机载LiDAR点云自适应滤波算法  

An Adaptive Filtering Algorithm for Airborne LiDAR Point Cloud

作　　者：吉雨田 张春亢[1] 尹耀 JI Yutian;ZHANG Chunkang;YIN Yao(The Mining College, Guizhou University, Guiyang 550025, China)

机构地区：[1]贵州大学矿业学院,贵州贵阳550025

出　　处：《测绘科学技术学报》2021年第2期142-147,共6页Journal of Geomatics Science and Technology

基　　金：国家自然科学基金项目(41701464);贵州大学培育项目(贵大培育[2019]26号)。

摘　　要：针对传统移动曲面拟合滤波算法难以使用单一且具有自适应性阈值滤波的问题,提出一种改进自适应阈值滤波算法。首先将点云格网化,利用混合最小二乘曲面拟合对多级移动曲面滤波算法进行部分优化;其次利用离散点云数据分布特征计算一级滤波阈值;最后计算格网中最大真实高程值与最小真实高程值之差,利用曲率极限点为该值设定一个自适应系数,对滤波阈值算法进行自适应改进后二级滤波。实验采用国际摄影测量和遥感学会(ISPRS)公布的激光雷达数据集验证算法,结果表明,该算法滤波总误差平均值达到6.26%,连续地形滤波总误差达到4%以下,可以较精确地区分地面点与地物点,精确度较高且适应性较强。Aimed at the problem that the traditional moving surface fitting filtering algorithm is difficult to set a single and adaptive threshold,an improved adaptive threshold filtering algorithm is proposed in the paper.Firstly,the point cloud is grided and the hierarchical moving curve surface filtering algorithm is partially optimized by mixed least squares(LS-TLS).Secondly,the first-level filter threshold is calculated by the discrete point cloud data distribution characteristics.Finally,adaptive coefficient is set by using the curvature limit point for the difference,which is calculated between the maximum and the minimum true elevation value in the grid.The filtering threshold algorithm is adaptively improved to accomplish secondary filtering.LiDAR data for experiments are provided by International Society for Photogrammetry and Remote Sensing(ISPRS).The results show that the average total error reaches 6.26%,and the total error can reach below 4%.The adaptive threshold algorithm can accurately distinguish ground points from non-ground points,which is more efficient and adaptable.

关 键 词：机载LIDAR 点云滤波 移动曲面 混合最小二乘 自适应阈值 

分 类 号：P208[天文地球—地图制图学与地理信息工程]