Spatial Information in Fusion Methods - Information Fusion in Signal and Image Processing - page 191

Figure 9.1. Two MRI acquisitions of the brain and the result
of the classification by Markovian fusion
fusion is performed on an intermediate level, on primitives obtained by these detectors
(segments or areas), and therefore which contain spatial information.
Figure 9.2 illustrates an example of an ERS-1 radar image of the Aix-en-Provence
region.
Figure 9.3 shows the results of three of the detectors applied to this image.
The results of these detectors serve as the basis for modeling using belief functions.
This theory is particularly well-suited here, since it allows us to model the behavior of
each detector. Thus, road, slope or river detectors are precise for the objects they are
designed to study, but are not capable of distinguishing the other classes. As a result,
the focal elements of a road detector are the class “road” and the set “non-road” (hence