基于多级种子点优化的移动曲面滤波算法  

Moving Surface Filtering Algorithm Based on Multilevel Seed Point Optimization

作　　者：朱磊 邓兴升 邢承滨 徐康 Zhu Lei;Deng Xingsheng;Xing Chengbin;Xu Kang(Key Laboratory of Highway Engineering of Ministry of Education,Changsha University of Science and Technology,Changsha,Hunan 410114,China)

机构地区：[1]长沙理工大学公路工程教育部重点实验室,湖南长沙410114

出　　处：《激光与光电子学进展》2020年第17期266-273,共8页Laser & Optoelectronics Progress

基　　金：长沙理工大学公路工程教育部重点实验室开放基金项目(kfj190203)。

摘　　要：针对移动曲面滤波算法的种子点粗差问题,提出了一种基于多级种子点优化的移动曲面滤波算法。首先直接剔除雷达点云数据中的异常值。然后通过格网化建立格网索引,并将点云数据分为两级格网,确定一级格网的种子点,利用一级种子点对二级备选种子点进行筛选。当种子点数量达不到要求时,以一级种子点为参考点进行种子点的表面生长。最后利用选取的种子点进行曲面拟合,计算雷达点云数据真实高程和拟合高程的高差,并采用顾及地形起伏的自适应高差阈值判断地面点和非地面点。与经典滤波算法比较,结果表明该滤波算法能有效减少三类误差。Ⅰ类误差、Ⅱ类误差、总误差精度的平均值分别提高7.30%、4.67%、5.57%。同时将该算法与国际摄影测量与遥感学会公布的8种算法进行比较,结果表明该方法具有较高的精度,自适应性强。Aiming at the seed point gross error of a moving surface filtering algorithm,a multilevel seed point optimization algorithm for mobile surface filtering is proposed in this study.First,the outliers in the radar point cloud data are directly removed.Then,we establish a grid index to divide the radar point cloud data into two-level grids,determine the seed points of the first-level grid,and use the first-level seed points to screen the second-level candidate seed points.When the number of seed points does not meet the requirements,the seed points are grown on the surface using the first-level seed points as the reference point.Finally,the selected seed points are used for surface fitting,the height difference between the true elevation and the fitted elevation of the radar point cloud data is calculated,and the ground and nonground points are assessed using an adaptive height difference threshold that considers the terrain fluctuations.Compared with the classical filtering methods,results show that the proposed filtering algorithm can effectively reduce three types of errors.The average of the accuracy of the error Ⅰ,the error Ⅱ,and the total error are respectively increased by 7.30%,4.67%,and 5.57%.Furthermore,this algorithm is compared with eight algorithms published by the International Society of Photogrammetry and Remote Sensing.Results show that the proposed method exhibits high accuracy and strong adaptability.

关 键 词：遥感 雷达点云 多级种子点 移动曲面 滤波 自适应高差阈值 

分 类 号：P23[天文地球—摄影测量与遥感]