Graph Based Image Saliency Detection

In this paper, we propose a new saliency detection method based on graph. The method firstly uses SLIC to segment the image into a set of superpixels which are not overlapped, and regards these superpixels as nodes of the graph, then constructs a partly connected graph with these nodes. Secondly, the saliency values of these nodes are computed via absorbing Markov chain and manifold ranking, and a saliency map are obtained corresponding to each image. Finally, it uses a contrast stretching function to correct the above saliency map for improving the quality of image. Experimental results demonstrate that the proposed method can more easily distinguish salient objects from background, and performs better than some methods in terms of robustness and performance. Keywords-Image Saliency; Absorbing Markov chain;