A Novel Saliency Model for Stereoscopic Images

In this paper, we propose a novel saliency model for stereoscopic images. To improve depth information for stereo saliency analysis, this model exploits depth information from three aspects: 1) we extract the low-level features based on the color-depth contrast features in a local and global search range (local-global contrast); 2) to extract the topological structural from a depth map, a surrounding map based on a Boolean map is obtained as a weight value to enhance the local-global contrast features; and 3) based on the saliency probability distribution in depth information, we employ stereo center prior enhancement to compute the final saliency. Experimental results on two recent eye-tracking databases show that our proposed method outperforms the state-of-the-art saliency models.