Ih P
N = 0
as
id
se
he
:d
he
as
Thomas Vógtle
A comparison of the resulting wireframe with a manually derived CAD model (accuracy +/- 0.20m) has shown co-
ordinate differences of about +/- 0.2m to 0.9m in position and about +/- 0.2m in elevation, which should be adequate for
most applications.
5 CONCLUSIONS
In this paper an approach for recognition and 3D reconstruction of buildings in urban environment is presented. First
experiences in our test areas have shown its efficiency concerning accuracy and reliability by using laser scanning
elevation data and spectral information. The main advantage of this approach is that the method is not limited to
predefined building types like gable or hip roofed ones. The only restriction is the reconstruction of building roofs by
planes. Therefore, the overwhelming majority of roof types can be modelled with the exception of conical, cylindrical
or other surfaces of higher degree. In these cases a 3D triangulation can be applied for modelling these roof types.
Another positive aspect is that no additional data sources like digital city maps or digital cadastral maps have to be
provided. Spectral information, as it is used in this approach, is already acquired in today's laser scanning flights by
means of video cameras, later on by multispectral scanners or intensity measurement by the laser scanner itself.
Nevertheless, further investigations on the building reconstruction without additional spectral information will be
carried out at our Institute.
In future, a more sophisticated strategy needs to be developed in order to handle fragmentary acquired building objects
in laser dDEM, e.g. caused by eliminated vegetation or disturbing 3D objects on the roof (antennas, wells etc.). Beside
this, the algorithm for merging neighbouring or enclosed building parts to form one unique object has to be improved to
get suitable CAD objects for further processing (rendering, texture mapping etc.) in 3D city models.
ACKNOWLEDGEMENTS
The presented work was done within the project part C5 of the collaborative research center 461: Strong earthquakes...
(http://www-sfb461.physik.uni-karlsruhe.de), supported by the DFG (Deutsche Forschungsgemeinschaft). The authors
would like to thank in addition Miss Grelois and Mr. Kiema (IPF) for the kindly support of this paper.
REFERENCES
Baltsavias, E. P., 1999. Airborne laser scanning: existing systems and firms and other resources. ISPRS Journal of
Photogrammetry & Remote Sensing, 54, pp. 164-198.
Douglas, D.H., Peucker, T.K., 1973. Algorithms for the reduction of the number of points required to represent a
digitized line or its character. In: Canadian Cartographer, 10(2), pp. 112-123
Ebner, H., Eckstein, W., Heipke, C., Mayer, H. (eds.), 1999. Automatic Extraction of GIS Objects from Digital
Imagery. International Archives of Photogrammetry and Remote Sensing (IAPRS), Munich, Germany, Vol. XXXII,
Part 3-2W5.
Fórstner, W., 1999, 3D-City Models: Automatic and Semiautomatic Acquisition Methods. In: Photogrammetric Week
99, Fritsch, D. and Spiller, R. (eds.). Wichmann, Karlsruhe (Germany), pp. 291-303.
Grün, A., Baltsavias, E. P. (eds.), 1997. Automatic Extraction of Man-Made Objects from Aerial and Space Images
(ID).Birkhäuser Verlag,Basel.
von Hansen, W., Vôgtle, T., 1999. Extraktion der Geländeoberfläche aus flugzeuggetragenen Laserscanner-Aufnahmen.
Zeitschrift Photogrammetrie, Fernerkundung, Geoinformation (PFG), Heft 4, Schweizerbartverlag Stuttgart, S. 229-236.
Lohr, U., 1999. High resolution Laserscanning, not only for 3D-City models. In: Photogrammetric Week 99. Fritsch, D.
and Spiller, R. (eds.). Wichmann, Karlsruhe (Germany), pp. 133-138.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 933