Modified Graph-Cut Algorithm with Adaptive Shape Prior

Abstract

In this paper, a new method to adaptively apply shape prior in “modified graph-cut segmentation technique” has been proposed. The modified graph-cut technique has a better performance in terms of speed and accuracy when compared to the conventional graph-cut approach, introducing adaptive shape prior to this novel (modified) graph-cut approach yields a more efficient and effective result. Adaptive shape prior takes care of noise or object occlusion in a graph-cut segmentation process, it can be realized via a shape probability map, whose presence helps to showcase regions where the presence of a shape is required in an image. If employing adaptive shape prior to a conventional graph-cut technique yielded a better result than the classical approach, it is evident that applying adaptive shape prior to a modified graph-cut would yield a far better result.