Enhancing and Learning Denoiser without Clean Reference

TL;DR

This paper introduces a novel deep image denoising approach that learns to denoise by modeling noise transfer, enabling effective removal of real-world noise without requiring clean reference images.

Contribution

The proposed method treats denoising as a noise transference task, allowing the network to learn denoising directly from corrupted samples without clean references.

Findings

Achieves promising results on real-world denoising benchmarks.

Outperforms traditional supervised and synthetic noise-based denoising methods.

Demonstrates strong generalization to real photographs.

Abstract

Recent studies on learning-based image denoising have achieved promising performance on various noise reduction tasks. Most of these deep denoisers are trained either under the supervision of clean references, or unsupervised on synthetic noise. The assumption with the synthetic noise leads to poor generalization when facing real photographs. To address this issue, we propose a novel deep image-denoising method by regarding the noise reduction task as a special case of the noise transference task. Learning noise transference enables the network to acquire the denoising ability by observing the corrupted samples. The results on real-world denoising benchmarks demonstrate that our proposed method achieves promising performance on removing realistic noises, making it a potential solution to practical noise reduction problems.