Research project T4/DD/42 (Research action T4)
This project aims at the interpretation of multi-sensor information within monitoring and warning systems. The diversity of sensors that observe the earth all give a unique picture of a certain phenomenon, as well in ground resolution as in the spectral range of the imagery. One sensor by itself is inadequate to give a complete interpretation of the observed scene. Having data available from complementary sensors permits a more accurate and detailed analysis.
One of the main factors which hinder the widespread use of multi sensor analysis is the spatial registration of several sensor images within a common geometric framework. Today's techniques are still based on manual identification of ground control points and are therefore very labour intensive. Especially in applications like monitoring, the time spent on processing and the amount of data are critical, and automating the registration process is certainly of importance. Automation however is not evident when using multi sensor data. Ground control points can differ strongly when seen on images with different ground resolution. Conventional control points (e.g. crossroads) that are being used in high-resolution images may not be visible on low-resolution images. On the other hand, possible ground control points (e.g. a cape) in a low-resolution image can have a different appearance in a high-resolution image.
The theme of this project aims at the automation of the registration process, by extension of the ground control point to the concept of "ground control object". This concept groups points (e.g. landmarks), curves (e.g. coastline, rivers) and areas (e.g. urban areas). The goal of this extension is to ease the registration process by exploiting the spatial information coupled with an object, namely the point co-ordinates, together with the shape and the structural features of the object. For example, not only a cape but also the characteristic shape of the entire coastline can be used to guide the registration process. This permits to compensate for :
- the sensor ground resolution, by the characterisation of the scaling effect, hereby modelling the evolution of a structure over several ground resolutions, and
- the spectral characteristic, by using structural features instead of intensity based (and therefore sensor specific) features.
The monitoring of drought stress in Mediterranean and semi-arid area is an urgent problem, and with this the monitoring of wild fires. Every year an area of about 5000 km2 is destroyed by fire in the Mediterranean region (Kailidis, 1992).
It is aimed to set up an operational fire fighting plan using multi sensor data, to:
1. monitor the drought stress of plants at "global? level (national),
2. localize the fires at "global? level (national),
3. monitor the spreading of fires at national to regional level,
4. map the fuel type at regional to local level.
This operational fire fighting plan must allow acquiring more knowledge about the wild fires control problem, as it concerns the prevention of fires and the possibility of fire spreading.
Kailidis D.S. (1992b). Forest Fire in Greece. In: Seminar for forest fire prevention, land use and people. Ministry of Agriculture, Secretariat general forests and natural environment, Athens, (Greece). pp. 27-41
