Edge-Aware Saliency Detection via Novel Graph Model

Edge information takes an important role in distinguishing salient objects from background. In this paper, the screened edge information is utilized to roughly locate the salient object, which is combined with the color and texture to construct the feature space. Based on the feature space and fast background connection, a novel graph is put forward to effectively obtain the local and global cues and ease the blurry surrounds of the saliency maps while dealing with the intrinsic discontinuity and non-homogeneity within the salient object. Visual qualitative comparisons and comprehensive quantitatively evaluations on three benchmark datasets demonstrate that our method outperforms other state-of-the-art unsupervised methods and even some powerful supervised methods.