Abstract:We propose a Retinex‐based LLIE network by integrating attention mechanism and U‐Net‐like architecture to alleviate the challenges. Specifically, the proposed model is composed of three modules namely decomposition module (DECM), reflectance recovery module (REFM), and illumination enhancement module (ILEM), aiming to decompose the low light image based on Retinex theory and enhance the decomposed reflectance and illumination maps by taking advantages of attention mechanism and U‐Net‐like architecture. Low‐light images are captured in environments with minimal lighting, such as nighttime or underwater conditions. These images often suffer from issues like low brightness, poor contrast, lack of detail, and overall darkness, significantly impairing human visual perception and subsequent high‐level visual tasks. Enhancing low‐light images holds great practical significance. Among the various existing methods for Low‐Light Image Enhancement (LLIE), those based on the Retinex theory have gained significant attention. However, despite considerable efforts in prior research, the challenge of Retinex decomposition remains unresolved. In this study, an LLIE network based on the Retinex theory is proposed, which addresses these challenges by integrating attention mechanisms and a U‐Net‐like architecture. The proposed model comprises three modules: the Decomposition module (DECM), the Reflectance Recovery module (REFM), and the Illumination Enhancement module (ILEM). Its objective is to decompose low‐light images based on the Retinex theory and enhance the decomposed reflectance and illumination maps using attention mechanisms and a U‐Net‐like architecture. We conducted extensive experiments on several widely used public datasets. The qualitative results demonstrate that the approach produces enhanced images with superior visual quality compared to the existing methods on all test datasets, especially for some extremely dark images. Furthermore, the quantitative evaluation results based on metrics PSNR, SSIM, LPIPS, BRISQUE, and MUSIQ show the proposed model achieves superior performance, with PSNR and BRISQUE significantly outperforming the baseline approaches, where (PSNR, mean BRISQUE) values of the proposed method and the second best results are (17.14, 17.72) and (16.44, 19.65). Additionally, further experimental results such as ablation studies indicate the effectiveness of the proposed model.

Multi-path parallel enhancement of low-light images based on multiscale spatially aware Retinex decomposition

Research on Low-Light Image Enhancement Algorithm Based on Filtering and Data-Driven Composite

Multi-Modular Network-Based Retinex Fusion Approach for Low-Light Image Enhancement

Retinex decomposition based low‐light image enhancement by integrating Swin transformer and U‐Net‐like architecture

Color and Luminance Separated Enhancement for Low-Light Images with Brightness Guidance

MARN: Multi-Scale Attention Retinex Network for Low-Light Image Enhancement

Toward Robust and Efficient Low-Light Image Enhancement:Progressive Attentive Retinex Architecture Search

A Retinex-based network for image enhancement in low-light environments

A non-regularization self-supervised Retinex approach to low-light image enhancement with parameterized illumination estimation

A depth iterative illumination estimation network for low-light image enhancement based on retinex theory

A Low-Light Image Enhancement Algorithm Incorporating Cross-Mixed Attention and Receptive Field Expansion Mechanism

Sparse Gradient Regularized Deep Retinex Network for Robust Low-Light Image Enhancement

Multi-Frequency Field Perception and Sparse Progressive Network for low-light image enhancement

Low-Light Enhancement Using a Plug-and-Play Retinex Model With Shrinkage Mapping for Illumination Estimation

End-to-End Retinex-Based Illumination Attention Low-Light Enhancement Network for Autonomous Driving at Night

Latent Disentanglement for Low Light Image Enhancement

Retinex theory-based nonlinear luminance enhancement and denoising for low-light endoscopic images

SCRNet: a Retinex Structure-based Low-light Enhancement Model Guided by Spatial Consistency

Dark2Light: multi-stage progressive learning model for low-light image enhancement

Image Restoration via Low-Illumination to Normal-Illumination Networks Based on Retinex Theory

Low-Light Image Enhancement With Semi-Decoupled Decomposition