Satellite image fusion using Dezert - Smarandache theory

Free and hybrid models of multisource satellite images fusion are developed using the plausible and paradoxical reasoning theory of Dezert-Smarandache. The aim of this work is to show the contribution of these fusion models for improving the thematic classification and the quantification of change. The maps obtained by the free model are composed by simple classes and compound classes. Nevertheless, they contain no significant thematic classes and require an important computing time. In the other hand, the hybrid model with a constraint introduced using a prior knowledge relatively of the study area, can have maps composed of more realistic classes in a reduced time. These models are implemented and tested on images acquired by SPOT HRV and Landsat ETM+ sensors.