Image Denoising of Real Photographs with Generative Adversarial Network for Data Augmentation

Yuan Fu,Cien Fan,Lian Zou,Ye Yang,Yifeng Liu
DOI: https://doi.org/10.1117/1.jei.28.5.053017
IF: 0.829
2019-01-01
Journal of Electronic Imaging
Abstract:Although image denoising methods based on a convolutional neural network (CNN) have achieved the state-of-the-art in Gaussian noise reduction, their performances are still very limited on real photographs even beyond the reach of traditional methods, such as block-matching and 3-D filtering and weighted nuclear norm minimization. We use a denoising benchmark to train a generative adversarial network for noise modeling and produce more data indistinguishable from the original ones, which can be regarded as a data augmentation scheme. Then, we utilize this extended dataset, including real images and synthetic images to train a CNN-based denoiser named DRNet for real photograph denoising. In the design of DRNet, we introduce a noise estimation module to improve the robustness of the single learning framework for handling unknown noise levels and a pair of reversible downsampling and upsampling operators to enlarge the receptive field. Experiments on real-world noisy images are conducted to evaluate our algorithm, and the results show that DRNet is effective for real photographs in comparison with other methods, especially in balancing the noise removal and the structure preservation. (C) 2019 SPIE and IS&T
What problem does this paper attempt to address?