Progressive Lightweight Enhancement and Denoising Network for Efficient Low-Light Image Enhancement

Existing low-light image enhancement methods often either overlook noise artifacts during rapid illumination correction or rely on large-scale models with high computational demands. To address these issues, we introduce PLED (Progressive Lightweight Enhancement and Denoising Network), an efficient framework designed for low-light image enhancement. PLED employs a progressive Retinex-based decomposition to achieve staged illumination refinement and noise suppression. It features a Progressive Multi-Scale Residual Illumination Enhancement Block (PMSRIEB) for adaptive brightness adjustment and a Detail-Enhancement and Adaptive Noise Reduction Block (DEANRB) to preserve structural details while reducing noise. Through extensive experiments on paired and unpaired datasets, PLED demonstrates competitive visual quality with significantly fewer parameters and lower computational costs, making it ideal for real-time applications and resource-constrained environments.