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.
出处
《南京师范大学学报（工程技术版）》
CAS
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)
