Underwater image segmentation with co-saliency detection and local statistical active contour model

Abstract

Listen

Translate article

Image segmentation of underwater environment with inhomogeneous intensity turns out to be one of the most challenging topics these years. In this paper, we try to combine co-saliency detection with local statistical active contour model together for underwater image segmentation. The cluster-based algorithm is first taken for co-saliency detection, which makes salient region in the underwater images be highlighted. The local statistical active contour model, a novel region-based level set method, is then made full use of to segment underwater images. It is shown in our simulation experiment that our proposed scheme could achieve great segmentation performance in both efficiency and quality for underwater images.