现代电子技术2025，Vol.48Issue(13)：36-42,7.DOI:10.16652/j.issn.1004-373x.2025.13.005
基于深度学习和Retinex理论的图像增强方法
Image enhancement method based on deep learning and Retinex theory
王德旭 1汪志锋 1闫文强1
作者信息
- 1. 上海第二工业大学 智能制造与控制工程学院,上海 201200
- 折叠
摘要
Abstract
This study aims to improve image quality under low-light conditions,primarily addressing issues of reduced visibility and color distortion.By combining image processing techniques based on Retinex theory and advanced neural network algorithms,an innovative image enhancement framework is proposed.This framework consists of two parts,named an image decomposition network and an image enhancement network.The former is responsible for decomposing the original image into illumination component and reflectance component,while the later is responsible for optimizing parameters and performing γ correction by the natural image quality evaluator(NIQE),adjusting the brightness and contrast of the illumination component,and then re-fusing the illumination component with the reflectance component,so as to enhance overall image quality.Tests on standard datasets LOL and LOL-V2 show that the proposed method outperforms most existing image enhancement algorithms in terms of the peak signal-to-noise ratio(PSNR)and the structural similarity index measure(SSIM),which demonstrates its effectiveness and practicality in the field of low-light image enhancement.关键词
神经网络/Retinex理论/NIQE/γ校正/PSNR/SSIM/图像增强/图像分解Key words
neural network/Retinex theory/NIQE/γ correction/PSNR/SSIM/image enhancement/image decomposition分类
电子信息工程引用本文复制引用
王德旭,汪志锋,闫文强..基于深度学习和Retinex理论的图像增强方法[J].现代电子技术,2025,48(13):36-42,7.
