Unsupervised change detection in multisource and multisensor remote sensing images

An automatic approach to unsupervised change detection on multisource and multisensor remote sensing images is proposed. Such an approach, unlike classical ones, allows the fusion of different information sources (e.g. optical data, SAR data) in the unsupervised change detection process. The fusion process is carried out according to a consensus theoretic approach by integrating the estimates of statistical terms achieved on the difference images associated with different sensors. The estimates of the statistical terms of each difference image are carried out by using a novel nonparametric and unsupervised technique.