[1]韩格格,李香芳,蒋伊泽.高速公路雾图像能见度的监测与识别方法研究[J].气象研究与应用,2024,45(02):50-56.[doi:10.19849/j.cnki.CN45-1356/P.2024.2.08]
　HAN Gege,LI Xiangfang,JIANG Yize.Research on monitoring and recognition technology of highway fog image visibility[J].Journal of Meteorological Research and Application,2024,45(02):50-56.[doi:10.19849/j.cnki.CN45-1356/P.2024.2.08]
点击复制

高速公路雾图像能见度的监测与识别方法研究()
分享到：

气象研究与应用[ISSN:1673-8411/CN:45-1356/P]

卷:
第45卷
期数:
2024年02期
页码:
50-56
栏目:
研究论文
出版日期:
2024-06-15

文章信息/Info

Title:
Research on monitoring and recognition technology of highway fog image visibility
作者:
韩格格1; 2;  李香芳1; 3;  蒋伊泽1; 4
1. 中国气象局 旱区特色农业气象灾害监测预警与风险管理重点实验室, 银川 750002;
2. 宁夏回族自治区气象信息中心, 银川 750002;
3. 宁夏回族自治区气象服务中心, 银川 750002;
4. 盐池县气象局, 宁夏 盐池 751599
Author(s):
HAN Gege1; 2;  LI Xiangfang1; 3;  JIANG Yize1; 4
1. Key Laboratory of Agro-Meteorological Disaster Monitoring, Early Warning and Risk Management in Arid Regions, China Meteorological Administration, Yinchuan 750002, China;
2. Ningxia Meteorological Information Center, Yinchuan 750002, China;
3. Ningxia Meteorological Service Center, Yinchuan 750002, China;
4. Yanchi Meteorological Bureau, Ningxia Yanchi 751599, China
关键词:
雾; 图像识别; 高速公路; 监测模型; 能见度
Keywords:
fog; image recognition; highway; monitoring model; visibility
分类号:
P427.2
DOI:
10.19849/j.cnki.CN45-1356/P.2024.2.08
摘要:
基于高速公路雾图像能见度识别的方法，对采集到的高速公路雾天气图像进行预处理，选出与能见度相关性较高的图像特征、监测因子、兴趣窗格，采用机器学习方法，探索图像特征与雾天气能见度之间的关系，构建雾天气能见度二元线性回归模型，并对监测模型输出结果进行验证。结果表明：（1）饱和度的均值、色度的方差与能见度相关性较高。饱和度和色度是能见度监测的关键性因素，而不是颜色。（2）通过划分不同的能见度等级，基于随机森林算法对图像能见度进行判定，该模型分类准确率达到 90%，对图像的能见度区间判定具有较强的分类能力；（3）构建不同能见度等级的二元线性回归模型，对能见度预测准确率较高，预测值均在正确范围内，其中70%的预测值很接近真实值。
Abstract:
In recent years，automatic visibility stations have been set up along expressways to monitor fog，which has played an important role in ensuring traffic safety. However，the automatic visibility stations are generally far away from each other，and cannot monitor local fogs and mass fogs in a small range. Therefore，this paper proposes a visibility recognition method based on fog images on expressways. The collected highway fog weather images are preprocessed，and the image features，monitoring factors and interest panes with high correlation with visibility are selected. Machine learning method is also adopted to explore the relationship between image features and visibility in fog weather，a binary linear regression model for visibility in fog weather is constructed，and the output results of the monitoring model is verified. The results show that: (1) through the experiment，it is proved that the mean value of saturation and the variance of chroma have a high correlation with visibility，while the three color features of red，green and blue have a low correlation with visibility，indicating that saturation and chroma are the key factors for visibility monitoring， rather than color. (2) By dividing different visibility levels，the image visibility is determined based on the random forest algorithm，and the classification accuracy of the model reaches 90%，which has a strong classification ability for the determination of the visibility interval of the images. (3) The binary linear regression model with different visibility levels is constructed，and the verification results of the verification data set show that the visibility prediction accuracy of the model is high，and the predicted values are all within the correct range，of which 70% of the predicted values were very close to the true values.

参考文献/References:

[1] 刘瑞莲,柴德美.气象条件对交通安全的影响分析[J].现代农业科技,2017(6):229,231.
[2] 冯蕾,邓毅萍,李蔼恂,等.雾天高速公路交通事故的特征及与能见度的关系[J].科技导报,2020,38(11):160-168.
[3] KWON T M. Final report on signal and image processing for road condition classification[R]. 2002.
[4] NARASIMHAN S G,NAYAR S K. Vision and atmosphere[J]. International Journal of Computer Vision,2001,48(3):233-254.
[5] NARASIMHAN S G,NAYAR S K. Shedding light on the weather[C]//The 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,June 18-20,2003,Madison,Wisconsin. Piscataway:IEEE Press, 2003:665-672.
[6] GALLEN R,HAUTIERE N,DUMONT E. Static estimation of meteorological visibility distance in night fog with imagery[J]. IEICE Transactions on Information and Systems,2010,93(7):1780-1787.
[7] GALLEN R,CORD A,HAUTIERE N. Nighttime visibility analysis and estimation method in the presence of dense fog[J]. IEEE Transactions on Intelligent Transportation Systems,2015,16(1):310-320.
[8] CHINCHOLKAR S,RAJAPANDY M. Fog image classification and visibility detection using CNN[C]//Intelligent Computing, Information and Control Systems:ICICCS 2019. Springer International Publishing,2020:249-257.
[9] JIANG Y,SUN C,ZHAO Y,et al. Fog density estimation and image defogging based on surrogate modeling for optical depth[J]. IEEE Transactions on Image Processing, 2017,26(7):3397-3409.
[10] 常荣.基于卷积神经网络的高速公路团雾图像分类的研究[D].合肥:安徽建筑大学,2021.
[11] 李从初,励臣儒,朱佳敏,等.基于迁移学习和Xception网络的海雾能见度等级估测研究[J].浙江气象,2023, 44(1):23-28.
[12] 徐德保.基于机器学习的高速公路团雾图像分类[D].合肥:安徽建筑大学,2021.
[13] 戴庞达,张玉钧,周毅,等.基于双光源图像特征的夜间能见度观测方法研究[J].光谱学与光谱分析,2014,34(1):1-5.
[14] 夏创文.高速公路网运行监测若干关键技术研究[D].广州:华南理工大学,2013.
[15] LIUCX,LUXB,JIS P,et al. A fog level detection method based on image HSV color histogram[C]//Progress in Informatics and Computing (PIC),May 16-18,2014, Shanghai,China. Piscataway:IEEE Press,2014:373-377.
[16] LIU Y,LU X B. Fog detection and classification using gray histograms[C]//2011 International Conference on Embedded System and Microprocessors,2010:175-178.
[17] 花毓幸.基于图像兴趣窗格测算大气能见度的方法研究[D].南京:南京信息工程大学,2016.
[18] 叶栋水.基于视频的雾天能见度监测方法研究[D].南京:南京信息工程大学,2018.
[19] 赵博文,张力夫,潘在峰,等.基于OpenCV的图像滤波方法比较[J].信息与电脑(理论版),2020,32(15):78-80.
[20] 饶敏.特殊虚线交通标线设计及图像识别技术研究[D].大连:大连理工大学,2023.
[21] 古晶.RGB到Lab颜色空间转换[J].广东印刷,2009(5):11-14.
[22] 杨康叶,邬春学.基于RGB模型颜色相似性的彩色图像分割[J].计算机系统应用,2013,22(3):128-131,160.
[23] 包新月,俞磊.计算机数字图像处理常用颜色空间及转换[J].电子技术与软件工程,2021(7):122-123.
[24] 户尊兰.基于颜色特征分析的色偏检测与校正算法研究[D].武汉:武汉大学,2017.

相似文献/References:

[1]黄子航,黄干淇.广西西江流域大雾的气候特征及大雾类型分类[J].气象研究与应用,2017,38(02):22.
　Huang Zihang,Huang Ganqi.Climatic characteristics and classification of heavy fog in Xijiang river basin of Guangxi[J].Journal of Meteorological Research and Application,2017,38(02):22.
[2]杨守懋,王娟怀,史成,等.四会雾的气候特征及天气成因分析[J].气象研究与应用,2016,37(02):47.
　Yang Shou-mao,Wang Juan-huai,Shi Cheng,et al.Characteristics of fog and its causes in Sihui[J].Journal of Meteorological Research and Application,2016,37(02):47.
[3]陈兴鹃,黄淑娥,樊建勇,等.基于FY-3A/VIRR卫星资料的江西省大雾遥感监测[J].气象研究与应用,2018,39(01):91.
　Chen Xingjuan,Huang Shue,Fan Jianyong,et al.Remote sensing monitoring of fog in Jiangxi based on FY-3A/VIRR satellite data[J].Journal of Meteorological Research and Application,2018,39(02):91.

备注/Memo

备注/Memo:
收稿日期:2023-9-30。
基金项目:中国气象局旱区特色农业气象灾害监测预警与风险管理重点实验室开放研究项目（CAMF-202206）
作者简介:韩格格（1991-），工程师，主要从事气象信息技术处理工作。E-mail：1499912520@qq.com。
更新日期/Last Update: 1900-01-01