On the Importance of Denoising when Learning to Compress Images

TL;DR

This paper demonstrates that explicitly training image codecs to denoise improves compression quality for noisy images, outperforming traditional methods and reducing computational complexity.

Contribution

It introduces a denoising-aware training approach for image codecs using diverse noise levels, achieving superior rate-distortion performance with fewer computations.

Findings

Single denoising-aware model outperforms separate denoising and compression pipelines.

Model trained on varied noise levels generalizes well to noisy and clean images.

Significant reduction in GMac operations compared to traditional methods.

Abstract

Image noise is ubiquitous in photography. However, image noise is not compressible nor desirable, thus attempting to convey the noise in compressed image bitstreams yields sub-par results in both rate and distortion. We propose to explicitly learn the image denoising task when training a codec. Therefore, we leverage the Natural Image Noise Dataset, which offers a wide variety of scenes captured with various ISO numbers, leading to different noise levels, including insignificant ones. Given this training set, we supervise the codec with noisy-clean image pairs, and show that a single model trained based on a mixture of images with variable noise levels appears to yield best-in-class results with both noisy and clean images, achieving better rate-distortion than a compression-only model or even than a pair of denoising-then-compression models with almost one order of magnitude fewer GMac operations.