This result shows that, there is a systematic error about 8 
meters, which actually is the tree height. It can be seen that 
in north part of the model the systematic errors are reduced 
(since the density of trees is reduced), but the random 
errors are increased, due to land cover patterns. 
4.3 Evaluation of rectification via automatic generated 
DEM 
Both left and right images from the study area have been 
rectified using automatic generated DEM in TRASTER T10 
with module ORTHO-IMAGE then both were exported to 
ILWIS and 100 points have been measured in both of ortho 
images. 
The standard deviation of the differences in coordinates is 
about 11 m. Regarding the resolution of the images (10 m) 
reveals that this DEM are reliable for image rectification 
(See figure 4.8). With regard to the figure 4.8 it is obvious 
that in the north and north west the amount of errors is 
increased with respect to other parts of the figure and this 
is due to the complex pattern of the area. In fact this method 
has got some problems in the area with no distinct ground 
objects and also may have large blunders due to the 
patterns of the ground. Therefore by examining the figure it 
can be seen that on that particular area image correlation 
for determination of height may become confuse. 
5. Conclusion and Recommendations 
We have reviewed the earliest work in matching stereo- 
images based on the idea of finding a matching patch, using 
a correlation function, or finding matching feature in two 
images. Area correlation has been found to be slow, but 
capable of sub-pixel accuracy. Feature matching is faster, 
but usually does not give sub-pixel accuracy. In feature- 
based as it has been mentioned, the integrity of method are 
based on distinct objects, therefore in area which is empty 
of distinct certainly it cause severe blunders and this 
method is not suitable to use. 
Early work on automatic DEM generation was based on 
aerial photographs, which could be rotated with relative 
ease to use epipolar processing scan lines, but satellite 
imagery, captured over an extended processing time during 
which both the satellite and the Earth are moving, present 
geometric problems to be overcome at the same time as the 
matching process. 
Image maps based on digital ortho images provide great 
advantages in comparison to their analogue counterparts, 
especially with respect to the flexibility, easy of use and 
understanding, production of derived products and 
combination with other datasets: containing more details, 
intended to wide range of users, produced by computer, 
able to be applied in GIS. 
Tests performed at ITC have shown that a SPOT 
stereomodel, controlled by about 10 evenly distributed 
ground control points and rectified using an automatically 
generated DEM ensures sufficient planimetric precision for 
778 
image mapping at scale 1:50,000. But the automatic 
generated DEM does not provide enough precision and 
fidelity to topographic features for contouring with 10 to 20 
m interval. The remedial action was to supplement the DEM 
with morphologic information. 
Quality control of image maps with respect to the process 
involved is actually the controlling of the three effecting 
elements on the overall quality of final result: firstly the 
quality of input data such as image, digital elevation model 
and ground control points, and secondly the method, 
equipment and finally operator skill and care applied during 
the processes such as rectification, mosaicking, screening 
etc.. Apparently reaching high quality orthophoto is 
impossible unless all these elements are established. 
REFERENCES 
Ackermann, F. 1984,"Digital image correlation: performance 
and potential application in photogrammetry" 
Photogrammetric Record, 64(11): pp 429 - 439. 
Angleraud, C. & Becek, K. & Donnely, B. E. & Trinder, 
J.C.,1992,"DEM determination from SPOT", 6th 
AustrallianRemote Sensing conference, Wellington, New 
Zealand. 
Brockelbank, D. C. & Tam, A. P., 1991,"Stereo Elevation 
determination techniques for SPOT imagery", P.E. & R.S., 
57(8): pp 1065-1073. 
Carrol, D. 1987,"An analytical solution for stereo-images 
from the SPOT Remote Sensing Satellite", Master's Thesis, 
University of New South Wales, Sydney, Australia. 
Charif, M., #1987, "optimum sampling for digital terrain 
modelling", ISPRS, Commission Ill, ITC, Enschede, The 
Netherlands. 
Charif, M. 1987,"Digital Terrain Model" ITC lecturenote, 
Enschede, The Netherlands. 
Day, T. & Muller, J.P. 1987,"digital Elevation Model 
Production by Stereo Matching SPOT images-pairs: A 
comparison of Algorithms", Image and vision computing, 
7(2): pp 95-101. 
Denegre, J. 1994," Thematic mapping from satellite imagery, 
a guide book", International cartographic assosiation, Institut 
Geographique National, 136 bis, rue de grenelle, 75700 
Paris, France 
Dowman, |. 1988,"The restitution of metric photography 
taken from space", ISPRS Congress, Kyoto, Japan. 
Elmhorst, A. & Muller, W. 1988,"Generation of DTM with 
space photographs", ISPRS Congress, Kyoto, Japan. 
Estes, J. E. 1980."Attribute of well-trained remote sensing 
technologist", Proceedings, Conference of Remote Sensing 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
CD) => rnm 
WD > (n