The detector based on the MAP criterion chooses the most
likely CI among all possible CIs, given the observed images.
The resulting probability of error is minimum among all other
detectors [12], [13]. The structure of the MAP detector is based
on statistical models. Therefore, the accuracy of the statistical
models for the given images as well as for the CI is crucial. In
this paper, we assume that the given images are obtained by the
summation of noiseless image models (NIMs) and image noises.
Both the NIMs and the CI are assumed to have MRF properties.
Furthermore, we assume that configurations of the NIMs
are equal for unchanged sites. In addition, the configurations of
changed sites from one NIM are independent of configurations
of changed sites from the other NIM when the configurations
of unchanged sites are given. Then, the a posteriori probability
is determined, based on the above assumptions, and the MAP
criterion is used to select the optimum CI.