三维激光扫描地形点云的分层去噪方法

A Denoising Method by Layering for Terrain Point Cloud from 3D Laser Scanner

针对三维激光扫描采集的地形点云,提出在高程方向进行切片分层的去噪方法。将分层提取的点云投影在平面上,对点云进行网格划分,把网格内的点云转换为二值图像,采用数字图像处理方法过滤离散的网格噪点;然后对分层提取的点云轮廓特征格网,通过对上下层轮廓边缘特征比较,获取点云的地面轮廓网格点,删除孤立噪点与非地面点。根据对不同地形的点云进行去噪实验,通过分层过滤非地面点可以得到很好的去噪效果。

For the terrain point cloud acquired by 3D laser scanner, a denoising method by layered slices in eleva-tion direction is presented in this paper. Firstly, the hierarchically extracted point cloud is projected on a plane and the point cloud is divided by grid. In the meanwhile, the point cloud in the grid is transferred into a binary image. So the discrete grid noises can be filtered by using the method of digital image processing. Then the terrain contour grid can be extracted from the point cloud slice by slice. By the comparison of the upper and lower edge features, the isolated and non-ground points can be removed. According to the denoising experiments on the point cloud of different terrain, a satisfied denoising outcome can be obtained by filtering the non-ground points hierarchically.