摘要
点云滤波是机载LiDAR点云后处理应用的必要环节。现有的大多数点云滤波方法往往在地形平坦的区域滤波效果比较好,而在地形起伏较大区域滤波效果较差。为进一步提升点云滤波方法的精度及对复杂环境的适应能力,提出一种基于多约束连通图分割的滤波方法。通过设定垂直性、高差、距离三个约束条件构建点云连通图,实现点云分割,并基于地面覆盖率和格网化高程实现地面种子点集获取与筛选。最后,基于点到邻近地面种子点集拟合平面的距离实现地面点集优化。采用国际摄影测量与遥感学会(ISPRS)网站发布的15组专门用于检验滤波效果的点云数据进行实验。实验结果表明,所提方法针对不同的地形环境均可以获得良好的滤波结果。在与其他四种滤波方法的对比中,所提方法能够取得最小的平均总误差(5.44%)。此外,所提方法的平均一类误差和平均二类误差都相对较小,表明所提方法在去除地物点的同时能够有效保护地形细节。
Point cloud filtering is a necessary step in the application of airborne LiDAR point cloud post-processing.Most existing point cloud filtering methods have a better filtering effect in areas with flat terrain but a poor filtering effect in areas with high terrain fluctuation. To improve the accuracy of point cloud filtering methods and their adaptability to complex environments, this paper proposes a filtering method based on multiconstrained connected graph segmentation. In this paper, three constraint conditions, verticality, height difference, and distance, were set to construct the point cloud connectivity graph to achieve point cloud segmentation, and the ground seed point set was acquired and screened based on the ground coverage rate and the grid elevation. Finally, the ground point set optimization was realized based on the distance between the points and the adjacent ground seed point set. To test the filtering effect, 15 sets of point cloud data published on the website of the International Society of Photogrammetry and Remote Sensing(ISPRS) were used. The experiment results show that the proposed method can produce good filtering results in various terrain environments. Compared with the other four filtering methods,the proposed method has the lowest average total error(5. 44%). In addition, the average type Ⅰ error and the average type Ⅱ error of the proposed method are relatively small, indicating that the proposed method can effectively protect terrain details while removing ground object points.
作者
惠振阳
胡海瑛
李娜
李卓宣
Hui Zhenyang;Hu Haiying;Li Na;Li Zhuoxuan(Faculty of Geomatics,East China University of Technology,Nanchang,Jiangxi 330013,China)
出处
《激光与光电子学进展》
CSCD
北大核心
2022年第4期417-426,共10页
Laser & Optoelectronics Progress
基金
中国博士后科学基金(2019M661858)
国家自然科学基金(41801325)
江西省自然科学基金(20192BAB217010)
江西省教育厅科技项目(GJJ170449)
江西省数字国土重点实验室开放基金(DLLJ201806)
东华理工大学博士启动基金(DHBK2017155)。
关键词
遥感
机载LIDAR
多约束连通图
点云分割
点云滤波
remote sensing
airborne LiDAR
multiconstrained connected graph
point cloud segmentation
point cloud filtering
