Abstract:We investigate the task of learning blind image denoising networks from an unpaired set of clean and noisy images. Such problem setting generally is practical and valuable considering that it is feasible to collect unpaired noisy and clean images in most real-world applications. And we further assume that the noise can be signal dependent but is spatially uncorrelated. In order to facilitate unpaired learning of denoising network, this paper presents a two-stage scheme by incorporating self-supervised learning and knowledge distillation. For self-supervised learning, we suggest a dilated blind-spot network (D-BSN) to learn denoising solely from real noisy images. Due to the spatial independence of noise, we adopt a network by stacking \(1\times 1\) convolution layers to estimate the noise level map for each image. Both the D-BSN and image-specific noise model (\(\text {CNN}_{\text {est}}\)) can be jointly trained via maximizing the constrained log-likelihood. Given the output of D-BSN and estimated noise level map, improved denoising performance can be further obtained based on the Bayes’ rule. As for knowledge distillation, we first apply the learned noise models to clean images to synthesize a paired set of training images, and use the real noisy images and the corresponding denoising results in the first stage to form another paired set. Then, the ultimate denoising model can be distilled by training an existing denoising network using these two paired sets. Experiments show that our unpaired learning method performs favorably on both synthetic noisy images and real-world noisy photographs in terms of quantitative and qualitative evaluation. Code is available at https://github.com/XHWXD/DBSN.

Learning Deep Representations Using Convolutional Auto-Encoders with Symmetric Skip Connections

Image Denoising Using Very Deep Fully Convolutional Encoder-Decoder Networks with Symmetric Skip Connections.

Image Restoration Using Convolutional Auto-encoders with Symmetric Skip Connections.

Image Restoration Using Very Deep Convolutional Encoder-Decoder Networks with Symmetric Skip Connections.

Denoising Auto-Encoders Toward Robust Unsupervised Feature Representation.

Dilated Residual Encode-Decode Networks for Image Denoising

Deep Convolutional Architecture for Natural Image Denoising

Stacked Convolutional Denoising Auto-Encoders for Feature Representation

Stacked Convolutional Sparse Auto-Encoders for Representation Learning

Image Denoising and Inpainting with Deep Neural Networks.

A Deep Encoder-Decoder Networks for Joint Deblurring and Super-Resolution

Discriminative Representation Learning with Supervised Auto-encoder

Learning a Deep Convolutional Network for Subband Image Denoising

Enhanced CNN for Image Denoising

Image Super-Resolution Using Convolutional Neural Network with Symmetric Skip Connections

A Hybrid CNN for Image Denoising

A Multiscale Image Denoising Algorithm Based On Dilated Residual Convolution Network

Learning visual co-occurrence with auto-encoder for image super-resolution

Noisy-as-Clean: Learning Self-Supervised Denoising From Corrupted Image

Fast and accurate image denoising via a deep convolutional-pairs network

Unpaired Learning of Deep Image Denoising