基于区域光照约束和Retinex的煤矿井下低光照图像增强  

Enhancement of low light images in coal mines based on regional lighting constraints and Retinex

作　　者：彭庆国 李志鹏 PENG Qingguo;LI Zhipeng(Chinacoal Intelligent Technology Co.,Ltd.,Huainan,Anhui 232000,China)

机构地区：[1]中煤智能科技有限公司,安徽淮南232000

出　　处：《计算机应用文摘》2025年第7期115-117,共3页

摘　　要：针对煤矿井下环境光线昏暗、光照不均等问题,文章提出了一种基于区域光照约束和Retinex理论的无监督低光图像增强方法。通过构建并优化基于Retinex理论的欠曝光模型,设计了特征增强模块以生成光照映射和增强图像,同时引入噪声抑制模块以降低图像噪声。通过光照约束的计算,有效避免了过度曝光和颜色不饱和等问题。实验结果表明,该方法在自有数据集上表现优异;与现有最先进方法相比,NIQE指标提升了5%,BRISQUE指标提升了3%。此外,该方法具有较低的计算复杂度和更快的训练速度,在矿井低光图像增强任务中展现出显著优势。This article proposes an unsupervised low light image enhancement method based on regional lighting constraints and Retinex theory to address the problems of dim and uneven lighting in the underground environment of coal mines.By constructing and optimizing an underexposure model based on Retinex theory,a feature enhancement module was designed to generate illumination maps and enhance images,while introducing a noise suppression module to reduce image noise.Through the calculation of lighting constraints,problems such as overexposure and color saturation have been effectively avoided.The experimental results show that this method performs well on its own dataset.Compared with the existing state-of-the-art methods,the NIQE index has increased by 5% and the BRISQUE index has increased by 3%.In addition,this method has lower computational complexity and faster training speed,demonstrating significant advantages in low light image enhancement tasks in mines.

关 键 词：低光图像增强 煤矿井下 RETINEX理论 深度学习 光照约束 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]