'- N '■ 
f evati °n should 
Z®, the î 
¡.As Figure h 
is 1 pixel; and 
nore I 
line and p’i 
ecision can be 
lines, zoom m 
vs . These lines 
;econ(i °“ 
0 w coordinate, 
1.7 pixels and 
ïerence in r0 ' v 
coordinate is 7.3 pixels. Therefore, the epipolar pairs do not 
exist for the entire scene, but exist locally. This conclusion 
consists with the property of epipolarity derived by Kim (2000). 
Thanks for the supporting from the 973 Program of the People’s 
Republic of China under Grant 2006CB701302 and the 
National Natural Science of China under Grant 407721001. 
The number of epipolar line on let scene 
Figure 6. Accuracy of epipolar lines on the left scene 
. REFERENCES 
Cho, W., Schenk, T., and Madani, M., 1992. Resampling digital 
imagery to epipolar geometry. International Archives of 
Photogrammetry and Remote Sensing, 29(B3), pp. 404-408. 
Sohn, H., Park, C., and Chang, H., 2005. Rational function 
model-based image matching for digital elevation models. 
Photogrammetric Record, 20(112), pp. 366-383. 
Hu, Y., and Tao, V., 2004. Understanding the rational function 
model: method and applications. 
http://www.geoict.net/Resources/Publications/IAPRS2004_RF 
M2394.pdf (accessed 16 Feb. 2008) 
Figure 7. The difference among the seven epipolar lines 
Kim, T., 2000. A study on the epipolarity of linear pushbroom 
images. Photogrammetric Engineering and Remote Sensing, 
62(8), pp. 961-966. 
5. CONCLUSIONS 
After analysing the epipolar geometry of linear array scanner 
scenes a method which can be used to generate the approximate 
epipolar line for linear array scanner scenes based on the 
forward and inverse transform of RFM was developed. The 
method proposed does not require DTMs or ground control 
points during the generation, but it is better to know the range 
of elevation of the area the scenes covered which can improve 
the accuracy of the epipolar line generated. A stereo pairs of 
IKONOS imagery are used to validate the feasibility of this 
method; the epipolar generated can achieve a precision better 
than 1 pixel which can meet the need of image matching. The 
existence of epipolar pairs are also discussed, the result of the 
experiment shows that the epipolar pairs do not exist in the 
range of entire scene, but exist in a small range. 
Orun, A.B., and Natarajan, K., 1994. A modified bundle 
adjustment software for SPOT imagery and photography 
tradeoff. Photogrammetric Engineering and Remote Sensing, 
60(12), pp. 1431-1437. 
Okamoto, A., and Fraser, C., 1998. An alternative approach of 
the triangulation of SPOT imagery. International Archives of 
Photogrammetry and Remote Sensing, 32(4), pp. 457-462 
The epipolar lines generated in this paper can be used in image 
matching, but the epipolar lines generated in this paper are 
approximated, the true correspondence point may not lie on the 
line but locate near from it. So the searching area for 
correspondence should be defined not just along the epipolar 
line but should be wider than it. 
Recommendations for future work include performing more 
experiments using more data such as SPOT and QuickBird