Locality sensitive hashing for LiDAR data processing

Terrain modeling and building reconstruction requires interpretation of millions of points acquired by terrestrial and airborne laser scanners. Nearest neighbors search is often used in LiDAR data processing based computation of local features in 3D space. Kd-tree and range tree etc. are usually employed for range search. In this paper we investige the feasibility of Locality Sensitive Hashing (LSH) for processing LiDAR data by means of approximate nearest neighbors search. And comparison experiments with kd-tree demonstrate its effectiveness and performance for supervised classification of laser scanning points.