DOI: 10.5593/SGEM2016/B22/S10.105


O. Trhan, M. Frastia, M. Marcis
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, 823-830 pp

The main part of the spatial city models are buildings. These buildings can be create
using various ways. Mostly are used data from LiDAR systems or from vectorization of
stereoscopic models from airborne photogrammetry.
In our case we use automatic obtained data from UAV photogrammetry. Nowadays, this photogrammetric method is widely spread, because of the speed of data obtaining,
financial demands and high precision. We use the digital surface model in the form of
the pointcloud and the true orthophoto as the input data. The main problem of automatic pointcloud processing is, that the building walls are not perpendicular and this leads to some errors in the true orthophoto.
In our work, we edit the photogrammetric data by some operations. In the first step, we
arrange the DSM to grid, which has the same size as true orthophoto. The second step is creation of digital terrain model. DTM is create using progressive morphological filter and is used to create the basic digital mask, which is used to building outlines detection.
In next step is digital mask modified and applied to the orthophoto. In areas of the mask are detected building edges, which can be after some editing considered as the building outlines. From these outlines are created the building areas. In these areas, we can generate spatial building model in LOD2. After this, we obtain the DSM with buildings, which are geometrically and positionally correct and the walls are perpendicular.

Keywords: UAV photogrammetry, DSM, DTM, orthophoto, spatial building model