A New Generative Adversarial Network for Texture Preserving Image Denoising

Abstract

In this paper, a new generative adversarial networks (GAN) is proposed for image denoising. The proposed GAN has a new generator network to produce denoised images with noisy images as input, and the entire network is trained using a new loss to represent the distance between the data distribution of clean images and denoised images. Based on quantitative and qualitative evaluating criteria, we made comparisons between our method and other denoising methods which shows the superiority of our approach. Keywords—Image denoising, Generative adversarial network, Loss function, Texture Preserving.