Multi-scale salient object detection using graph ranking and global–local saliency refinement

• Published in: Signal processing. Image communication 2016-09, Vol.47, p.380-401
• Main Authors: Filali, Idir, Allili, Mohand Saïd, Benblidia, Nadjia
• Format: Article
• Language: English
• Subjects: Feature relevance; Multi-layer graphs; Random forests; Region and boundary information; Salient object detection (SOD)
• ISSN: 0923-5965; 1879-2677
• DOI: 10.1016/j.image.2016.07.007

We propose an algorithm for salient object detection (SOD) based on multi-scale graph ranking and iterative local–global object refinement. Starting from a set of multi-scale image decompositions using superpixels, we propose an objective function which is optimized on a multi-layer graph structure to diffuse saliency from image borders to salient objects. This step aims at roughly estimating the location and extent of salient objects in the image. We then enhance the object saliency through an iterative process employing random forests and local boundary refinement using color, texture and edge information. We also use a feature weighting scheme to ensure optimal object/background discrimination. Our algorithm yields very accurate saliency maps for SOD while maintaining a reasonable computational time. Experiments on several standard datasets have shown that our approach outperforms several recent methods dealing with SOD. •Salient object detection using multi-layered graph ranking.•Local-global saliency refinement using objectness-measure and random forests.•Region and boundary information for salient object detection.•Feature relevance for object/background discrimination.