森林场景下不同径向基函数插值DEM精度的分析  

Analysis of DEM accuracy using different radial basis function interpolation in forest scenarios

作　　者：张先慧 杨飞 ZHANG Xianhui;YANG Fei(Geological Prospecting and Development Bureau Team 101 of Guizhou Province,Kaili,Guizhou 556000,China)

机构地区：[1]贵州省地质矿产勘查开发局101地质大队,贵州凯里556000

出　　处：《北京测绘》2025年第4期522-527,共6页Beijing Surveying and Mapping

基　　金：贵州省地质矿产勘查开发局地质科研项目(黔地矿科合[2022]21号)。

摘　　要：高精度数字高程模型(DEM)在森林资源调查、地形分析、环境监测等领域具有重要意义。然而,森林冠层的遮挡导致机载激光雷达(LiDAR)获取的地面点云数据稀疏且分布不均,给DEM构建带来挑战。因此,选择合适的插值方法对于获取高精度森林DEM至关重要。本研究旨在分析森林场景下不同径向基函数(RBF)插值方法构建DEM的适用性和精度,并探讨冠层密度对插值结果的影响。研究选取了四种不同冠层密度的森林区域作为试验区,并采用机载LiDAR获取地面点云数据。实验中,使用五种不同的径向基函数(RBF)的核函数(薄板样条函数(TPS),张力样条函数(ST),规则样条函数(CRS),高次曲面函数(MQ),反高次曲面函数(IMQ))对90%的地面点云进行插值,并利用剩余的10%数据作为验证集评估插值精度。研究结果表明,冠层密度对DEM插值精度有显著影响,随着冠层密度的增加,DEM插值误差也随之增大。此外,五种RBF核函数中,TPS和ST函数插值获得的DEM精度最高,CRS函数获得的DEM精度最低。综合考虑精度和效率,TPS函数是最适合森林场景DEM插值的核函数。本研究为森林场景DEM构建提供了重要的参考,建议在森林场景DEM构建中优先选择TPS函数进行RBF插值,并注意冠层密度和控制地面点密度,以提高DEM插值精度。High-precision digital elevation model(DEM) are of great significance in areas such as forest resource surveys,terrain analysis,and environmental monitoring.However,the canopy cover in forests causes airborne light detection and ranging(LiDAR) to obtain sparse and unevenly distributed ground point cloud data,which poses challenges in DEM construction.Therefore,selecting an appropriate interpolation method is crucial for obtaining a high-precision forest DEM.This study aims to analyze the applicability and accuracy of different radial basis function(RBF) interpolation methods for constructing DEM in forest scenarios and explore the impact of canopy density on interpolation results.Four forest areas with varying canopy densities were selected as test regions,and airborne LiDAR was used to acquire ground point cloud data.In the experiment,five different radial basis function(RBF) kernels(thin plate spline(TPS),tension spline(ST),compactly supported radial spline(CRS),multiquadric(MQ),and inverse multiquadric(IMQ)) were used to interpolate 90% of the ground point cloud,with the remaining 10% used as a validation set to evaluate interpolation accuracy.The results show that canopy density significantly affects DEM interpolation accuracy.As canopy density increases,the DEM interpolation error also increases.Among the five RBF kernels,the TPS and ST functions yielded the highest DEM accuracy,while the CRS function produced the lowest accuracy.Considering both accuracy and efficiency,the TPS function is the most suitable kernel for DEM interpolation in forest environments.This study provides important insights for DEM construction in forest scenes and recommends prioritizing the use of the TPS function for RBF interpolation in forest environments while paying attention to canopy density and controlling ground point density to improve DEM interpolation accuracy.

关 键 词：径向基函数(RBF) 机载激光雷达(LiDAR) 数字高程模型(DEM) 插值 

分 类 号：P224[天文地球—大地测量学与测量工程]