Especially, in the proximity of the buildings, this situation 
causes interference in the segmentation phase. In the first step, 
classes were assigned and the convenient criteria were selected 
to include the segment in those classes. Results of the 
classification procedure are shown in Fig. 4. 
re ) S ^ 
T E i y 
A eM 
| cost 
road 
buildings white rocf 
e se: 
$9 rubbish, area 
Q non dassified 
Figure 4. Result of object-oriented classification 
Regarding the results gained from the created class hierarchy, 
most of the buildings and roads could be identified. However, 
manual revision of the classification could not be avoided and 
the objects that are misclassified with buildings and roads 
should be manually erased from these classes. Classification 
quality seems strongly depends on the quality of the initial 
segmentation and the DEM information used in the generation 
of pan-sharpened image. In this case, the geometrical shift and 
noise of DEM data used should be taken into consideration. 
Based on the classification results, eCognition software can 
produce statistical information for the users. Table has an 
emphasis because of it shows the error matrix in addition to the 
different accuracy values. Kappa of 0.84 shows the results suits 
with the expectation, however, for more reliable results suitable 
vector layers can be additionally be used. 
5. CONCLUSIONS 
Because of its high spatial resolution, Ikonos data is well suited 
to extract buildings and roads. To take advantage of its spectral 
properties, principal component image enhancement method 
can be used. In this case, image with 1m ground pixel size, but 
covering four spectral channels can be generated. It was seen 
that object-oriented analysis technique can reveal satisfied 
result for extracting the main land objects, e.g. roads and 
buildings. 
6. REFERENCES 
Benz, U. C., Hofmann, P., Willhauck, G., Lingenfelder, I. and 
Heynen, M. 2003. Multi-resolution, object-oriented fuzzy 
analysis of remote sensing data for GIS-ready information, 
   
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
ISPRS Journal of Photogrammetry & Remote Sensing, 58 
(2004) pp. 239-258 
Buyuksalih, G., Kocak, G., Oruc, M., Akcin, H. and Jacopsen, 
K. 2003. Handling of Ikonos Images from Orientation Up to 
DEM Generation. Proceeding of the Workshop on Mapping 
from Space 2003, Hannover, (on CD-ROM) 
^ 
eCognition User Guide 3. 2003. Definiens Imaging, pp.3.2-108 
Hofmann, P., 2001a. Detecting buildings and roads from 
IKONOS data using additional elevation information. In: GIS 
Geo-Information-System, 6/2001. 
Hofmann, P., 2001b. Detecting informal settlements from 
IKONOS image data using methods of object oriented image 
analysis - an example from Cape Town (South Africa). In: 
Jürgens, Carsten (Editor): Remote Sensing of Urban Areas 
Fernerkundung in urbanen Räumen.  (-Regensburger 
Geographische Schriften, Heft 35), Regensburg. 
Hofmann, P., 2001c, Detecting urban features from IKONOS 
data using an object-oriented approach. In: RSPS 2001, 
Geomatics, Earth Observation and the Information Society, 
2001 
7. ACKNOWLEDGEMENTS 
Parts of the presented results have been supported by 
TUBITAK, Turkey and the Jülich Research Centre, Germany. 
The authors wish to thank Serkan KARAKIS for his help during 
this study. 
        
  
    
   
   
   
   
    
   
   
  
  
   
    
  
   
   
     
     
    
   
  
  
   
   
    
    
   
   
  
  
  
   
   
     
  
  
  
  
  
  
  
  
   
    
   
  
   
State 
KEY \ 
ABST! 
In this 
images 
memor 
and en 
as self 
nets, is 
of aeri; 
charac 
prelim 
clonal 
pathog 
Paralle 
algorit 
Drawi 
researc 
up. Th 
and ac 
last fe 
artifici 
immur 
tasks | 
Zuben 
few ay 
this pa 
analys 
In ren 
impro 
classif 
for re 
Howe 
require 
all cla: 
diversi 
often I 
the sh 
system 
Comp: 
classif 
the a 
follow 
adapti 
withou 
form 
functic 
functic