GROUP-MEMBERSHIP REINFORCEMENT FOR STRAIGHT EDGES BASED ON BAYESIAN NETWORKS

A probabilistic approach to edge reinforcement is proposed that is bas ed on Bayesian networks of two-dimensional (2-D) fields of variables. The proposed net is composed of three nodes, each devoted to estimatin g a field of variables. The first node contains available observations . The second node is associated with a coupled random field representi ng the estimates of the actual values of observed data and of their di scontinuities. At the third node, a field of variables is used to repr esent parameters describing the membership of a discontinuity into a g roup. The edge reinforcement problem is stated in terms of minimizatio n of local functionals, each associated with a different node, and mad e up of terms that can be computed locally. It is shown that a distrib uted minimization is equivalent to the minimization of a global reinfo rcement criterion. Results concerning the reinforcement of straight li nes in synthetic and real images are reported, and applications to syn thetic aperture radar (SAR) images are described.