Disjunctive fusion, contrary to conjunctive fusion, assumes that there
is only one reliable source among the sources to be fused, but we do not
know which. This assumption is not valid in the case of satellite images
because, as seen previously, the spectral bands of an image, as well as
the out-image data, are concordant most of the time, and can be taken as
reliable.
Disjunctive fusion should not thus be efficient for the fusion of satellite
images. However, as it occurs in the adaptive fusion used in the following
pages, it is useful to observe its behaviour here.
Principle
Let us note again that disjunctive fusion consists in applying, for a pixel x, the operator ``maximum'' onto the degrees of possibility of the membership of pixel x to a class c.
Those degrees are provided by the set of the p sources :
|
(55) |