Saliency Detection via Background Seeds by Object Proposals
Muwei Jian, Runxia Zhao, Junyu Dong, Kin‐Man Lam · 2018
In the recent research of saliency detection, many graph-based algorithms are applied, which use the border of an image as a background query. This frequently leads to undesired errors and retrieval outputs when the boundaries of the salient objects concerned touch, or connect with, the image's border. In this paper, we propose a novel bottom-up saliency-detection algorithm to tackle and overcome the above issue. First, we utilize object proposals to collect the background seeds reliably. Then, the Extended Random Walk algorithm is adopted to propagate the prior background labels to the rest of the pixels in an image. Finally, we refine the saliency map by taking both the textural and structural information into consideration simultaneously. Experiments on publicly available data sets show that our proposed approach achieves competitive results against the state-of-the-art methods.