Generative Adversarial Network-based Image Dehazing of Electric Power Facilities
Power facility monitoring can effectively avoid the occurrence of major power accidents,but under the influence of natural phenomena such as heavy fog in bad weather,the monitoring images of power facilities are blurred and difficult to work appropriately.Aiming at this problem,this paper proposes a dehazing method for power facility images based on generative adversarial network (GAN).This method performs deep dehazing by decomposing the surveillance color image into three RGB channels.Moreover,a novel joint loss function is designed to enhance the learning ability of generative ad-versarial networks.Extensive experimental verification on power image dehazing dataset confirms better image dehazing performance of the proposed method compared with the comparative methods,as the PSNR and SD are improved by 1 .1 and 2 .97 ,respectively.