Statistically independent region models applied to correlation and segmentation techniques
Philippe Réfrégier, François Goudail, Christophe Chesnaud · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1999
Recently new approaches for location and/or segmentation of objects with unknown gray levels embedded in non-overlapping noise have been proposed. These techniques are based on the Statistically Independent Region (SIR) model and are optimal in the maximum likelihood sense. In this paper, we review their theoretical bases and propose a unified approach which enlarges their field of application.