Image Denoising Using an Improved Generative Adversarial Network with Wasserstein Distance

The image denoising discriminant model has received extensive attention in recent years due to its good denoising performance. In order to solve the problems of denoising of traditional generative adversarial networks, which are difficult to train and easy to collapse, traditional denoising methods will damage the visibility of important structural details after the image is denoised, this paper proposes an improved generative adversarial network algorithm. The novel algorithm obtains more image features by adding multi-level convolution of the generative network, and adds multiple residual blocks and global residuals to extract and learn the features of the input noisy image to avoid the loss of features. The network uses the weighted sum of feature loss and perceptual loss, which can effectively retain the detailed information of the image while removing noise. In the experiments, we selected PSNR and SSIM as the indicators, and the experimental results show that the novel algorithm can effectively remove image noise and improve the visual perception of the image.