n this article, a method for the automatic detection of power
lines in a horizontal xy plane from airborne and terrestrial
laser scanning data is presented. The workflow is composed
of four main steps: pre-processing with classification of a
point cloud, filtering of the point cloud, the detection of
points on the power lines by applying the Hough transform
(HT), and vectorisation of power line locations and their
intersections. In the pre-processing step, most of the points
that are not representing power lines are eliminated via
classification of the point cloud. We present our filter, which
reduces the number of points in the point cloud further and
thus accelerates data processing and increases the reliability
of processing in the next steps. We detect the points on the
power lines with the HT on the vector points in the xy
plane. The final track of the power lines is derived from the
straight segments computed by the method of the least squares
from the points that HT recognised on the power lines. The
results are assessed visually and via relative comparison of
the computed intersections coordinates with the reference
data manually extracted from the filtered point cloud. The
proposed method detects almost all power lines in the test
area for both data sets
|
