Conclusion

In digital fusion for image processing, efforts in recent years have led to a better understanding of the various theories from other fields, such as belief functions and fuzziness. We now know the right frameworks for applying these theories, their strengths and limitations in image processing, how to represent and model digital, symbolic or structural information and data in each of these formalisms, and how to combine them. Numerous developments have been made, particularly for the typical applications of multisource image classification and recognition of structures or objects in images.

The points still to be developed concern :

• conflict management, which is often difficult to distinguish from the complementarity of sources, and for which it is not always easy to know whether it should be resolved or not;...