A site entropy rate and degree centrality based algorithm for image co-segmentation

• Published in: Journal of visual communication and image representation Vol. 33; pp. 20 - 30
• Main Authors: Chakraborty, Souradeep, Mitra, Pabitra
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.11.2015
• Subjects: Co-saliency; Co-segmentation; Degree centrality; k-partite graph; Markov chain; Object proposal; Site entropy rate; Stationary distribution; Markov chain; Site entropy rate; Co-segmentation; Co-saliency; Degree centrality; k-partite graph; Object proposal; Stationary distribution
• ISSN: 1047-3203; 1095-9076
• DOI: 10.1016/j.jvcir.2015.08.016

•Site entropy rate of graph nodes can be utilized to segment common object in images.•Degree centrality of graph nodes can also be used to obtain co-segmented regions.•Guided by co-saliency map, we combine the two methods to get image co-segmentations.•The proposed algorithm co-segments multiple instances of a common object in images.•The proposed algorithm co-segments images with multiple classes. In this paper, we propose a graph based algorithm that efficiently segments common objects from multiple images. We first generate a number of object proposals from each image. Then, an undirected graph is constructed based on proposal similarities and co-saliency maps. Two different methods are followed to extract the proposals containing common objects. They are: (1) degree centrality of nodes obtained after graph thresholding and (2) site entropy rate of nodes calculated on the stationary distribution of Markov chain constructed on the graph. Finally, we obtain the co-segmented image region by selecting the more salient of the object proposals obtained by the two methods, for each image. Multiple instances of the common object are also segmented efficiently. The proposed method has been compared with many existing co-segmentation methods on three standard co-segmentation datasets. Experimental results show its effectiveness in co-segmentation, with larger IoU values as compared to other co-segmentation methods.