Image co-segmentation using dual active contours

Abstract

In this article a novel algorithm is proposed to segment a pair of images simultaneously (co-segmentation) for extracting common objects. The task of co-segmentation has been performed using the dual geometric active contour model. Both the contours (of the objects) are initialized and evolved simultaneously in the given images. As the contours proceed towards the boundary of the common object(s) present in the images, energy value gets reduced. The contours are allowed to evolve until both the inner and the outer contours coincide at the object boundary. The resultant images formed are known as the co-segmented images. The proposed approach is evaluated on 20 benchmark datasets and compared with the state-of-the-art methods. Results show that the performance of the proposed method is better than the compared methods.