A Novel Visual Saliency Detection Method Using Motion Segmentation

In this paper, we propose a novel visual saliency detection method using motion segmentation. We group the corner point trajectories using a two stage clustering algorithm. The most stable trajectories are pre-clustered using mean shift in the first stage. Then, we propose an unsupervised clustering method to cluster the trajectories and detect the number of motions automatic. At last, the motion saliency map is generated with the segmented spare feature points. Experimental results show that our proposed method is capable of achieving both good accurate and the stable performance.