Finding saliency object via an integration approach

Saliency detection is a hot topic in the community of computer image and vision. In this paper, we present a new saliency detection method. Given an input image, our method first uses Harris corner detection technique to approximately locate the salient region, and then assign the saliency scores to each pixel, getting the center-prior based map. In addition, we employ Bayesian formula to further optimize it, obtaining the center-Bayesian map. On the other hand, we use the image boundary to generate boundary-based map. Finally, we merge them into a saliency map as our final saliency map. A large number of experimental results demonstrate that the proposed algorithm is superior to most existing algorithms.