Salient object detection in RGB-D image based on saliency fusion and propagation

Automatic detection of salient objects in images attracts much research attention for its usage in numerous multimedia applications. In this paper, we propose a saliency fusion and propagation strategy based salient object detection method for RGB-D images, in which multiple cues are fused to provide high precision detection result and saliency propagation is utilized to improve the completeness of salient objects. To each RGB-D image, we firstly generate the saliency maps based on color cue, location cue and depth cue independently. Then, we fuse the saliency maps and obtain a high precision saliency map. Finally, we propagate saliency to obtain more complete salient objects. We evaluate the proposed method on two public data sets for salient object detection, NJU400 and RGBD Benchmark. The experimental results demonstrate saliency fusion and propagation are effective in salient object detection and our method outperforms the state-of-the-art methods.