Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Identification and correction of road courses by merging successive segments and using improved attributes

: Bulatov, Dimitri; Häufel, Gisela; Pohl, Melanie

Postprint urn:nbn:de:0011-n-4265749 (10 MByte PDF)
MD5 Fingerprint: fc97b42a1d2e7d010be8840290a299d0
Copyright Society of Photo-Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.
Created on: 4.1.2017

Erbertseder, T. ; Society of Photo-Optical Instrumentation Engineers -SPIE-, Bellingham/Wash.:
Remote Sensing Technologies and Applications in Urban Environments : 26-27 September 2016, Edinburgh, United Kingdom
Bellingham, WA: SPIE, 2016 (Proceedings of SPIE 10008)
ISBN: 978-1-5106-0420-9
ISBN: 978-1-5106-0421-6
Paper 100080V, 11 pp.
Conference "Remote Sensing Technologies and Applications in Urban Environments" <2016, Edinburgh>
Conference Paper, Electronic Publication
Fraunhofer IOSB ()
Dominant Direction; vectorization; urban terrain; road database

Both in military and civil applications, there is an urgent need for a highly up-to-date road data, which should be ideally semantically structured (into main roads, walking paths, escape ways, etc.) with application-driven attributes, such as road width, road type, surface condition and many others. A vectorization algorithm processing aerial images recently acquired yields an up-to-date road vector data, which are, however, often represented by wriggly, noisy polylines without semantics. The reasons for zigzagged street courses are insufficiencies in the intermediate results of sensor data processing (orthophotos, elevation maps) and occlusions caused by trees, buildings, and others. In the current contribution, an improved computation of geometric attributes will be explained which makes a difference between straight and circular (or elliptic) polylines. Using improved attributes, the candidates for polylines having identical course and sharing a junction are determined. From such candidates, we form chains of polylines. These chains correspond better to the intuitive perception of the term street than the previously used road polylines, because, even after being interrupted by narrower side roads, a chain maintains its label. The generalization of chains with simultaneously adjusting positions of junctions is evidently performed. We apply a generalization with the purpose-based modification of a well-known polyline simplification algorithm once chain-wise and once polyline-wise in order to show - by means of qualitative results - the advantages of the chain-wise generalization.