ate
ch
nt
>
are
jle
of
es.
rst
ut
he
Eberhard Steinle
A numerical comparison between the results of first and last pulse derived wireframe models has shown mean
differences of about 0.8m to 1.5m in position and about 0.2m to 0.6m in elevation. Regarding the fact that these models
are systematically too large (first pulse mode), respectively too small (last pulse mode), these results seems to be
adequate for most applications concerning 3D city models. Compared to manually measured CAD models (accuracy +/-
0.20m), the deviations are in the order of about +/- 0.2m to 0.9m in position and about +/- 0.2m in elevation.
4 CONCLUSIONS
In this paper the different characteristics of first and last pulse modes of laser scanners were explained, as well as the
effect on the building recognition and reconstruction process. It was found that data acquired by first pulse mode does
coincide better with the contour lines of cadastral maps or manually measured CAD models. On the other hand, last
pulse mode was found to be advantageous concerning the roof shapes, because it is not affected as much as the first
pulse measurement by disturbing objects on the roofs (e.g. antennas, chimneys). Another advantage of last pulse
measurement is that most vegetation is not acquired. In first pulse mode this has to be taken into account by specific
preprocessing procedures. But the necessary spectral information is already acquired by operational systems by means
of video cameras or spectral scanners.
More generally, the results of building modelling depend on the quality of the building hypotheses. Mainly vegetation
but also objects on the roofs may disturb a correct separation of 3D objects and therefore, a correct reconstruction of the
buildings. Problems could also occur, if complex building structures are acquired where the surfaces consists of a lot of
small planes partly occluded.
Nevertheless, first experiences show the principle suitability of laser scanning data for building reconstruction. The
differences founded between building reconstruction based on first and last pulse derived DEMs compared to CAD
models indicates that they are of sufficient accuracy for most applications. In most cases the buildings could be
successfully reconstructed regarding their shape and dimensions. In future, further investigations have to be done to
verify the first experiences concerning the differences of first and last pulse based building reconstruction.
ACKNOWLEDGEMENTS
The presented work was done within the project part C5 of the collaborative research center 461: Strong earthquakes...
(http://www-sfbd61.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
Brunn, A., Weidner, U., 1998. Hierarchical Bayesian nets for building extraction using dense digital surface models.
ISPRS journal of Photogrammetry & Remote Sensing, 53, pp. 296 — 307.
Haala, N., Cramer, M., Kilian, J., 1996. Sensor Fusion for Airborne 3D Data Capture. In: Second International Airborne
Remote Sensing Conference and Exhibition, Vol. I, San Francisco, Ca, pp. 344-353.
Haala, N., Brenner, C., Anders, K.-H., 1998. 3D urban GIS from laser altimeter and 2D map data. In: International
Archives for Photogrammetry and Remote Sensing (IAPRS), Columbus, Ohio, Vol. XXXII, Part 3/1, pp. 339-346.
Lohr, U., Eibert, M., 1995. The TopoSys Laser Scanner-System. In: Photogrammetric Week '95. Fritsch, D. and
Hobbie, D. (eds.). Wichmann, Heidelberg (Germany), pp. 263-267.
Maas. G.. Vosselmann, G.. 1999. Two algorithms for extracting building models from raw laser altimetry data. ISPRS
Journal of Photogrammetry & Remote Sensing, 54, pp. 153-163.
Vogtle, T., Steinle, E., 2000. 3D Modelling of Buildings using Laser Scanning and Spectral Information. In: IAPRS,
Amsterdam, The Netherlands, this volume.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 865
