CMRT09: Object Extraction for 3D City Models, Road Databases and Traffic Monitoring - Concepts, Algorithms, and Evaluation 
LIDAR instead of programmetry. In many areas of the inner 
city, Cologne has an extreme building density, which 
complicated a clean separation of building geometry and roof 
forms, even though the building outlines contained in the 
ground cadastre map were examined beforehand for their 
accuracy. 
In addition, there were many special building structures such as 
churches that had to be extracted from the airborne LiDAR 
data. Pre-processing efforts were further complicated by the fact 
that Cologne’s ground plan data was outdated or incomplete as 
several new buildings that had been erected and still others had 
been tom down since the last update made to the ground 
cadastre map. 
Figure 14. 3D city model of Cologne. 
representation of the building heights. Because the inner city 
buildings will receive realistic façade textures, highly accurate 
building heights and roof structures as well as building details 
were a key project requirement. 
The entire model will be used a decision making tool for urban 
planning and serves as a visualisation tool and complement to 
Cologne’s Master Plan. The amount of overall manual post 
editing required with the software has been reduced since 
working on the East Berlin model to 15 percent. 
4. CONCLUSION AND FUTURE WORK 
We have presented an approach for the automatic 
reconstruction of 3D building models from LIDAR data and 
existing ground plans. It is based on an algorithm to decompose 
given footprints into sets of nonintersecting cells, for which 
roof shapes are then determined from the normal directions of 
the LIDAR points. The validity of this approach has been 
proven effective, as can be judged by the 3D city models of East 
Berlin and Cologne. 
The next step is to increase the amount of detail by loosening 
some of the restrictions of our shapes and by making them more 
flexible. This is already possible in manual editing. However, to 
increase both the richness in detail and the automation, we plan 
to integrate a segmentation of the roof points to selectively 
decompose the footprints without generating more cells. 
After a careful study of the digital ground map it was 
determined that first several adjustments had to be made, for 
example removing underground buildings and structures such 
as parking garages and identify tom down buildings. This 
required examining the discrepancies between the DTM, DSM 
and building outlines to create the final 3D city model. 
Finally, many larger buildings appeared in several different 
attribute tables containing sometimes conflicting information, 
therefore presented a challenge for both the client as well as the 
operators because these buildings still needed to be 
reconstructed without altering their original building footprints. 
Figure 15. 3D city model of Cologne. 
We have completed a wide-area 3D city model in LOD2 for the 
whole of Cologne by the end of September 2008. This new 
model will be integrated into the existing model, thereby 
replacing the GIS data with a much more accurate 
5. REFERENCES 
Arefi, H., Engels, J., Hahn, M. and Mayer, H., 2008. Levels of 
Detail in 3D Building Reconstruction from LIDAR Data. In: 
International Archives of Photogrammetiy, Remote Sensing 
and Spatial Information Sciences, Vol. XXXVIl-B3b. 
Berlin 3D, 2009. 3D-Stadtmodell Berlin, http://www.3d- 
stadtmodell-berlin.de (accessed 6 April 2009) 
Brenner. C., 2005. Building Reconstruction from Images and 
Laser Scanning. In: International Journal of Applied Earth 
Observation and Geoinformation (Theme Issue ‘Data Quality 
in Earth Observation Techniques), Vol. 6 (3-4), p. 187-198. 
Kada, M. 2007. Scale-Dependent Simplification of 3D Building 
Models Based on Cell Decomposition and Primitive Instancing. 
In: Spatial Information Theory: 8 th International Conference, 
COSIT 2007, pp. 222-237. 
Kolbe, T.H., 2009. Representing and Exchanging 3D City 
Models with CityGML. In: Lee, Zlatanova (Eds.): 3D Geo- 
Information Sciences, Springer-Verlag Berlin Heidelberg, pp. 
15-32. 
Moser, S., Wahl, R. and Klein, R., 2009. Out-Of-Core 
Topologically Constrained Simplification for City Modeling 
from Digital Surface Models. In: International Archives of 
Photogrammetry’, Remote Sensing and Spatial Information 
Sciences, Vol. XXXVIII-5/W1. 
Sohn, G., Huang, X. and Tao, V., 2008. Using a Binary Space 
Partitioning Tree for Reconstructing Polyhedral Building 
Models from Airborne Lidar Data. In: Photogrammetric 
Engineering & Remote Sensing, Vol. 74 No. 11, pp. 1425-1438.