论文部分内容阅读
点云滤波是机载LiDAR数据后处理的基础工作,本文提出一种基于多尺度虚拟网格与坡度阈值的机载LiDAR点云滤波方法。该方法采用类似影像金字塔的方式构建不同尺度即不同分辨率的虚拟网格,各级网格都以每个方格内最低点作为地面种子点,然后根据坡度阈值以分辨率由低到高的方式逐层对种子点进行平滑处理,最后以最高分辨率即最小尺度虚拟网格地面种子点作为基准种子点对整个数据集进行滤波处理。本文分别采用城区与郊区两块机载LiDAR数据进行了实验。实验表明,该方法能够有效地提取出地面点,运算效率也比较高,具有一定的实用价值。
Point cloud filtering is the basic work of airborne LiDAR data postprocessing. In this paper, an airborne LiDAR point cloud filtering method based on multi-scale virtual grid and gradient threshold is proposed. The method uses a similar image pyramid to construct virtual grids of different scales or different resolutions. The grids in each grid are ground seed points with the lowest point in each grid, and then according to the gradient threshold, the resolution is from low to high The seed points are smoothed layer by layer. Finally, the entire data set is filtered by using the highest resolution, the smallest scale virtual grid ground seed point as the reference seed point. In this paper, two pieces of airborne LiDAR data of urban area and suburban area respectively are used for experiments. Experiments show that this method can effectively extract the ground points, and the computational efficiency is relatively high, which has some practical value.