Saliency Detection on Graph Manifold Ranking via Multi-scale Segmentation

Saliency detection is an essential task in the field of computer vision. Its role is to identify significant or prominent regions from an image, weigh the visual information, and then better understand the visual scene. The existing saliency detection methods based on graph manifold ranking usually achieve good results in salient object detection tasks. However, they do not consider the influence of different segmentation scales on the saliency detection results. To solve this problem, based on the original saliency detection method based on graph manifold ranking, this paper performs multi-scale segmentation. It constructs a multi-scale fusion saliency detection algorithm. The experimental results show that the proposed method achieves better than seven classical saliency detection algorithms on four data sets.