Salient Region Detection Based on SLIC and Graph-based Segmentation
Xiaofei Sun, Wenwen Pan, Xia Wang, Xuhong Li, Guan Wang · 2016
At present, the most recent saliency detection algorithms are still not satisfactory.A saliency detection algorithm based on SLIC and graph-based segmentation is proposed.Firstly, graph-based segmentation is used to obtain larger image partitions, and the partitions with good contours are acquired.Then the relatively detailed partitions are obtained using SLIC image segmentation.Sparse color histogram is applied to these two methods.By using the color and spatial distance information, the saliency of each partition is calculated.Contours of salient objects from saliency maps generated using Graph-Based Segmentation are acquired and the gray scale values of the two kinds of saliency maps in the area surrounded by contours are merged.Experimental results show that compared with other detection methods, the proposed algorithm can effectively detect the salient object in the image.