Multi-mapping saliency based on global-local structural information
Feng Xu, Shengzhou Xiong, Yihua Tan · 2016
This paper proposes a saliency detection approach using multi-mapping based on global-local structural information. Firstly, a global-background-suppression(GBS) image and a Lab color space hybrid(CSH) image are obtained through different mapping rules. Secondly, both of them are segmented several times to produce multiple shape images. Thirdly, the above shape images are multiplied with GBS image to build intermediate saliency maps. At the end, all of the maps are fused to generate the final saliency map. The MSRA images are used to verify the proposed algorithm.