DMP-KDO-PCLoSA: a LiDAR point cloud visibility analysis method for urban remote sensing using depth map projection and KNN optimization

Abstract

Accurate line-of-sight (LoS) analysis in LiDAR point clouds is vital for urban remote sensing and geospatial applications, yet sparse and occluded data challenge reliability. We propose DMP-KDO-PCLoSA, a novel method integrating Depth Map Projection (DMP) with K-Nearest Neighbour Depth Optimization (KDO), featuring a Variable Correction Threshold (VCTS) for adaptive depth refinement. Compared to density-based and voxel-based methods, our approach achieves 92.23% visibility detection accuracy (VDA), reduces false detection rates by 1.09%, and decreases computation time by 40%. Experiments on urban datasets demonstrate its robustness in complex city environments. DMP-KDO-PCLoSA enhances LiDAR point cloud processing for urban modelling, supporting smart city planning, environmental monitoring, and disaster management. Its scalability and efficiency offer practical solutions for large-scale geospatial analysis, advancing remote sensing applications in diverse urban contexts.

Keywords

• LiDAR
• point cloud
• visibility analysis
• urban remote sensing
• city planning