Ground point filtering method for terrestrial laser scanning point clouds using iterative relative density analysis and its mapping applications

Terrestrial laser scanning (TLS) is widely utilized for acquiring point cloud data, with ground filtering playing a pivotal role in data processing. Variations in scanning angles and station setup conditions often lead to a broader distribution range of non-ground points compared to ground points within TLS data. However, prevailing ground filtering methods commonly assume a uniform distribution of ground points across the entire scanning scene, which may not hold true in TLS contexts. To tackle this issue, this study proposes an iterative analysis approach for ground filtering based on relative density, focusing on the scanning lines of the point cloud. The methodology involves restoring the angular resolution and scanline information of the disordered ground laser point cloud. Subsequently, a scanline-by-scanline relative density analysis is performed to identify both the ground candidate point set and the non-ground point set. Once this criterion is met, the iteration concludes, and the remaining candidate points are designated as the final ground point set. Experimental validation of the proposed method is conducted using three datasets, accompanied by comparisons with existing approaches. The results demonstrate the efficacy of the method in constraining the range of ground points, leading to a more precise filtering effect, particularly in transitional zones between ground and non-ground surfaces, such as wall footings, curbs, and poles. These findings emphasise the suitability of the method for ground filtering and mapping applications in TLS data.