ul 2004 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
  
In the second occasion where there was no 
histogram matching the image that came up is the 
following (fig.4). 
  
  
  
  
  
Figure 4. Synthetic image 2. Municipality of 
Thessaloniki. 
The software ERDAS IMAGINE 8.5 was used for 
the processing of the images while the GCP'S 
mensurations were actualized by the Leica 500 
receiver (with the AT502 antenna) and their 
processing was actualized by the LEICA SKI-PRO 
and ISTOS — 2000 programs. 
2.2.2 Evaluation of spectral quality of the 
synthetic image: In a merging procedure it is 
extremely important to maintain the primitive 
spectral information in the final image as well. For 
this reason the synthetic images were studied for 
their spectral accuracy, through statistical criteria 
such as mean values, typical variations and 
coefficient correlation. 
Analyzing the two synthetic images led to the 
following conclusions: 
With a simple visual observation of the two 
synthetic images one can easily discover that the 
synthetic | image has better appearance. The results 
of the spectral examinations through statistical 
criteria clearly indicate as better method of merging 
the one where histogram matching between the 
layers of the multispectral and the layer of the 
panchromatic is attained, that is the synthetic | 
image (Tab. 5). 
  
  
  
  
  
  
Statistical 
criteria 
Synthetic 1 Y 
Synthetic 2 
  
Table 5. Evaluation of the two merging methods 
used after examining their spectral accuracy. 
3. SYNTHETIC IMAGE - APPLICATIONS 
The second stage regarded to the applications that 
took place in the synthetic image like: 
a) Visual interpretation and 
b) Classification 
For the classification of the synthetic image and its 
estimation, the ERDAS IMAGINE 8.5 software was 
used. 
3.1 Visual interpretation of the synthetic 1 image 
With the assistance of cartographic data an attempt 
to improve tracking down urban characteristics (e.g. 
hospitals, buildings of administration, monuments 
etc.) via visual interpretation takes place. Here are 
some examples (Fig.6,7,8) 
  
Figure 8. Prefecture of 
Thessaloniki 
Figure 7. Hospital 
Improvement of tracking down urban characteristics 
can be useful for creating city plans that 
communicate better with the user. It can also 
become a basis for a G.LS. application. 
3.2. Classification of the synthetic 1 and 
estimation of the classification 
3.2.1 Classification: The type of classification 
selected for the synthetic 1 image was supervised 
and was accomplished with the maximum likelihood 
method. The selection and receipt of the samples 
was based on the V.LS. model (Vegetation- 
Impervious surface-Soil),(Ridd 1995, Ming C. H. 
2002). Estimation of the classification through 
close examination of the error matrix and the Kappa 
statistics followed. The selected classes were: 
* Grassland, trees 
- healthy green grass vegetation and tree 
and/or shrub vegetation. 
* Buildings 
- bright impervious surface, like rooftops, 
metals and tiles. 
* Streets 
- medium impervious surface, like concrete 
and weathered asphalt.