Abstract:In recent years, most denoising methods based on deep convolutional neural networks heavily rely on massive noisy–clean image pairs. Collecting massive noisy–clean image pairs is expensive and not practical in real scenes. Currently, few-shot learning has been applied to many areas to cope with the absence of data. The few-shot learning, however, in image denoising severely suffers from domain gap problems, including dataset domain gap and feature domain gap, especially for the real noisy images. Therefore, this article proposes a multilevel noise contrastive network (MNC-Net) performing few-shot image denoising. MNC-Net consists of two training stages: 1) using contrastive learning to self-supervise the training of multilevel noise contrastive learner (MNCL) on the pure synthetic noisy images with multiple Gaussian noise levels to ease the acute dataset domain gap and 2) features generated by the MNCL on limited data are fused to the second stage and alleviate the feature domain gap using our proposed denoising network. Specifically, the MNCL consists of a contrastive feature extractor (CFE) and a contrastive feature projector (CFP). MNCL learns the rich and complex content-invariant degradations and general multiple-level noise representations. The denoising network in the second stage is composed of guided feature encoder (GFE) and adaptive denoising decoder (ADD). The GFE uses contrast features from CFE to guide the produced representations on the specific input noisy images. Then, such output features are fed into the ADD to adaptively denoise the noisy images on the corresponding noise distribution. To the best of our knowledge, this work is the first attempt to jointly use the few-shot learning and contrastive learning in the deep denoising field. Extensive experiments on CBSD68, Kodak24, Set12, SIDD, and DND show that our method achieves promising denoising performances in the absence of data.

Comparing and Combining MLPs and CNNs for Image Denoising

A Neural Network Based Low-Light Image Denoising Method

Multinoise-type Blind Denoising Using a Single Uniform Deep Convolutional Neural Network.

Image Denoising Based On A Cnn Model

Image denoising with multi-layer perceptrons, part 1: comparison with existing algorithms and with bounds

On Combining CNN with Non-Local Self-Similarity Based Image Denoising Methods.

Convolutional Neural Networks For Image Denoising And Restoration

D3CNNs: Dual Denoiser Driven Convolutional Neural Networks for Mixed Noise Removal in Remotely Sensed Images

Denoising Convolutional Neural Network Inspired Via Multi-Layer Convolutional Sparse Coding

A Comparison of Image Denoising Methods

Block-Matching Convolutional Neural Network for Image Denoising

Considering Image Information and Self-similarity: A Compositional Denoising Network

A Comprehensive Comparison of Multi-Dimensional Image Denoising Methods

MMPDNet: Multi-Stage & Multi-Attention Progressive Image Denoising

Combining Multi-Level Loss For Image Denoising

A Parallel and Serial Denoising Network

Comparing Noisy Patches for Image Denoising: A Double Noise Similarity Model

Image Denoising Method Based on a Deep Convolution Neural Network

3D Filtering by Block Matching and Convolutional Neural Network for Image Denoising

Multilevel Noise Contrastive Network for Few-Shot Image Denoising

Deep learning on image denoising: An overview