Co-saliency Detection Based on Feature Fusion and Multiple Constraints

In order to solve the problem of the confusion of foreground and background information and the wrong indication of salient objects for image groups with complex environments,we propose a co-saliency detection model based on feature fusion and multiple constraints.Firstly,we calculate the inter-aliency values by fusing depth information optimized by objectness probability and color information.Then,we calculate the intrasaliency values by background priors with multiple constraints guided by depth probability.The saliency maps are also optimized.Finally,we fuse the saliency maps with region-wise proposal.The least squares method is used to obtain the final results.The experimental results on public datasets indicate that the consistent relationship among multiple images is used more effectively and background information is better suppressed.The salient objects are closer to the truth values.Overcoming the impact of complex environmental factors,the evaluation results of this algorithm have been significantly improved.