5 CONCLUSION AND OUTLOOK 
It is quite clear that reasoning in scale space will be a key to 
come to a satisfying solution for edge detection and location. To 
approach the reasoning we analyzed simulated and real image 
data to gain insight into problems of tracking and matching in 
multiscale images. Aspects due to the tracking of edges located 
by the Fórstner Operator are observed and discussed. Those 
edges which can be tracked through a large scale range are most 
important for solving the matching process. They may serve for 
the elimination of spurious edges and give approximate informa- 
tion for matching edges which are missed after smoothing with 
a small scale parameter. Further we expect that other knowledge 
like smoothness of the displacement field will be a complemen- 
tary and helpful information, suited to support the tracking and 
matching in combination. 
REFERENCES 
Ackermann, F., et al. (1989): MOMS-02-D2 Ein multispektrales 
Stereo-Bildaufnahmesystem für die zweite deutsche 
Spacelab-Mission D2. In: Geoinformationssysteme 3, S. 
5-11. 
Ackermann, F., Hahn, M. (1991): Image Pyramids for Digital 
Photogrammetry. In: Ebner, H. et al.: Digital Photogram- 
metric Systems, Wichmann-Verlag, Karlsruhe. 
Sester, M. (1990): Multiscale representation for knowledge 
Based recognition and Tracking of Objects in Image 
Sequences. Int. Arch. of Photogrammetry and Remote 
Sensing Vol 28 3/2, pp. 868-888. 
Bergholm, F. (1986): Edge Focusing. IEEE PAMI 9, No 6, pp 
726-741. 
Canny, J. (1986): A Computational Approach to Edge Detec 
tion, IEEE PAMI Vol 8, No. 6, pp 679-689. 
Fórstner, W., Gülch E. (1987): A Fast Operator for Detection 
and Precise Location of Distinct Points, Corners and 
Centres of Circular Features. Proc. of the Intercom. 
Conf. on Fast Processing of Photogrammetric Data, 
Interlaken. 
Fórstner, W. (1988): "Statistische Verfahren für die automati 
sche Bildanalyse und ihre Bewertung bei der Objekter- 
kennung und Vermessung", Habilitationsschrift, Uni 
Stuttgart. 
Hahn, M. (1990): Estimation of the Width of the Point Spread 
Function within Image Matching. Int. Arch. of Photo- 
grammetry and Remote Sensing, Vol 28-3/2, Wuhan, 
Peoples Rep. of China, pp 246-267. 
Hahn, M., Schneider, F. (1991): Feature Based Surface Recon 
struction - A Hierarchical Approach Developed for 
MOMS-02 Imagery. International Geosience and Remote 
Sensing Symposium, Vol 3, pp 1747-1750, Helsinki, 
University of Technology Espoo, Finland. 
Hahn, M., Fórstner, W. (1988): The Applicability of a Feature 
Based Matching and a Least Squares Matching Algo- 
rithm for DEM-Acquisition. Int. Arch. of Photogram- 
302 
metry and Remote Sensing, Vol 27, B9, pp 321-331. 
Heikkild, J. (1989): Multiscale Representation with Fórstner 
Operator. The Photogrammetric Journal of Finnland. Vol 
11, No. 2, pp 40-54. 
Lindeberg, T., Eklund, J-O. (1990): Scale Detection and Region 
Extraction from Scale-Space Primal Sketch. Proc. 3rd 
Int. Conf. on Comp. Vision, Osaka, pp. 416-426. 
Lu, Y., Jain R. L. (1992): Reasoning about Edges in Scale 
Space. IEEE PAMI 14, No. 4, pp. 450-468. 
Marr, D. (1982): Vision. San Francisco: W. H. Freeman. 
Witkin, A. (1983): Scale Space filtering. Proc. Int. Joint Conf. 
Artif. Intell., Karlsruhe, pp. 1019-1021. 
Yuille, A., Poggio, T. (1986): Scaling Theorems for Zero Cros- 
sings, IEEE PAMI, Vol 8, No. 1, pp. 15-25. 
Int 
base 
jor ı 
estir 
lyin, 
from 
acqu 
the 
The 
whic 
will 
extr 
is CO 
The: 
non 
com 
mot; 
cons 
tion 
dien 
as t|