A global architecture for temporal changes analysing system using satellite images and cartographic data

The authors present the development of a temporal change analyzing expert system. The proposed system acts in three sequential steps. First, the dataset given in input is analyzed and depending on the data nature, a change detection method is selected. Second, the change detection method is applied and a fuzzy changes image is created. Changes are detected at a pixel level. Third, the nature of the changes is interpreted using a rule-based reasoning approach. Here the changes are identified at a polygon level. The authors present their method by analyzing a dataset composed of a base forest map at time T/sub 0/ (1993) and a Landsat-TM image at time T/sub 1/ (1995) in the context of a forest map updating paradigm.