CLASSIFIER-FREE DETECTION OF POWER LINE PYLONS FROM POINT CLOUD DATA
High density airborne point cloud data has become an important means for modelling and maintenance of a power line corridor. Since, the amount of data in a dense point cloud is huge even in a small area, an automatic detection of pylons in the corridor can be a prerequisite for efficient and effec...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2017-11-01
|
| Series: | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-4-W4/81/2017/isprs-annals-IV-4-W4-81-2017.pdf |
| Summary: | High density airborne point cloud data has become an important means for modelling and maintenance of a power line corridor. Since,
the amount of data in a dense point cloud is huge even in a small area, an automatic detection of pylons in the corridor can be a
prerequisite for efficient and effective extraction of wires in a subsequent step. However, the existing solutions mostly overlook this
important requirement by processing the whole data into one go, which nonetheless will hinder their applications to large areas. This
paper presents a new pylon detection technique from point cloud data. First, the input point cloud is divided into ground and nonground
points. The non-ground points within a specific low height region are used to generate a pylon mask, where pylons are found
stand-alone, not connected with any wires. The candidate pylons are obtained using a connected component analysis in the mask,
followed by a removal of trees by comparing area, shape and symmetry properties of trees and pylons. Finally, the parallelism property
of wires with the line connecting pair of candidate pylons is exploited to remove trees that have the same area and shape properties as
pylons. Experimental results show that the proposed technique provides a high pylon detection rate in terms of completeness (100 %)
and correctness (100 %). |
|---|---|
| ISSN: | 2194-9042 2194-9050 |