GAN2GAN: Generative Noise Learning for Blind Denoising with Single Noisy Images

Abstract

We tackle a challenging blind image denoising problem, in which only single noisy images are available for training a denoiser and no information about noise is known, except for it being zero-mean, additive, and independent of the clean image. In such a setting, which often occurs in practice, it is not possible to train a denoiser with the standard discriminative training or with the recently developed Noise2Noise (N2N) training; the former requires the underlying clean image for the given noisy image, and the latter requires two independently realized noisy image pair for a clean image. To that end, we propose GAN2GAN (Generated-Artificial-Noise to Generated-Artificial-Noise) method that can first learn to generate synthetic noisy image pairs that simulate independent realizations of the noise in the given images, then carry out the N2N training of a denoiser with those synthetically generated noisy image pairs. Our method consists of three parts: extracting smooth noisy patches to learn the noise distribution in the given images, training a generative model to synthesize the noisy image pairs, and devising an iterative N2N training of a denoiser. In results, we show the denoiser trained with our GAN2GAN, solely based on single noisy images, achieves an impressive denoising performance, almost approaching the performance of the standard discriminatively-trained or N2N-trained models that have more information than ours, and significantly outperforming the recent baselines for the same setting.Preprint. Under review.

NoiseBreaker: Gradual Image Denoising Guided by Noise Analysis

“…These percentages refer to the amount of pixels, in the image, corrupted by the given noise. This type of spatial mixture noise has been used in the experiments of Generated-Artificial-Noise to Generated-Artificial-Noise (G2G) [10]. An example of sequential mixture noise is used to test the recent Noise2Self method [11].…”

“…Compared methods Our solution is evaluated in comparison with BM3D [4], DnCNN [5] and N2V [27]. Although related to our proposal, G2G [10] is evaluated on other mixture noises. Furthermore, no code is publicly available yet.…”

“…h is then composed of the set {η(ρ), ρ ∈ x}. This type of noise is called a spatial mixture [10,11]. The mixture noise can also be considered as the result of n primary noises applied with distributions ηi, i ∈ {0..n − 1} to each pixel ρ of the image x.…”

“…These denoisers are trained following the strategy mentioned in the original paper[5] except for the noisy data, which is made of the primary noise (according to Table1).Compared methods Our solution is evaluated in comparison with BM3D[4], DnCNN[5] and N2V[27]. Although related to our proposal, G2G[10] is evaluated on other mixture noises. Furthermore, no code is publicly available yet.…”

Remote Sensing Image Denoising Based on Deep and Shallow Feature Fusion and Attention Mechanism

“…In addition, many methods can generate remote sensing image denoising datasets. In [16,17], after noise extraction is performed on the uniform area in the noise image, the generative adversarial network (GAN) is trained to estimate the noise distribution over the input noisy images and generate new noise samples. Then, the paired training set can be generated from the noise map obtained in the previous step, and in turn, train a neural network to denoise.…”