2004 
jde, 
tho- 
lave 
red 
The 
een 
ated 
g at 
be 
1ber 
the 
) of 
del. 
‘the 
atic 
hod 
and 
and 
such 
ural 
989, 
note 
INTEGRATED DEM AND PAN-SHARPENED SPOT-4 IMAGE IN URBAN STUDIES 
G. Doxani, A. Stamou 
Dept. Cadastre, Photogrammetry and Cartography, Aristotle University of Thessaloniki, GREECE 
gdoxani@hotmail.com, katerinoudi@hotmail.com 
Youth Forum 
KEY WORDS: Remote Sensing, SPOT Satellite images, DEM, orthorectification, fusion, urban, 3D visualization 
ABSTRACT: 
Digital orthoimages are becoming increasingly of interest as they provide up-to-date information for a wide range of applications. 
Facing the need for accurate spatial information this paper aims to the production of orthoimages and their further processing, as 
well as to the three dimensional visualization of the resulting images. The data used are one panchromatic image from Spot-3, with 
spatial analysis 10 m, and one multispectral from Spot-4, with spatial analysis 20 m. Both images depict the city of Thessaloniki, in 
Greece. In order to eliminate distortions arising from the topographic relief and improve the accuracy of orthoimages, a DTM 
(Digital Terrain Model) of the same area with grid size of 25 meters was used. It should be noted that the ground control points were 
collected by GPS measurements and two orthophotos, which were produced from 1:10,000 and 1:20,000 aerial photos. Furthermore, 
the panchromatic orthoimage and the multispectral orthoimage were fused with the methodology of principal components analysis. 
Prior to the fusion process, the registration of the image with the lower spatial analysis (the multispectral) on the image with the 
higher spatial analysis (the panchromatic) was necessary. In order to analyze the reliability and the accuracy of the spectral and 
spatial information of the synthetic image an evaluation was accomplished based on statistical tests. Additionally, supervised 
classification was made in the synthetic image and its accuracy was evaluated. Finally, a panoramic video was created which 
included a supervised flight through the area of interest for the virtual representation of the anaglyph. 
1. INTRODUCTION beginning, choosing the 7 most important land cover and use of 
the area of interest. 
As the demand for digital products is getting bigger, the digital 
orthoimages are becoming more and more popular these days. Finally the 2D and 3D image data are elaborated with a view to 
Once the orthoimages are free from distortions arising from the produce various forms of visualization. 
topographic relief, they are more accurate and could be used 
diversely. 
2. DATA 
In order to eliminate geometric distortions of the images 
obtained by satellites is essential to alter their geometric form. The used data are one panchromatic image from Spot-3, with 
This is accomplished through the process of geometric spatial analysis 10 m and one multispectral from Spot-4, with 
correction by using one geometric model, in this case the Spot spatial analysis 20 m. Both images depict the city of 
geometric model. The result is the registered digital image. Thessaloniki, in Greece. Also a DTM (Digital Terrain Model) 
Subsequently, the resampling of the registered image, which of the same area with grid size of 25 meters is used. 
calculates the new pixel values with the method of bicubic 
interpolation, is necessary for the production of the orthoimage. Ground control points collected by GPS measurements and two 
After this procedure the central perspective image is orthophotos, which were produced from 1:10,000 and 1:20,000 
transformed into an orthogonal projection. aerial photos. 
To broaden the uses of orthoimages with different radiometric, For the development of this paper is used the software of the 
geometric and spectral characteristics, is important to fuse all Erdas Imagine 8.5. 
this information and produce a synthetic image. A synthetic 
image combines the spatial resolution of the high spatial 
analysis image and the spectral resolution of the one with the 
high spectral analysis. Its characteristics make possible the full 
exploitation of the image data. 
Additionally, from the classical cartographic point of view, is 
considered essential the classification of the image pixels into 
different types of classes, representing specific kinds of land. 
The result of this procedure is a thematic map, which provides 
useful information. In this paper, with the method of supervised 
classification, the types of classes were determined from the 
139