基于径向基函数神经网络的机载LiDAR点云空洞填补方法  

Filling Method of Airborne LiDAR Point Cloud Hole Based on the Radial Basis Function Neural Network

作　　者：蔡香玉 杨林[1,2,3,4] 吕海洋 

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,江苏南京210023 [2]江苏省地理环境演化国家重点实验室培育建设点,江苏南京210023 [3]江苏省地理信息资源开发与利用协同创新中心,江苏南京210023 [4]南京师范大学地理科学学院,江苏南京210023

出　　处：《南京师范大学学报（工程技术版）》2017年第3期57-62,共6页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：国家自然科学基金(41631175;41471102)

摘　　要：机载LiDAR技术为地表三维数据的获取和DEM、DSM的构建提供了有利的条件.由于建筑物和植被遮挡等原因,造成了点云的缺失,形成区域的空洞,给地表建模带来不便,需要对LiDAR点云数据进行插值处理以修复缺失的数据.对径向基函数(RBF)神经网络构建插值模型进行了研究,利用该模型对点云中缺失的空洞区域进行修复.通过利用一部分采样点对RBF神经网络进行学习训练,得到模型中参数的具体值,然后利用这些参数值对空洞区进行插值.实验验证了RBF神经网络模型的有效性及插值精度.Airborne LiDAR technology provides favorable conditions for the acquisition of 3D data and the construction of DEM, DSM. Such reasons as buildings and vegetation shelter result in the lack of point cloud and the formation of region-al holes,which make the surface modeling inconvenient. LiDAR point cloud data Interpolation is needed to repair the missing data. The RBF neural network interpolation model is studied by using the model to repairempty area in the point cloud. A part of the sampling points is used to train RBF neural network to get the specific values of parameters in model,then these parameters are used to interpolate the empty area. Through experiments, the effectiveness and the interpolation precision of the RBF neural network model are verified.

关 键 词：空间插值 LIDAR点云 空洞填补 RBF 神经网络 

分 类 号：P208[天文地球—地图制图学与地理信息工程]