Now you see me, now you don't

Abstract

In modern warfare and airborne operations, such as those carried out by drones and helicopters, precise visibility calculations are critical to mission success. However, current automated methods for calculating line-of-sight (LoS) often rely on simplified gridded, meshed, or voxelized 2.5D representations and either ignore vegetation's impact or unrealistically model it as fully opaque. To address these limitations, this research proposes a novel method for calculating fully 3D, non-binary intervisibility directly from airborne LiDAR point clouds. The method leverages Cloud Optimized Point Cloud (COPC) streaming and k-d tree spatial indexing to rapidly query large, remote datasets. Vegetation is automatically segmented using the deep learning Myria3D framework. The algorithm computes viewsheds by modeling terrain, buildings, and semi-transparent vegetation, resulting in a continuous probability score rather than a traditional binary output. This is achieved by adapting the Beer-Lambert law to apply exponential attenuation along a discrete, volumetric LoS. Finally, the results are visualized as 3D spatial volumes or 2D map overlays for practical integration into mission planning workflows, such as safe flight height mapping and visibility-aware routing. This direct point cloud analysis bridges the gap between raw data and high-level spatial decision-making, offering a framework for safer, data-driven mission planning.