依据多切面投影的机载激光雷达点云滤波方法  

Airborne lidar point cloud filtering method based on multi-sections projection

作　　者：张杰[1] 顾和和[1] 王飞[1] 贺清清[1] 陈宏念 

机构地区：[1]中国矿业大学环境与测绘学院,江苏徐州221116

出　　处：《激光杂志》2017年第10期47-51,共5页Laser Journal

基　　金：基金项目:江苏省高校优势学科建设工程(SZBF2011-6-B35)

摘　　要：针对机载Li DAR点云滤波问题,提出一种依据多切面投影的滤波方法。该方法首先对利用频数分布和格网进行粗差剔除后的点云,按一定间隔和方向进行切面分组并投影,对各组内的投影点利用拟合一次函数直线来滤波;然后,通过设置不同的切面方向使得每个点都具有多个判定结果;最后,综合各判定结果来实现点云的滤波。以ISPRS第三委员会提供的样本对该方法进行测试,并与8种经典的滤波算法进行比较,表明依据多方向切面投影进行的滤波是有效可行的,但其对地形的适应性并不强有待进一步改进。A method based on multi-sections projection was put forward for differentiating ground from nonground lidar point cloud in the light of the needs of airborne lidar data filter.Outlier elimination based on frequency distribution and grid,section variations of point cloud and linear fittings of points projections divided by group were the first step in this method.Then each point will has many judges through changing the orientation of sections.Finally,point cloud filtering was finished through integrating every judges.The benchmark dataset provided by the International Society for Photogrammetry and Remote Sensing（ ISPRS） commission were used to compare the performance of this method with eight kinds of classic filtering algorithms.The results showed that the method based on multi-sections projection was feasible and effective,but has limited ability in adapting to different terrain environment that needs further improved.

关 键 词：机载激光雷达 粗差剔除 多切面 线性拟合 点云滤波 

分 类 号：TN209[电子电信—物理电子学]