根据地形坡度旋转的车载激光点云形态学滤波

摘要

The existing filtering algorithms,based on the mathematical opening operation,have some limitations when removing the on-ground points in the slope topography.To solve these problems,a new filtering algorithm,which adapted to the slope topography,was proposed.Firstly,divide the whole area into different blocks and build the ground triangulated irregular network to determine the terrain slope and direction of each sub-block.Secondly,according to the terrain slope and direction,the data of each sub-block is rotated to make the terrain slope close to the level,which is more suitable for the filtering method based on the morphological opening operation.Thirdly,the ground points are determined using the multi-scale morphological opening operation method.Finally,the proposed algorithm is verified experimentally using points cloud data of the flat and slope topography,which possesses different surface features,such as buildings,soil,and vegetation.Experimental results suggest that the proposed algorithm is feasible to points cloud data of various terrain characteristics,while showing a better performance in the slope topography.%针对传统的基于开运算的形态学滤波方法在坡度连续变化较大的区域无法有效滤除非地面点的问题,提出了一种可适应斜坡地形的滤波新方法.首先对全区域进行分块,构建地面不规则三角网确定各分块区域的地形坡度和方向;然后根据地形坡度和方向对各分块区域内的数据进行旋转,使地形坡度接近于水平,从而更适用于基于开运算的形态学滤波方法;最后对点云数据执行多尺度形态学开运算确定地面点.分别选择平坦地形和斜坡地形的车载激光点云数据进行了实验,两类点云数据均包含植被、建筑、裸地等,结果表明,提出的方法能够有效滤除各种类型地物,并且对斜坡地形具有较好的滤波效果.