3. Istanbul 2004 
  
urea 
;hing stereopairs 
many features, 
passes and even 
s are apparent, 
he test area, the 
. Even with the 
ible to recognise 
Ss of high ground 
t this image has 
e to the fact the 
  
t area 
d in perspective 
t is clear to the 
sentation of the 
graphic features 
if any, obvious 
, 
ferent geometric 
olution satellite 
jures have been 
nts, triangulating 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
individual points, and performing a visual analysis. 
Consequently these results show that the object space geometric 
constraint (based on the affine projective model) is certainly as 
good as the image space geometric constraint (based on the 
epipolar model). An advantage of the affine-based constraint is 
that the matching parameters can be specified in ground 
coordinates. If approximate terrain heights are known, the 
matching search space can be specified very accurately indeed, 
and hence processing time and the number of potential errors 
can be reduced. (With the image space constraint, the matching 
parameters are much more arbitrary.) Additionally, these results 
further confirm the usefulness of the affine projective model in 
geometric processing of high resolution satellite imagery. 
Images not aligned to epipolar coordinates (the San Diego data) 
have also been successfully matched by use of a quadratic 
epipolar model. Since the results of this model were as good as 
those from the affine model, it can safely be assumed that use 
of the quadratic epipolar model is justified. This is an important 
result since it means that a high resolution image pair (i.e., not a 
stereopair) can be matched as successfully as a stereopair. 
Success of these empirical matching constraint models is 
largely contingent on the images being free of scanning non- 
linearities. Fortunately Ikonos appears to be largely free of such 
effects. Future work will investigate the possibilities of using 
multi-temporal, same-sensor images for terrain modelling, as 
well as multi-temporal, multi- sensor images. 
8. ACKNOWLEDGEMENTS 
This work was carried out while the author was a research 
fellow at the Department of Geomatics, University of 
Melbourne. The author would like to thank Prof. Clive Fraser, 
Dr. Harry Hanley, Mr. Simon Cronk and Mr. Takeshi 
Yamakawa for their assistance with various aspects of the 
research. 
9. REFERENCES 
Fraser, C. S., Baltsavias E. P., Gruen, A., 2001. Ikonos Geo 
stereo images: geometric potential and suitability for 3D 
building reconstruction. In: Photogrammetrie-Fernerkundung- 
Geoinformation: Geodaten schaffen Verbindungen, E. Seyfert 
(Ed.), Publications of the German Society for Photogrammetry 
and Remote Sensing, 10: 113-121. 
Fraser, C. S., Baltsavias E. P., Gruen, A., 2002a. Processing of 
lkonos imagery for submetre 3D positioning and building 
extraction. /SPRS Journal of Remote Sensing, 56(3): 177-194. 
Fraser, C. S., Hanley, H. B. and Yamakawa, T., 2002b. Three- 
dimensional geopositioning accuracy of Ikonos imagery. 
Photogrammetric Record, 17(99): 465-479. 
Gonzalez, R. C. and Woods, R. E. (1992). Digital image 
processing. Addison-Wesley, New York, 716 p. 
Luong, Q. T. and Faugeras, O., 1996. The fundamental matrix: 
Theory, algorithms and stability analysis. /nternational Journal 
of Computer Vision, 17(1): 43-76. 
Orun, A. B. and Natarajan, K., 1994. A modified bundle 
adjustment software for SPOT imagery and photography: 
tradoff. Photogrammetric Engineering & Remote Sensing, 
60(12): 1431-1437. 
Samadzadegan, F., Azizi, A., Lucas, C., Hahn, M., 2002. 
Automatic Surface Reconstruction Based on the Fusion of 
Fuzzy Logic and Robust Estimation Techniques. 
Photogrammetric Record, 17(100): 651-674. 
Zhang, L., Pateraki, M., Baltsavias, E., 2002. Matching of 
Ikonos Stereo and Multitemporal GEO Images for DSM 
Generation. Proceedings of Map Asia 2002, Asian Conference 
on GIS, GPS, Aerial Photography and Remote Sensing, 
Bangkok, Thailand (on CD-ROM).