基于改进Retinex和分数阶微分的矿井红外图像增强算法  

Mine Infrared Image Enhancement Algorithm Based on Improved Retinex and Fractional Differentiation

作　　者：杨云帆 孙宗剑[2] YANG Yunfan;SUN Zongjian(Department of Basic Course Teaching,Hebei Vocational University of Industry and Technology,Shijiazhuang 050091,China;School of Sciences,Hebei University of Science and Technology,Shijiazhuang 050000,China)

机构地区：[1]河北工业职业技术大学基础课教学部,河北石家庄050091 [2]河北科技大学理学院,河北石家庄050000

出　　处：《金属矿山》2024年第10期196-201,共6页Metal Mine

基　　金：河北省自然科学基金项目(编号:A2022208007)。

摘　　要：矿井红外图像增强是提高井下工作环境安全性和可视化监控能力的关键技术之一。然而,由于井下光照复杂、红外图像质量较差等因素,现有算法在增强效果和细节保留方面存在不足。为此,提出了基于改进Retinex和分数阶微分的矿井红外图像增强算法。首先针对现有Retinex算法存在的局部对比度不足问题,引入颜色恢复模块,在进行反射传递函数估计时利用彩色空间信息增强图像的局部对比度。该模块通过计算多个彩色通道的统计特征,并结合颜色转换矩阵,实现了更准确的反射传递函数估计,从而提高了图像整体对比度。其次通过分析图像的局部灰度特征,根据分数阶微分算子的响应特性自动调整指数参数,以实现对不同纹理细节的合理增强。采用多组真实矿井红外图像对所提算法性能进行了试验验证,结果表明:该算法在图像增强效果和细节保留方面相对于Retinex等算法具有一定的优势,为提升矿井复杂光照条件下的红外图像对比度提供了支持。Mine infrared image enhancement is one of the key technologies to improve the safety of underground working environment and visual monitoring ability.However,due to complex illumination in mine environment and poor infrared image quality,the existing algorithms have shortcomings in enhancement effect and detail retention.Therefore,a mine infrared image enhancement algorithm based on improved Retinex and fractional differentiation is proposed.Firstly,aiming at the problem of insufficient local contrast existing in Retinex algorithm,a color recovery module is introduced to enhance the local contrast of images by using color spatial information in the estimation of reflection transfer function.By calculating the statistical characteristics of multiple color channels and combining with the color conversion matrix,the module achieves a more accurate reflection transfer function estimation,thereby improving the overall contrast of the image.Secondly,by analyzing the local gray features of the image,the index parameters are automatically adjusted according to the response characteristics of fractional differential operators to achieve reasonable enhancement of different texture details.The performance of the proposed algorithm is verified by experiments with multiple sets of real mine infrared images.The results show that the proposed algorithm has certain advantages over Retinex and other algorithms in terms of enhancement effect and detail retention,and provide a reference for improving the contrast of infrared images in complex mine lighting conditions.

关 键 词：矿井红外图像 改进Retinex 分数阶微分 图像增强 

分 类 号：TD672[矿业工程—矿山机电] TP391[自动化与计算机技术—计算机应用技术]