DBPapers
DOI: 10.5593/SGEM2016/B22/S10.119

LINE EXTRACTION FOR ARTIFICIAL OBJECTS USING UNMANNED AERIAL VEHICLE IMAGERY

J. Yeom
Thursday 8 September 2016 by Libadmin2016

References: 16th International Multidisciplinary Scientific GeoConference SGEM 2016, www.sgem.org, SGEM2016 Conference Proceedings, ISBN 978-619-7105-59-9 / ISSN 1314-2704, June 28 - July 6, 2016, Book2 Vol. 2, 931-938 pp

ABSTRACT
Unmanned aerial vehicle (UAV) imagery has been increasingly attracting public
attention due to its advantages. In particular, feature extraction is a technique that
increases the applicability of UAV imagery. Line features derived from artificial objects
are important vector data used to generate basic digital maps, which are closely related
to human social activity. In this study, line features in UAV images were extracted by
the regular grid-based Hough transform (RGHT), which is known for better
performance in local line extraction. Appropriate local line extraction methods were
required because UAV images have extremely high spatial resolution. In addition, a
selective method was devised to extract the lines of artificial objects. This method of
line filtering was achieved by unsupervised classification and designation of built-up
land cover classes. Finally, the line extraction results for artificial objects were
evaluated, and the main causes of errors were analyzed.

Keywords: line extraction, artificial object, regular grid-based Hough transform,
unmanned aerial vehicle image