655 
4 Conclusions 
In this paper we stated that due to the availability of generic road models, the extrac 
tion of roads from satellite images becomes less shallow. We have tested our hypothesis 
by three case studies based on a combination of image data, GIS data, computer vision 
paradigms and optimization techniques. This resulted in semi-automatic systems for road 
delineation under guidance of GIS related knowledge applicable for certain types of terrain 
and landcover. One of the big problems is the validation of the results. Other generic road 
network models and other knowledge sources need to be exploited. The error prone nature 
of GIS data (and DTM in particular) has to be carefully investigated in this process. 
Acknowledgements 
The SPOT data under a SPOT-IMAGE(R) license and the Landsat TM data are acquired 
in collaboration with EUROSENSE BELFOTOP N.V., Belgium. Many thanks are due to 
C. Steenmans and M. Martens for providing us with their expertise and the GIS data. The 
hydrography related work has been largely implemented by D. Claus and T. Creemers. 
References 
[1] F. Fierens, J. Van Cleynenbreugel, P. Suetens, and A. Oosterlinck. I-see : an Al-tool 
for image understanding. In Proceedings Workshop on Tools for AI, pages 314-320, 
IEEE, Fairfax, VA, October 1989. 
[2] M. Kass, A. Witkin, and D. Terzopoulos. Snakes: active contour models. International 
Journal of Computer Vision, 1(4):321—331, 1987. 
[3] A.N. Strahler. Physical Geography, pages 370-388. John Wiley &; Sons, second edition, 
1960. 
[4] P. Suetens, P. Fua, and A.J. Hanson. Computational strategies for object recognition. 
September 1989. submitted for publication to ACM Computing Surveys. 
[5] E.R. Tello. Object-oriented Programming for Artificial Intelligence, pages 196-214. 
Addison-Wesley Publishing Company, Inc., 1989. 
[6] J. Ton, A.K. Jain, W.R Enslin, and W.D. Hudson. Automatic road identification and 
labeling in Landsat 4 TM images. Photogrammetria, 43(5):257—276, 1989. 
[7] J. Van Cleynenbreugel, F. Fierens, P. Suetens, and A. Oosterlinck. Delineating road 
structures on satellite images by a GIS-guided technique. Photogrammetric Engineering 
and Remote Sensing, 56(6), June 1990.