Terrestrial laser scanning point cloud ground filtering method utilizing iterative relative density analysis
Terrestrial laser scanning(TLS)is a widely adopted method for acquiring point cloud data,and ground fil-tering represents a crucial step in the processing of such data.Due to scanning angle variations and station setup con-ditions,the distribution range of non-ground points within TLS data tends to be significantly broader than that of ground points.Moreover,existing ground filtering methods often assume a uniform distribution of ground points throughout the entire scanning scene,a presumption not entirely applicable in the context of TLS.To address this challenge,this study proposes an iterative analysis approach for ground filtering utilizing relative density and based on the scanning lines of the ground-based point cloud.The methodology involves the restoration of the angular resolution and scanline information of the disordered ground laser point cloud.Subsequently,scanline-by-scanline relative densi-ty analysis is conducted to derive both the ground candidate point set and the non-ground point set.The ground candi-date point set is incrementally integrated into the relative density analysis,iteratively combined with the scanline,until the number of newly generated non-ground points falls below the predefined threshold.Upon satisfying this criterion,the iteration is terminated,and the remaining candidate points are designated as the final ground point set.The pro-posed method is experimentally validated using three sets of data,and comparisons with existing approaches are pres-ented.The results demonstrate the method's effective constraint on the range of ground points,showcasing a more ac-curate filtering effect,particularly in the transitional zones between ground and non-ground,such as wall footings,curbs,and poles.These results highlight the method's suitability for ground filtering in TLS data.
terrestrial laser scanningpoint cloud filteringrelative density analysisiterationscanlinedensity
万方数据
维普
