A graph-cut based algorithm for approximate MRF optimization

Aymen Shabou, Florence Tupin, Jérôme Darbon

Abstract: This paper copes with the approximate minimization of Markovian energy with pairwise interactions. We extend previous approaches that rely on graph-cuts and move making techniques. For this purpose, a new move is introduced that permits us to perform better approximate optimizations. Some experiments show that very good local minima are obtained while keeping the memory usage low.