International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
  
Table 8. The error matrix, the overall accuracy and the kappa 
coefficient of the two classified image parts 
Given that the ideal value for the overall accuracy is less than 
70-75% and for the Kappa coefficient less than 75% it should 
be noted that the classification has a high accuracy (Overall 
Accuracy of 84.43% and 81.25% and Kappa coefficient of 
78.33% and 75.00% for each case). 
6. VISUALIZATION 
The interest in observing the environment was the motive for 
the development of the visualization of 3D objects. 
Visualization aims at examining the earth's surface with 
perspective by combining remotely sensed data and Digital 
Terrain Models. Perspective visualization assists in a better 
understanding of complex geographical areas and geological 
structures. For this reason, in the last few years it has been used 
in many applications, such as 
Urban planning: the geometry and the planning of an 
engineering project (a bridge or a dam) and the visual 
effects of planned reconstructions can be examined 
through visualization. 
Environmental studies: the ecological impacts of a 
proposed project can be thoroughly examined. 
Telecommunication applications: the morphology of 
the terrain has an important effect in the signal's 
transmission, so the 3D simulation of the anaglyph 
contributes in the establishment of telecommunication 
network stations (Graf, 1995). 
6.1 Visualization of the DTM 
A Digital Terrain Model can be visualized in different ways 
through terrain analysis, which involves the processing and 
graphic simulation of elevation data. In the following Figures 
some images are presented which are the products of the terrain 
analysis: 
— Shaded relief image: illustrates variations in terrain by 
differentiating areas that would be shadowed by a 
light source simulating the sun. 
Slope image: illustrates changes in elevation over 
distance 
Aspect image: illustrates the prevailing direction that 
the slope faces at each pixel (Erdas, 1999). 
  
Figure 9. A painted shaded relief, which is created by splitting 
the elevation data into 25 equal levels and assigning 
a distinct colour to that level, draped over the DTM. 
190 
  
  
  
  
  
  
  
Figure 10. The slope (a) and the aspect image (b) 
6.2 Visualization of the pan-sharpened image 
The textural properties of an image can be useful in many 
applications, such as classification. Certain algorithms that 
detect textural features in imagery have been developed 
including contrast, energy, entropy, homogeneity and variance. 
In this study the variance operator was used, which is expressed 
as: 
> (x, - M) 
n-]1 
(1) 
Variance — 
where 
x; = value of pixel (i, /) 
n = number of pixels in a window 
M = Mean of the moving window 
  
Figure 11. Textural features detected by a 3X3 window of the 
variance operator. a) Original and b) Texture image. 
A window of 3X3 dimensions was used and the resulting 
image highlights the edges of the buildings, streets and 
vegetated areas. 
7. CONCLUSIONS 
This study focused on the production of accurate orthoimages 
as well as their further processing. During the orthorectification 
of the panchromatic and multispectral image a certain 
methodology was adopted so as to derive reliable results. It was 
shown that by improving the accuracy of the DTM, refining the 
RFCs and processing the image separately in three parts the 
error of the orthorectification was significantly reduced (from 
3510 35m).