using topographic relief to simplify the ground point cloud data,and Kd-tree was built on the streamlined point cloud to improve the search efficiency of k-nearest neighbor and r-nearest neighbor.The experimental results show that the simplified point cloud retains the general undulation feature information of the ground on the premise of obtaining a higher compression rate,meeting the requirements of subsequent ridge(valley)point extraction.(3)In view of the large deviation between ridge(valley)points extracted in Reverse Engineering and actual ridge(valley)points,this paper studies the forming characteristics of local geomorphology,and proposes an improved ridge(valley)feature point extraction method.By fitting the local point cloud,the main direction of the curvature extreme point was obtained,and the maximum point of the normal line in the main direction was taken as the ridge point,and the valley point was similar.It is proved theoretically and proved experimentally that the target points extracted by this method are more consistent with the ridge(valley)points.(4)There is a proposal method of a ridge-valley feature line extraction method based on progressive growth method.The improved uniform grid method was used to simplify the point cloud data.The point cloud was divided into blocks according to the point cloud density.The local point cloud was fitted to a cubic polynomial surface using the least square method,the rough feature lines were obtained by the gradual growth of the local highest point to the r-nearest neighbor,and the broken lines,short branches and cross lines were processed,and finally the ridge-valley feature lines were obtained by fine processing.In the experiment,the extraction effect of feature line was analyzed by comparing the contour line generated based on ground point cloud data,and the feasibility and effectiveness of the method were verified.