弱光线下道路交通标志图像增强算法.pdfVIP

计算机时代2025年第1期

DOI:10.16644/j.cnki.cn33-1094/tp.2025.01.002

弱光线下道路交通标志图像增强算法

张书涵，王婷，郭钊汝

(新疆农业大学计算机与信息工程学院，新疆乌鲁木齐830052)

摘要：针对弱光线下道路交通标志特征不明显、细节颜色丢失，目标检测算法识别准确度低的问题，本文基于HSV空

间提出针对V分量的改进MSR-CLAHE算法、针对S分量的UM-GF算法。对V分量利用双边滤波代替高斯滤波改进

MSR算法，进行局部对比度增强，并结合CLAHE算法以限制对比度的过度放大。对S分量采用非锐化掩模图像进行引

导滤波，保持边缘锐化，同时也有效保留了色彩信息。并在CCTSDB数据集中选取弱光线图片进行了测试实验，实验的

主观和客观评价指标表明本文提出的算法能够有效提高图像对比度，增强细节信息，为后续交通标志识别提供支持。

关键词：图像增强；MSR算法；CLAHE算法；非锐化掩模；引导滤波

中图分类号：TP391.4文献标识码：A

Roadtrafficsignimageenhancementalgorithmunderlowlightconditions

ZhangShuhan,WangTing,GuoZhaoru

(CollegeofComputerandInformationEngineering,XinjiangAgriculturalUniversity,Urumqi,Xinjiang830052,China)

Abstract:Inordertosolvetheproblemthatthecharacteristicsofroadtrafficsignsarenotobviousandthedetailsarelostunder

weaklight,andtherecognitionaccuracyoftargetdetectionalgorithmislow,thispaperproposesanimprovedMSR-CLAHE

algorithmfortheV-componentandaUM-GFalgorithmfortheS-componentbasedontheHSVspace.FortheVcomponent,the

MSRalgorithmisimprovedbyusingbilateralfilteringinsteadofGaussianfiltering,andthelocalcontrastisenhancedby

combiningCLAHEalgorithmtolimittheexcessiveamplificationofcontrast.ForScomponent,theunsharpenedmaskimageis

usedtoguidethefiltering,whichkeepstheedgesharpening,andalsoefectivelyretainsthecolorinformation.Low-lightimages

wereselectedintheCCTSDBdatasetfortestingexperiments.Theexperimentsshowthatthealgorithmadoptedinthispapercan

effectivelyimproveimagecontrastandenhancedetailsinbothsubjectiveandobjectiveevaluationtests,andprovidesupportfor

subsequenttrafficsignrecognition.

Keywords:Imageenhancement,MS