Superpixel fats for fast foreground extraction
Kang Liu, Xuelong Li, Yongsheng Dong · 2015
Fast foreground extraction is an important and challenging problem. Although GrabCut can perform well in foreground extraction, the average accuracy is not satisfactory, and more importantly, its computational cost is large. In this paper, we propose a fast interactive foreground extraction based on superpixel GrabCut and matting. Specifically, we use a superpixel method to process images in order to improve the efficiency of the GrabCut. To obtain a refined mask, we employ the Shared matting method whose trimaps are implemented by our constructed fast and adaptive trimaps (FATs). Experimental results demonstrate the proposed method is more competitive and effective in the visual sense, E-Time and MSE criterions.