Night-to-Day Translation via Illumination Degradation Disentanglement

TL;DR

This paper introduces N2D3, a novel approach for night-to-day translation that disentangles illumination degradation to improve visual quality and semantic preservation in nighttime images, outperforming previous methods.

Contribution

The paper proposes a degradation disentanglement and contrastive learning framework guided by physical priors for effective night-to-day translation under unpaired conditions.

Findings

Significant visual quality improvements on public datasets

Enhanced preservation of semantic content during translation

Potential benefits for downstream vision tasks

Abstract

Night-to-Day translation (Night2Day) aims to achieve day-like vision for nighttime scenes. However, processing night images with complex degradations remains a significant challenge under unpaired conditions. Previous methods that uniformly mitigate these degradations have proven inadequate in simultaneously restoring daytime domain information and preserving underlying semantics. In this paper, we propose \textbf{N2D3} (\textbf{N}ight-to-\textbf{D}ay via \textbf{D}egradation \textbf{D}isentanglement) to identify different degradation patterns in nighttime images. Specifically, our method comprises a degradation disentanglement module and a degradation-aware contrastive learning module. Firstly, we extract physical priors from a photometric model based on Kubelka-Munk theory. Then, guided by these physical priors, we design a disentanglement module to discriminate among different illumination degradation regions. Finally, we introduce the degradation-aware contrastive learning strategy to preserve semantic consistency across distinct degradation regions. Our method is evaluated on two public datasets, demonstrating a significant improvement in visual quality and considerable potential for benefiting downstream tasks.