改进坡度的LiDAR点云形态学滤波算法  

SLOPE IMPROVED MORPHOLOGICAL FILTERING ALGORITHM FOR LIDAR POINT CLOUDS

作　　者：李峰[1,2] 崔希民[1,2] 袁德宝[1,2] 王果[1,2] 张玲[1,2] 

机构地区：[1]中国矿业大学(北京)煤炭资源与安全开采国家重点实验室,北京100083 [2]中国矿业大学(北京)地球科学与测绘工程学院,北京100083

出　　处：《大地测量与地球动力学》2012年第5期128-132,共5页Journal of Geodesy and Geodynamics

基　　金：国家自然科学基金(41071328)

摘　　要：针对经典形态学滤波算法中存在过多预置参数的现状,通过自动计算全区域地形的平均近似坡度,估计窗口迭代过程中用到的高差阈值。在得到初始格网的类别后,再利用格网坡度和高差阈值进一步区分格网内点的类别,最终达到移除非地面点类的目的。该算法只需输入最大建筑物尺寸、激光点云的误差,就可以有效地过滤绝大多数的非地面点类,同时减少了Ⅰ类误差和总误差的大小。In order to solve the problems on overmuch predefine parameters prior to performing classical morphological filter,this paper proposes an improved slope morphological filter algorithm.Firstly,an average approximate slope of the whole data sets is automatically calculated,and then the height difference threshold used in window iteration is estimated.Furthermore,the sorts of several point clouds in a grid are identified according to a slope of a grid and height difference threshold.As a result,terrain points are retained and non-terrain points are removed.The experimental results show that this approach is able to remove majority off-ground points effectively and reduce type Ⅰ errors along with total errors by means of inputting a maximum building size and errors of LiDAR.

关 键 词：LIDAR 形态学 坡度 滤波 地面点类 

分 类 号：P207[天文地球—测绘科学与技术]