A Connected Sets Detection Morphological Filter for Airborne LiDAR DTM Extraction under Urban Area.

In recent years, Point Cloud signal processing has received increased attention. Airborne LiDAR can measure the ground and generate point clouds in a cost-effective and rapid way. In order to generate an accurate Digital Terrain Model (DTM), non-ground point such as buildings, vehicles, and vegetation must be removed, that is, point cloud filtering. In this paper, we propose an adaptive morphological filtering algorithm designed specifically for urban area. The morphological filtering algorithm transforms the point cloud into a grayscale image and identifies non-ground locations using morphological operations. However, the method requires setting the filter window size manually. Improper parameters will affect the accuracy of DTM. In order to improve the adaptivity and robustness of the morphological filtering algorithm. In this paper, we first segment the point cloud images using the integral image adaptive segmentation algorithm, and then detect the buildings using the proposed connected sets detection algorithm to automatically determine the size of the filtering window. In addition to this, we also propose dynamic thresholding for point cloud filtering, which performs better compared to previous methods where it is set to a constant threshold. The experimental results on 15 samples demonstrate the effectiveness of our proposed method. Our code will be available at https://github.com/xdu-whh/ICCCS2023-point-cloud-filtering.