International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
effect of matching and large remaining blunders, which 
cannot be avoided. That's why in the distorted case there was 
deterioration, while in the automated DEMSs, there was no 
noticeable difference. 
The error distribution plots show a noticeable improvement 
after applying the algorithm. The measure of this effect is 
kurtosis, which has been improved hugely. 
6. CONCLUSIONS AND FURTHER 
RESEARCH. 
From the arithmetic point of view, the main problem of the 
automatically collected DEMs is the dispersion from the real 
surface due to the matching and sampling of the ground. 
It is quite obvious that the method can calculate corrections 
in DEMs and improve their accuracy (RMS) by 37% (real 
data case), and their precision (SD, MAD) by 40% 
approximately. 
The most interesting fact is that the algorithm was able to 
improve MAD, SD and RMS, ending in the same values in 
any case, irrespectively of the magnitude of the initial error. 
This fact confirms the initial statement that the method does 
not need iterations to work. 
MAD has been reduced to the theoretical height accuracy of a 
single measurement. This is particularly promising especially 
if one considers that the comparison enclose a necessary step 
of interpolation, which deteriorates the results (Al-Tahir et 
al., 1992; Zhilin 1993b). 
What's makes the method even more attractive is the fact that 
it can be used in a number of cases such as: 
correction and creation of a more accurate DEM 
checking of automatically created DEMs 
updating of previously existing DEMs, using recent aerial 
photographs 
change detection based on activities concerning DEM 
change, such as road creation, quarry development, 
urbanisation, etc. 
Extensive tests on a number of different aerial or close range 
photographs, with different scale, created by different digital 
stereoplotters, hugely distorted DEMs etc, are currently 
running with promising results. 
Since the algorithm can effectively correct the DEM, these 
corrections can be used for checking. An efficient way to 
investigate the quality of the created DEMs is also under 
investigation. 
ACKNOWLEDGEMENTS 
Financial support from State Scholarship Foundation for a 
Ph.D. research for the first author, must be acknowledged. 
REFERENCES 
Al-Tahir, R., Schenk, T., 1992. On the interpolation problem 
of automated surface reconstruction. /nternational Archives 
of Photogrammetry and Remote Sensing, 29(3):227-232. 
Fórstner, W., 1986. On feature based correspondence 
algorithm for image matching and least squares matching, 
International Archives of Photogrammetry and Remote 
Sensing, Vol. 26, P3, Rovaniemi, pp. 150-166. 
Georgopoulos, A., Skarlatos, D., 2003. A novel method for 
automating the checking and correction of digital elevation 
models using orthophotograps. Photogrammetric Record, 
18(102):156-163. 
Un 
ISM, 1998. The fundamentals of Digital Photogrammetry. 
ISM International Systemap Corporation, Canada. 54 pages. 
Kraus, K., 1992. Photogrammetry. Volume I. Fundamentals 
and standard processes. Dümmler, Bonn. 397 pages. 
Norvelle, F., R., 1996. Using iterative orthophoto refinements 
to generate and correct digital elevation models (DEM's). 
Digital Photogrammetry: An addendum to the Manual of 
Photogrammetry. American Society for Photogrammetry and 
Remote Sensing. Bethesda, Maryland, USA. 247 pages:151- 
155. 
Norvelle, F.R., 1994. Using iterative orthophoto refinements 
to generate and correct digital elevation models (DEM). 
Proceedings: Mapping and remote sensing tools for the 21st 
century. American Society for Photogrammetry and Remote 
Sensing, Washington:134-142. 
Skarlatos D., 2000. Image matching towards maturity. 
IAPRS XXXIII, Amsterdam 2000, TP 111-03-03. 
Skarlatos, D., Georgopoulos, A., 2004. A new matching 
algorithm using elliptical areas: Results, accuracy, 
advantages and disadvantages. IAPRS XXXIV, Istanbul, 
2004. 
Sonka, M., Hlavak, V., Boyle, R., 1993. Image processing, 
analysis and machine vision. Chapman & Hall, London. 555 
pages. 
Zhilin L., 1993a. Mathematical models of the accuracy of 
digital terrain model surfaces linearly constructed from 
square gridded data. Photogrammetric Record, 14(82): 661- 
674. 
Zhilin L., 1993b, Theoretical models of the accuracy of 
digital terrain models: an evaluation and some observations. 
Photogrammetric Record, 14(82):651-660. 
Zhilin, L., 1988. On the measure of digital terrain accuracy. 
Photogrammetric Record, 12(72):873-877. 
KEY V 
ABSTI 
In ! 
techniq 
23 km 
topogra 
Thi 
points 
Results 
is satis 
particul 
control 
1:50.00 
1.1 In: 
Photogi 
the pos: 
digital « 
the first 
digital « 
pairs is 
source | 
other ve 
algorith 
the da 
uniform 
À digite 
from ste 
on DEN 
the oris 
availabl 
commoi 
Depend 
when tl 
Sensing 
from ri 
(Slama, 
Digital 
decades 
satellite: 
arise un 
satellite: 
SPOT ( 
1995), 
(Welch 
pushbro 
(Decem