Fourier-based two-stage low-light image enhancement network via mutual learning

Jianming Zhang, Zhijian Feng, Jia Jiang, Yan Gui

Abstract: Highlights•Propose a Fourier-based low-light image enhancement network that enhances amplitude and phase sequentially in two stages.•Explore the correlations between amplitude and luminance, and phase and texture.•Design an amplitude enhancement module and a phase enhancement module.•For the two modules, design an amplitude enhancement block and a phase enhancement block based on FFT and IFFT.•For the two blocks, design a spatial unit and a frequency unit adopting a mutual learning strategy.