Fusion of Deep Learning and Global-Local Features of the Image Salient Region Calculation

A novel algorithm of saliency detection combined with a multi-environment context deep learning framework is proposed to improve the efficiency of saliency detection. Firstly, color uniqueness and compactness of the foreground are used to highlight the foreground regions. Then a method of combining global and local information to fully consider the relationship between local properties and global context features is adopted. In order to refine the entire network essentially, a contextual reweighting recurrent feedback network module is proposed to transfer high-level semantic information from the top convolutional layer to shallower layers in a feedback manner, and filter the noise repeatedly to reduce the influence of background information. The algorithm of this paper is tested in the ECSSD database and DUT-OMRON database, respectively. And the experiment results show that the proposed algorithm is better than the other popular algorithms. © 2019, Beijing China Science Journal Publishing Co. Ltd. All right reserved.