After a presentation of the basic data taken into account in information fusion, we will present various data-fusion operators from:

• probability theory (Bayes rule),
  
• possibility theory (T-norms, adaptive operators, priority operators).
  Several fusion operators having different fusion behaviour are available in possibility theory. The main basic types of operators are the following :

• Conjunction operators (such as T-norms) act strictly for reliable and sources which agree with each other,
• Disjunction operators (such as T-conorms) act with precaution for conflicting sources,
• and between the two, we find average type operators.
  
• belief theory (Dempster-Shafer's orthogonal rule).

Each of these 3 centers of interest is composed of a theoretical part and an application part.

The application part deals with practical applications of fusion methods in the aim of improving classification in LANDSAT satellite images. Fusion operators need to be adapted for use in image processing, in particular the possibility theory operators. We present the constraints of information representation specific to satellite imagery, and the modifications required to implement fusion operators.