Detection of Pole-like Objects from LIDAR Data

This paper presents novel approach for detection of pole-like objects from LIDAR data. The designed method uses directional vector to detect pole-like structures in unordered point clouds. A new segmentation algorithm is presented as well. The segmentation is designed to overcome a common problem found in LIDAR data of urban environments, where a lot of poles are connected together with various types of wires. The method is tested on real world point clouds captured during mobile mapping process.