slope 
"ases 
differ 
more 
S the 
'S by 
/ and 
rther 
vias 
their 
1 the 
t any 
t. In 
he 6 
ssful 
f the 
(2) 
4 an 
se of 
nage 
v no 
rther 
ures. 
hich 
ate. 
  
Dimitris Skarlatos 
  
  
Figure 4. The application of the double image checking algorithm, with 7x7 fixed template size. 
Since this is a first report, other types of photography are to be used, with different scales and resolutions. Close range 
images will be used when the initial approximation problem would have been addressed. Combinations over the 
proposed methods in order to find the best way to match images without any a priori knowledge (relative orientation), 
with the highest reliability factor are the next step. Theoretically the re weighting in conjunction with adaptive template 
to be a good solution and therefore research will be focused towards that direction. Computational effort is no object 
since computer power raises geometrically. 
Check could also facilitate the affine parameters (Baltsavias, 1991) in terms of rotation, scale and xy distortions. In 
such case application of thresholds on these parameters should be avoided and an adaptive algorithm should be used. 
Other LS parameters which should be adapted automatically are: 
- the window search area, 
- the use of less than 8 parameters in each point, 
- dynamically reducing the parameters by fixing those which do not change a lot (research is already being done on 
this subject, but results haven't been analyzed yet), 
- the maximum number of iterations, 
- the minimum corrections in affine parameters to end the iterations 
It is expected that the use of the epipolar geometry would strengthen the matching. 
In author's opinion the LS method is still alive; all it needs is a little refinement (Skarlatos, 1999). 
ACKNOWLEDGEMENTS 
The author would like to acknowledge the financial support from State Scholarship Fundation. 
REFERENCES 
Kennedy, J. B., Neville, A. M., 1964. Basic Statistical Methods for Engineers & Scientists. Dun-Donnelley, New York, 
p. 490. Chapters 3 and 4. 
Calitz, M., F., Rither, H., 1996. Least absolute deviation (LAD) image matching. ISPRS Journal of Photogrammetry & 
Remote Sensing 51, pp. 223-229 
Cross, P. A.,1990. Advanced Least Squares Applied to Position-Fixing. Polytechnic of East London, London, p. 205. 
Baltsavias, E., P., 1991. Multiphoto Geometrically Constrained Matching. Ph.D. thesis, Institute for Geodesy and 
Photogrammetry, Zurich. 
Pilgrim, L., 1996. Robust estimation applied to surface matching. ISPRS Journal of Photogrammetry & Remote Sensing 
51, pp. 243-257. 
Shao, J., Fraser, C., 1998. Multi-Image matching for automatic object reconstruction, University of Melbourne, 
Australia. jsha 9 sli.unimelb.edu.au 
Zhang, B., Miller, S. 1997. Adaptive automatic terrain extraction. Helave Associates Incorporated, California, USA. 
Helava  gdesystems.com 
Gruen, A., 1985. Adaptive least squares correlation: A powerful image matching technique. South African Journal of 
Photogrammetry, Remote Sensing and Cartography. (14)3, pp.175-187. 
Atkinson, K., B., 1996. Close Range Photogrammetry and Machine Vision. Whittles, Caithness, UK, p. 371. 
Skarlatos, D., 1999. Orthophotograph production in urban areas. Photogrammetric Record, 16(94), pp643-650. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 849