Density Based Recovery of Urban Power Lines Using Vehicle-Mounted LiDAR

Many methods have been proposed for the automated recovery of power lines using LiDAR. The common approach is to use a bottom-up process where conductors are built up from primitive components. This requires the conductors to have enough clearance for the calculation of local features, as nearby objects can pollute these descriptors. In urban environments, this is often not the case. This paper presents a method which can match the performance of these bottom-up approaches for conductor configurations that would otherwise prove challenging. By using the utility poles to perform a structured search of the conductor region, a density-based clustering algorithm can cluster and reconstruct closely packed power lines in cluttered environments.