An Automatic Technique for Power Line Pylon Detection from Point Cloud Data

This paper proposes a new pylon detection technique from point cloud data. Two masks are created from the non-ground points that mainly represent trees and power line components. The first mask is the power line mask M-m, which contains the power line components and trees and where successive pylons are found connected with wires. The second mask is the pylon mask M-p, where successive pylons are found disconnected, and thus is exploited to obtain candidate pylons using a connected component analysis. By contrasting the area, shape and symmetry properties between trees and pylons majority of the false candidates (trees) are removed from M-p. Finally, long straight lines that represent wires between successive pylons are extracted from M-m, and used to remove the remaining trees from M-p. Experimental results show that the proposed technique provides a high pylon detection rate in terms of both completeness (100%) and correctness (100%).