Filtering of Airborne LiDAR Point Cloud with a Method Based on Kernel Density Estimation (KDE)
X-G. Tian, L-J. Xu, X-L. Li, T. Xu and J-N. Yao
In this paper a new method is proposed for filtering airborne light detection and ranging (LiDAR) point cloud data based on kernel density estimation (KDE). The point cloud data is divided into a number of blocks at different sizes step by step. In each block, the kernel probability density of the elevation values of all points is estimated, and a threshold value is selected for data filtering by referring the elevation value of the maximum probability density point. The points whose elevation values are lower than the threshold are classified as ground points. Because the method does not focus on the calculation of individual points, the computation complexity is greatly reduced. Experimental results show that the filtering method is valid and efficient for massive point cloud filtering.
Keywords: Airborne LiDAR, point cloud, filtering, kernel density estimation (KDE), statistical features