An adaptive filtering method for airborne LiDAR point clouds in mining areas based on bending energy optimization

LIN Xiaohu; GUO Jiyao; YAO Wanqiang; LOU Ning; ZHENG Junliang; MA Xiongwei; GAO Yinan; ZHANG Qingyu

doi:10.13225/j.cnki.jccs.KT25.0181

Abstract：The high-precision three-dimensional spatial data required for areal subsidence monitoring in mining areas are provided through the application of airborne Light Detection and Ranging(LiDAR)technology.However,challenges such as densely distributed surface features,abrupt terrain gradients,and high similarity in elevation characteristics between surface objects and ground points in complex mining scenarios significantly degrade the accuracy of existing point cloud filtering methods,severely limiting the precision of ground point extraction and the reliability of subsidence monitoring.To address these issues,an adaptive filtering method based on bending energy optimization is proposed for airborne LiD-AR point clouds in mining areas.It achieves accurate ground point extraction in complex mining scenarios and provides high-precision three-dimensional spatial data support for monitoring areal subsidence in mining regions.Firstly,potential seed ground points are extracted using a one-dimensional discrete smoothing spline method,and residual non-ground points are eliminated through multi-scale morphological opening operations.Secondly,a quantitative relationship between terrain bending energy and rebound ratio is established to develop an adaptive cloth stiffness adjustment method for com-plex terrains,enabling the dynamic generation of high-precision reference terrain.Finally,precise ground point extraction is achieved by determining elevation difference thresholds between points and the reference terrain.In order to verify the effectiveness of the proposed method,multi-scenario experiments are carried out in complex mining areas,and the results show that the proposed method had significantly lower errors than the existing method in terms of class I and class II er-rors,with an average total error of 6.08%,which was 49.04%lower than that of the existing methods.The challenge of high-precision ground point cloud extraction in complex mining scenarios is successfully resolved through this method,enabling reliable data support for safety monitoring tasks including areal subsidence monitoring and slope stability analys-is in mining areas.Technical support is simultaneously provided for establishing an integrated sky-air-ground intelligent monitoring system dedicated to mine safety.