Phase-based salient object detection

In this paper, we propose an efficient approach to computing saliency maps and determining salient areas in images. Biologically inspired feature maps are first constructed for a given image, and the phase information is then extracted from each feature map in the frequency domain. Finally, a saliency map is formed by using the phase information, followed by a smoothing operation. Experiment results show that our proposed algorithm is computationally efficient and can detect salient objects accurately.