Ansgar Brunn
From a preprocessing, which is based on the feature extraction (Fuchs and Förstner, 1995) of the DSM, we get the
geometry of the approximate building description (cf. fig. 7(a)). The a priori likelihood estimate leads to the shown simplex
classification. The misclassification of a least one face is obvious. The iterative process of classification and reconstruction
results in the building shown in fig. 7(b). Only essential simplices are shown in the visualization. Misclassifications and
the geometry of the surface have been corrected automatically in this second step.
We found that the error tolerance of the approach is quite high, which is also known from MRFs in general. Correction of
single misclassification could be done, but regions of false classifications cannot be corrected due to the defined low sizes
of the neighborhoods.
6 CONCLUSIONS AND OUTLOOK
This article presents a new approach for automatic analysis of multi sensorial data. The integration of interpretation and
reconstruction in a common algorithm was a major goal. A new structure for the representation of buildings based on their
topology was introduced. The example shows the correction capability of the local classification and local reconstruction.
In future research, the integration of changes in topology during the classification has to done. The implementation and
modeling should be extended to CW-complexes. For an empirical evaluation of the approach, additional models, large
datasets and additional sensor types have to be covered.
Acknowledgments: This work has been founded by the European Community as part of the ESPRIT project “Image
Processing for Automatic Cartographic Tools (IMPACT)", No. 20.243.
REFERENCES
Baillard, C., Schmid, C., Zisserman, A. and Fitzgibbon, A., 1999, Automatic line matching and 3d reconstruction of
buildings from multiple views. In: ISPRS Conference on Automatic Extraction of GIS Objects from Digital Imagery,
IAPRS Vol.32, Part 3-2W5.
Brunn, A. and Weidner, U., 1997. Extracting buildings from digital surface models. In: 3D Reconstruction and Modeling
of Topographic Objects, Vol. 32, IAPRS.
Brunn, A., Gülch, E., Lang, F. and Fórstner, W., 1998. A hybrid concept for 3D building acquisition. JPRS 53(1),
pp. 119-1238.
Fischer, A., Kolbe, T. H. and Lang, F., 1999. On the use of geometric and semantic models for component-based building
reconstruction. In: SMATI 99, pp. 101—120.
Fuchs, C. and Fórstner, W., 1995. Polymorphic grouping for image segmentation. In: 5th International Conference on
Computer Vision, Proceedings, pp. 175-182.
Haala, N. and Brenner, C., 1997. Interpretation of urban surface models using 2D building information. In: A. Gruen,
E. Baltsavias and O. Henricsson (eds), Automatic Extraction of Man-Made Objects from Aerial and Space Images (II),
Birkhauser Verlag, Basel, Switzerland, pp. 213-223.
Halmer, A., Heitzinger, D. and Kager, H., 1996. 3d-surface modelling with basic topologic elements. International
Archives of Photogrammetry and Remote Sensing XXXI(B4), pp. 407-412.
Jánich, K., 1994. Topologie. 4 edn, Springer, Berlin.
Koch, K. and Schmidt, M., 1994. Deterministische und stochastische Signale. Diimmler Bonn.
Lang, F., 1999. Geometrische und semantische Rekonstruktion von Gebäuden durch Ableitung von 3D-Gebäudeecken.
Berichte aus der Informatik, Shaker Verlag. 184 Seiten, 60 Abbildungen, ISBN 3-8265-6619-X.
Li, S., 1995. Markov Random Field Modeling in Computer Vision. Springer-Verlag Tokyo.
Moons, T., Frere, D., Vandekerckhove, J. and Gool, L. V., 1998. Automatic modelling and 3d reconstruction of urban
house roofs from high resolution aerial imagery. Computer Vision - ECCV'98, Freiburg pp. 410—425.
Rissanen, I., 1987. Minimum description length principle. Encyclopedia of Statistical Sciences.
Vosselmann, G., 1999, Building reconstruction using planar faces in very high density height data. In: Automatic
Extraction of GIS objects from Digital Imagery, ISPRS.
124 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.
