Cycle-Retinex: Unpaired Low-Light Image Enhancement Via Retinex-Inline CycleGAN

Kangle Wu,Jun Huang,Yong Ma,Fan,Jiayi Ma

DOI: https://doi.org/10.1109/tmm.2023.3278385

IF: 7.3

2024-01-01

IEEE Transactions on Multimedia

Abstract:Low-light image enhancement aims to recover normal-light images from the images captured under dim environments. Most existing methods could just improve the light appearance globally whereas failing to handle other degradation such as dense noise, color offset and extremely low-light. Moreover, unsupervised methods proposed in recent years lack reliable physical model as the basis, thus universality is greatly limited. To address these problems, we propose a novel low-light image enhancement method via Retinex-inline cycle-consistent generative adversarial network named Cycle-Retinex, whose training is totally dependent on unpaired datasets. Specifically, we organically combine Retinex theory with CycleGAN, by which we decouple low-light image enhancement task into two sub-tasks, i.e. illumination map enhancement and reflectance map restoration. Retinex theory helps CycleGAN simplify low-light image enhancement problem and CycleGAN provides synthetic paired images to guide the training of Retinex decomposition network. We further introduce a self-augmented method to address the color distortion and noise problem, thus making the network learn to enhance low-light images adaptively. Extensive experiments show that the proposed method can achieve promising results.