004 
USING THE DATA FUSION TECHNIQUE FOR PRODUCING THEMATIC MAP 
M.R. Della Rocca *, M. Fiani*, A. Fortunato, P. Pistillo* 
* Dipartimento di Ingegneria Civile, Università di Salerno, Via Ponte Don Melillo, 1, 84084 Fisciano (Salerno), Italy 
(r.dellarocca, m.fiani, a.fortunato, ppistillo)@unisa.it 
KEY WORDS: Remote Sensing, Fusion, Satellite, High resolution, Multisensor, Mapping 
ABSTRACT: 
The University of Salerno (Italy) has recently been equipped with a pair of aerials that are able to receive data transmitted from various 
satellite platforms such as the NOAA and the Terra-1 and Aqua-1 EOS. These satellites are fitted with sensors that pick up information 
regarding soil and cloud temperature, the humidity level of the atmosphere, the presence of extraneous water, the presence of certain 
substances such as chlorophyll. the surface colour of the ocean and the presence of plankton with a spatial resolution that at its greatest 
reaches 250 m. Great use is therefore made of a vast amount of data that concerns our geographical area which allows for a fairly 
approximate assessment of the state of health (desertification, deforestation, abundance of surface water) and the potential risks 
(landslides, fires, plant and crop infections or diseases) to which the territory is prone. Nevertheless, the data is supplied with differing 
precision to the ground and with different pixel dimensions, therefore it is not possible to generate, taking as a starting point the data 
alone, (thematic map) which, besides furnishing qualitative information, also supplies correct and precise metric information. By using 
the Data Fusion technique it is possible to project the data supplied by various sensors onto a higher resolution image in order to obtain 
a representation that allows for a synthesis of all available information. 
1. INTRODUCTION 3. Decision Level 
The University of Salerno (Italy) has recently been equipped with Processing and analysing data at Pixel level means working, as is 
a pair of aerials that are able to receive data transmitted from also shown in figure 1, directly on the signal supplied by the 
various satellite platforms such as the NOAA and the Terra-1 sensors and applying algorithms on the images without carrying 
and Aqua-1 EOS. out any further operation other than a simple radiometric or 
These satellites are fitted with sensors that pick up information geometric correction. In the other two cases however, pre- 
regarding soil and cloud temperature, the humidity level of the processing of the radiometric information is always required, and 
atmosphere, the presence of extraneous water, the presence of this information subsequently undergoes the process of Image 
certain substances such as chlorophyll, the surface colour of the Fusion. Figure 1 shows a diagram of the different methods. 
ocean and the presence of plankton with a spatial resolution that 
at its greatest reaches 250 m. Great use is therefore made of a 
vast amount of data that concerns our geographical area which 
allows for a fairly approximate assessment of the state of health 
(desertification, deforestation, abundance of surface water) and 
the potential risks (landslides, fires, plant and crop infections or 
diseases) to which the territory is prone. Nevertheless, the data is 
supplied with differing precision to the ground and with different 
pixel dimensions, therefore it is not possible to generate, taking 
as a starting point the data alone, thematic map which, besides 
furnishing qualitative information, also supplies correct and 
precise metric information. By using the Data Fusion technique it 
is possible to project the data supplied by various sensors onto a 
higher resolution image in order to obtain a representation that 
allows for a synthesis of all available information. The term Data 
  
Fusion is somewhat vague and does not fully describe the work Figure 1. Flow diagram of the different Data Fusion methods 
that it carries out. ^ sa matter or, *, every time one attempts to (from Pohl, 1999) 
give a definitive definition, one ends up by providing merely one 
aspect of it or one of its potentials, as has been shown by Wald 
(Wald, 1999). The methodologies and the techniques utilised to 2. IMAGE FUSION TECHNIQUES 
provide the results presented in this paper can in fact be traced 
back to one of its sub-sets, known as Image Fusion. The Hue Saturation Value (HSV) algorithm used for 
We can distinguish between three different degrees of processing experimental purposes is an Image Fusion type and works at 
in Data Fusion (Pohl C. 1999): pixel level in order to allow for the combination of high 
| resolution, panchromatic images with low resolution, multi- 
| |. Pixel spectral images. The operation is conducted by keeping the 
2. Feature spectral resolution while the spatial resolution of the lowest 
721