Change detection approach using evidential fusion of change indices

In this paper, we present fusion and classification process of change indices using multitemporal satellites images in the aim to detect the change of surface states after a flood. This process is performed in the framework of Dempster Shafer Theory (DST), which takes into account the imprecision and the ignorance related to data. We apply this process to a study site located at south west of England, traversed by Severn river, which have undergone in October 2000 an important flood. For the detection of the flood damage, we have used two change indices: difference values and texture evolution. We find that change index fusion overcomes the limits of change mono-index classification.