机载激光雷达的数字高程模型孔洞修补方法及精度分析

Hole repair method and accuracy analysis of digital elevation model for airborne LiDAR

针对因地物遮挡、激光散射、复杂地形等因素影响,激光雷达在扫描过程中不可避免产生数据孔洞,进而导致生成数字高程模型(DEM)局部质量不佳的问题,该文提出在激光雷达数据TIN网格化基础上,检测网格孔洞边界,并在特征平面上按阈值新增三角面片,结合径向基函数调整网格节点至隐式曲面,完成孔洞修补。实验利用获取的山区机载激光雷达点云数据,进行网格模型孔洞实例修补并开展不同地形、不同微地貌的孔洞修补,生成对应的DEM并对网格孔洞模型修补效果进行精度分析。结果表明:地形复杂度指标和孔洞面积越大孔洞修补精度越低;在山地实验区,地形复杂度≤0.5,孔洞面积≤200 m^(2)时,修补精度能满足1∶500 DEM产品要求,≤500 m^(2)时,满足1∶1000要求,≤1000 m^(2)时,满足1∶2000要求。

In response to the inevitable generation of data holes during the scanning process of LiDAR due to factors such as ground object occlusion,laser scattering,and complex terrain,which can lead to poor local quality of the generated DEM,this paper proposes to detect the boundaries of grid holes on the basis of LiDAR data TIN gridding,and add triangular patches according to the threshold on the feature plane.Combined with radial basis functions,the grid nodes are adjusted to implicit surfaces,Complete hole repair.The experiment utilizes the obtained airborne LiDAR point cloud data in mountainous areas to perform hole instance repair on grid models and carry out hole repair on different terrains and micro landforms,generating corresponding DEM and analyzing the accuracy of the mesh hole model repair effect.The results indicate that the larger the terrain complexity index and hole area,the lower the accuracy of hole repair;In the mountainous experimental area,when the terrain complexity is≤0.5 and the hole area is≤200 m^(2),the repair accuracy can meet the requirements of 1∶500 DEM products,1∶1000 for≤500 m^(2),and 1∶2000 for≤1000 m^(2).