一种参考既有DEM的机载点云漏洞修复方法  

An Airborne point cloud vulnerability repair method based on existing DEM

作　　者：林志君 LIN Zhijun(China Railway SIYUAN Survey and Design Group Co.,Ltd,Wuhan 430063)

机构地区：[1]中铁第四勘察设计院集团有限公司勘察院,武汉430063

出　　处：《铁道勘测与设计》2024年第4期1-4,9,共5页Railway Survey and Design

摘　　要：机载雷达在成像过程中受到地形起伏、地物反射率、物体遮挡等因素的影响,在局部区域形成漏洞,漏洞的存在严重影响DEM的制作。针对机载点云存在漏洞这一现象,提出了一种参考既有DEM的机载点云漏洞修复方法。首先根据点云密度与点云平均距离将三维点云投影到二维平面上,经过网格化与邻域检测,提取出漏洞边界的点云,若存在多个漏洞,则将边界点云进行聚类分割。进而根据点云密度与平均距离插值出漏洞区域内缺失点云的平面坐标,再将DEM对应点处的高程赋值给漏洞内的点云,最后结合边界点云与DEM的高程差,使用反距离加权法(IDW)对漏洞内点云的高程进行高程改正。实验结果表明,该方法对含有多个和不同大小的点云漏洞的数据同样有较好的修复效果,对较大区域的点云漏洞依旧适用,其中5米DEM修复结果的高程中误差为0.82米,1米DEM修复结果的高程中误差为0.19米,具有较高的修复精度,在机载点云漏洞修复中有一定的应用价值。Airborne radar is affected by terrain undulations,ground reflectivity,object occlusion and other factors in the imaging process,forming loopholes in local areas,and the existence of loopholes seriously affects the production of DEM.Aiming at the phenomenon that there are vulnerabilities in the airborne point cloud,a method to repair the vulnerabilities of the airborne point cloud based on the existing DEM is proposed.First,according to the point cloud density and the average distance of the point cloud,the 3D point cloud is projected onto the 2D plane.After gridding and neighborhood detection,the point cloud at the boundary of the vulnerability is extracted.If there are multiple vulnerabilities,the boundary point cloud is processed Cluster segmentation.Then interpolate the plane coordinates of the missing point cloud in the vulnerability area according to the point cloud density and the average distance,then assign the elevation at the corresponding point of the DEM to the point cloud in the vulnerability,and finally combine the elevation difference between the boundary point cloud and the DEM,and use the inverse distance The weighting method(IDW)corrects the elevation of the point cloud within the vulnerability.The experimental results show that the method has a good repair effect on data containing multiple and different size point cloud vulnerabilities,and it is still applicable to point cloud vulnerabilities in a larger area.The error in the elevation of the 5-meter DEM repair result is 0.82.M,the 1 m DEM repair result has a median elevation error of 0.19 m,which has a high repair accuracy and has certain application value in the repair of airborne point cloud vulnerabilities.

关 键 词：LIDAR 漏洞检测 最近距离聚类 点云修复 IDW 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]