Detection of major changes in satellite images

We address in this paper the probleni of detecting changes between two photographs of the same scene taken at different dates. Considering that both images are already accurately registered our problem reduces to a) detect changes between corresponding pixels, and b) determine whellier these changes are a clue o f a major change in the scene. The solution ofa) involves the use of spectral invariant features, since we consider the general situation of comparing photapphs coming froin different spectral channels. The answer to b) implies the computation of an absolute threshold above which a region of the iniage will be detected as having "meaningful changes". In order to do that, we apply a recently introduced methnd yielding accurate false alarm rates for each detection. We show that this method pcrmits lo help significantly photo-interprets in their search for major changes in a scene. This will be illustrated by a smking application, the detection of the explosion ofthe AZF chemical plant in Toulouse (France) in Septemher2001