WSEAS Transactions on Computers
Print ISSN: 1109-2750, E-ISSN: 2224-2872
Volume 13, 2014
Data-Driven Saliency Region Detection Based on Undirected Graph Ranking
Authors: , ,
Abstract: Highlighting saliency region is still a challenging problem in computer vision. In this paper, we present a data-driven salient region detection method based on undirected graph ranking. It consists of two steps: we first compute priori saliency map on super-pixel image by combining region contrast and center prior information, and then extract saliency map by optimized ranking function based on a new affinity matrix. It is simple and efficient. Furthermore, salient objects can be successfully highlighted with precise details and high consistency. We evaluate the proposed method with three image datasets. The experimental results show that the proposed approach has a good performance in terms with the PR curve, the ROC curve.