International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
Following the segmentation with small scale values as applied 
in Figure 6 the classification is resulted in for a series of 
buildings as shown in Figure 8. Although the building blocks 
should have been separated from each other, the buildings are 
classified altogether apparently. The overlapping manually and 
automatic digitized objects are shown in Figure 9. 
It is too normal that, there can be some inconsistency between 
classifiers fail at separating two different buildings as one 
building as seen in Figure 9 marked with 1. Besides, some 
buildings can not be extracted from the image, although these 
constructions can be digitized manually. This situation can be 
seen in Figure 9 marked by 2. Originally these three buildings 
exist in the 1:1000 line maps. However, some buildings like in 
Figure 9 marked by 3 have the same size obtained by manual 
and automatic digitizing. 
5. CONCLUSION 
In this study, manual on-screen digitizing and the automatic 
object oriented image analysis methods have been compared 
using KVR-1000 orthoimage. By manual method, almost all 
building and road details that are available or not available 
could be derived. Although the effective pixel size of KVR- 
1000 orthoimage is about 2 pixel, experience and function of 
operator are the main factors on the success rate. However, 
accuracy of the coordinate transformation of about £12 m does 
not provide the required position accuracy. The reason for this 
is that the KVR-1000 orthoimage was generated by the DEM 
with 20 m height accuracy. As a rule of thumb 10 times of the 
pixel size gives the scale factor (Jacobsen, 2002). For KVR- 
1000 case 10 times of the pixel size is 15.6 m and this 
corresponds to 1:16000 map scale. Individual structures in a 
forest can be located significantly due to their distinct grey 
values using KVR-1000 images. Such a study was made by 
Karathanassi et al. (2003). Their concern was not the geometric 
accuracy of the classification. But our study has attempted the 
accuracy potential of from KVR-1000 image digitized vector 
maps. The study comes to the conclusion that pixel size does 
not dictate the map scale of end product to be extracted from 
the satellite images such as KVR-1000. 
Expected success rate could not be reached on the KVR-1000 
ortho-image using eCognition 3.0 object-oriented image 
analysis software not enough contrast, monochromatic image, 
and negative influence of DEM on orthoimage generation fails 
the segmentation phase, then the following classification 
produce did not work as efficient as possible. In contrast to 
automatic method, manual, method produced expected success 
for the object extraction purpose. 
6. REFERENCES 
Bjorgo, E. 2000. Using very high spatial resolution 
multispectral satellite sensor imagery to monitor refugee camps. 
International Journal of Remote Sensing, 21(2), pp. 611-616. 
Jacobsen, K., 2002. Mapping with IKONOS images. In: 
EARSeL Symposium “Geoinformation for European-wide 
Integration", Prague, Czech Republic, pp. 149-156. 
Karathanassi, V., lossifidis, Ch. and Rokos, D., 2003. Remote 
sensing methods and techniques as a tool for the 
implementation of environmental legislation. The Greek Forest 
118 
Law case study. International Journal of Remote Sensing, 
24(1), pp. 39-51. 
Kostka, R., 2002. The world mountain Damavand: 
documentation and monitoring of human activities using remote 
sensing data. ISPRS Journal of Photogrammetry & Remote 
Sensing, 57(2002), pp. 5-12. 
Parker, J.R., 1997. Algorithms for Image Processing and 
Computer Vision. Wiley Computer Publishing, pp. 3-5. 
Robinson, A.H., Sale, R.D., Morrison, J.L. and Muehrcke, P.C., 
1984. Elements Of Cartography. John Wiley and Sons Inc., 
Canada. 
SOVINFORMSPUTNIK, 2004. Information from official web- 
site, Moscow, Russia.  http://www.sovinformsputnik.com 
(accessed 18 Feb. 2004). 
Topan, H., Buyuksalih, G. and Jacobsen, K., 2004. Comparison 
of information contents of high resolution space images. In: XX. 
Congress of ISPRS, Istanbul, Turkey, this issue. 
Yan, G., 2003. Pixel based and object oriented image analysis 
for coal fire research. Msc. Thesis, International Institute. for 
Geo-Information Science and Earth Observation Enschede, The 
Netherlands. 
7. ACKNOWLEDGEMENTS 
Parts of the presented results have been supported by 
TUBITAK, Turkey and the Jülich Research Centre, Germany. 
The authors wish to thank Dr. Gurcan Buyuksalih, Dr. Karsten 
Jacobsen and M. Guven Kocak for their supervisions during this 
study. 
KI 
AE 
Fo 
po 
ma 
In 
7 = 
est 
err 
eff 
em 
act 
itei 
Co 
per 
Th 
fre 
elli 
pra 
Th 
Tre 
ort) 
the 
ge 
In 
the 
rea 
Th 
the 
sut 
sea 
The 
but 
elli 
a Si 
*