An auto-selection method for saliency detection and its application to segment the object

Based on the analysis to the three methods of IG, GBVS and PQFT, and on the comparison of their advantages and disadvantages, an auto-selection method for saliency detection is proposed by calculating the consistency between two methods. The salient object in the image is separated through combining GrabCut algorithm with the saliency map generated. The results of experiments demonstrate that our method is not only has a good performance on saliency detection but also can effectively separate the salient object.