A New Depth Saliency Method Based on Selective Difference

oleh: Chen Qiu-Xia, Fu Li-Hua, Li Can-Can, Ding Hao-Gang

Format: Article
Diterbitkan: EDP Sciences 2017-01-01

Deskripsi

Stereopsis plays an important role in the human visual system. In this paper, we propose a novel regional contrast based saliency method, which works on depth images based on selective difference. In this work, we measure the saliency of a region based on regional contrast and take depth difference in horizontal direction into consideration. Besides, center prior is used here to refine the initial saliency map. The proposed method can better reflect the influence of depth information on saliency detection, and the results are more in line with human vision.