RSEND: Retinex-based Squeeze and Excitation Network with Dark Region Detection for Efficient Low Light Image Enhancement

2024-06-14

arXiv.org

AI Summary

Existing CNN-based low-light image enhancement methods suffer from insufficient accuracy on complex datasets (e.g., LOL-v2), high computational overhead, and cumbersome training. To address these issues, this paper proposes a single-stage Retinex decomposition framework that jointly models illumination and reflectance components. It introduces a novel dark-region-aware mechanism to guide illumination enhancement and integrates Squeeze-and-Excitation attention throughout the entire pipeline for efficient detail recovery. Furthermore, it combines element-wise fusion with a multi-scale denoising module to balance decomposition accuracy and inference efficiency. Experiments demonstrate that our method achieves PSNR gains of 0.44–4.2 dB over state-of-the-art CNN approaches on LOL-v2, and outperforms Transformer-based baselines on the LOL-v2-real subset. The proposed framework delivers superior performance, high computational efficiency, and streamlined training—offering a practical and effective solution for low-light enhancement.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Images captured under low-light scenarios often suffer from low quality. Previous CNN-based deep learning methods often involve using Retinex theory. Nevertheless, most of them cannot perform well in more complicated datasets like LOL-v2 while consuming too much computational resources. Besides, some of these methods require sophisticated training at different stages, making the procedure even more time-consuming and tedious. In this paper, we propose a more accurate, concise, and one-stage Retinex theory based framework, RSEND. RSEND first divides the low-light image into the illumination map and reflectance map, then captures the important details in the illumination map and performs light enhancement. After this step, it refines the enhanced gray-scale image and does element-wise matrix multiplication with the reflectance map. By denoising the output it has from the previous step, it obtains the final result. In all the steps, RSEND utilizes Squeeze and Excitation network to better capture the details. Comprehensive quantitative and qualitative experiments show that our Efficient Retinex model significantly outperforms other CNN-based models, achieving a PSNR improvement ranging from 0.44 dB to 4.2 dB in different datasets and even outperforms transformer-based models in the LOL-v2-real dataset.

Problem

• Enhancing low-light images efficiently and accurately
• Reducing computational resources in Retinex-based methods
• Improving performance in complex datasets like LOL-v2

Innovation

• Retinex-based one-stage framework for enhancement
• Squeeze and Excitation network for detail capture
• Dark region detection and denoising for efficiency