Invertible Network for Unpaired Low-light Image Enhancement

TL;DR

This paper introduces an invertible network framework for unpaired low-light image enhancement, improving stability and detail preservation over traditional GAN-based methods by leveraging reversibility and progressive self-guided enhancement.

Contribution

It proposes a novel invertible network architecture with specialized loss functions for stable, detail-preserving low-light image enhancement without paired data.

Findings

Achieves superior performance compared to state-of-the-art methods.

Reduces artifacts and over-exposure issues.

Demonstrates stable training through reversibility loss.

Abstract

Existing unpaired low-light image enhancement approaches prefer to employ the two-way GAN framework, in which two CNN generators are deployed for enhancement and degradation separately. However, such data-driven models ignore the inherent characteristics of transformation between the low and normal light images, leading to unstable training and artifacts. Here, we propose to leverage the invertible network to enhance low-light image in forward process and degrade the normal-light one inversely with unpaired learning. The generated and real images are then fed into discriminators for adversarial learning. In addition to the adversarial loss, we design various loss functions to ensure the stability of training and preserve more image details. Particularly, a reversibility loss is introduced to alleviate the over-exposure problem. Moreover, we present a progressive self-guided enhancement process for low-light images and achieve favorable performance against the SOTAs.