1.长安大学电子与控制工程学院,陕西 西安 710064
2.汉滨区农村公路养护中心,陕西 安康 725000
温立民(1976年生),男;研究方向:图像处理与机器视觉;E-mail:lmwen@chd.edu.cn
网络出版日期:2025-01-23,
收稿日期:2024-11-26,
录用日期:2024-12-19
移动端阅览
温立民, 杨睿, 聂磊, 等. 基于NGDR和Logistic模型的高速公路图像雾浓度检测算法[J/OL]. 中山大学学报(自然科学版)(中英文), 2025,1-10.
WEN LIMIN, YANG RUI, NIE LEI, et al. Detection algorithm for highway image fog concentration based on NGDR and Logistic model. [J/OL]. Acta scientiarum naturalium universitatis sunyatseni, 2025, 1-10.
温立民, 杨睿, 聂磊, 等. 基于NGDR和Logistic模型的高速公路图像雾浓度检测算法[J/OL]. 中山大学学报(自然科学版)(中英文), 2025,1-10. DOI: 10.13471/j.cnki.acta.snus.ZR20240336.
WEN LIMIN, YANG RUI, NIE LEI, et al. Detection algorithm for highway image fog concentration based on NGDR and Logistic model. [J/OL]. Acta scientiarum naturalium universitatis sunyatseni, 2025, 1-10. DOI: 10.13471/j.cnki.acta.snus.ZR20240336.
提出了基于Logistic函数拟合S型散点图的雾浓度评定算法。首先,提取LIVE标准图集归一化灰度差-比散点图先验;基于散点曲线与视场雾浓度的一一对应关系,引入Logistic函数并推导出适合回归分析的模型。其次,采用迭代搜索法确定纵向高斯分布的最佳回代样本点,以提高检测精度。最后,建立参数估计(
<math id="M1"><mover accent="true"><mi>β</mi><mo>^</mo></mover><mo>
</mo><mover accent="true"><mi>γ</mi><mo>^</mo></mover></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=72790431&type=
3.80999994
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=72790419&type=
5.16466665
)的查找表,采用计算相关系数和遍历搜索查找的方法实现雾浓度等级评定。同场景不同浓度图像样本1的测试表明,真实图像的PM2.5与查找表PM2.5的相关系数达0.99,检测误差小于2.9%;近似场景不同浓度高速公路图像样本2的测试表明,真实图像PM2.5与查找表PM2.5值的相关系数达0.98,检测误差小于1.8;执行效率对比测试表明,本文算法对于300 Kb样本图像的处理时间为19.8 s,低于同精度数据驱动的深度视觉算法;检测精度对比测试表明,本文算法优于其它典型算法。
A fog concentration evaluation algorithm based on logistic function fitting S-type scatter plot was pro
posed. Firstly, a scatter-plot prior of normalized gray difference-ratio(NGDR) from the standard LIVE image set was extracted,and Logistic functions was introduced to derive a regression analysis model based on the one-to-one correspondence between the scatter curve and the fog concentration. Secondly, the iterative search method was used to determine the optimal sample points of longitudinal Gaussian distribution to improve the detection accuracy. Finally, a lookup table for parameter estimation(
<math id="M2"><mover accent="true"><mi>β</mi><mo>^</mo></mover><mo>
</mo><mover accent="true"><mi>γ</mi><mo>^</mo></mover></math>
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=72790432&type=
4.40266657
https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=72790420&type=
7.95866632
)was established, and both calculating the correlation coefficient and traversal search were used to evaluate concentration grade. The test of image samples with different concentrations in the same scene 1 shows that the correlation coefficient between PM2.5 in the real image and PM2.5 in the lookup table was 0.99, and the detection error was less than 2.9%. The test results of highway image sample 2 with different concentrations in the approximate scene show that the correlation coefficient is 0.98, and the detection error is less than 1.8. The comparative test of execution efficiency shows that the processing time of the proposed algorithm for 300 Kb sample images is 19.8 s, which is lower than that of the data-driven depth vision algorithm with the same precision. The comparative test of detection accuracy shows that the proposed algorithm is better than other typical algorithms.
高速公路图像雾浓度检测NGDRLogistic模型回归分析查找表
highwayimagefog concentration detectionNGDRLogistic modelregression analysislookup table
唐晓庆,范赐恩,刘鑫,2015. 基于边缘保持滤波的单幅图像快速去雾[J].西安交通大学学报,49(3):143-150.
王雪冬,张超彪,王翠,等,2022. 基于Logistic回归与随机森林的和龙市地质灾害易发性评价[J]. 吉林大学学报(地球科学版),52(6):1957-1970.
温立民,巨永锋,闫茂德,2017. 基于自然统计特征分布的交通图像雾浓度检测[J].电子学报,45(8):1888-1895.
张琪东,迟静,陈玉妍,等,2024. 基于雾浓度分类与暗-亮通道先验的多分支去雾网络[J]. 计算机研究与发展,61(3):762-779.
周佳, 2018. 蒙特卡罗模拟在计量经济学中的应用——以样本容量对回归分析的影响为例[J].中国商论,22:161-163.
中国气象局, 2020. 气象与交通专家解析如何科学防御减轻雾的影响[EB/OL]. https://www.cma.gov.cn/2011xzt/2012zhuant/20120102/2012010204/201110/t20111028_3093508.htmlhttps://www.cma.gov.cn/2011xzt/2012zhuant/20120102/2012010204/201110/t20111028_3093508.html.
GALLEN R,CORD A,HAUTIÉRE N,et al,2015. Nighttime visibility analysis and estimation method in the presence of dense fog[J].IEEE Trans Intell Transp Syst,16(1):310-320.
GUY H, BROOKS I M , TURNER D D, et al,2023. Observations of fog-aerosol interactions over central Greenland[J]. JGR: Atmospheres,128(13):1-24.
JAD B, ERIC G, 2022. Factor and factor loading augmented estimators for panel regression with possibly non-strong factors[J]. J Bus Econ Stat,41(1):270-281.
JIANG Y, ZHANG N,LI A X, et al, 2020. Effects of weather on highway traffic capacity in China: Characteristics and causes of roadblocks due to fog events [J].Pure Appl Geophys,177(10):5027-5040.
MITTAL A, MOORTHY A K, BOVIK A C, 2012. No-reference image quality assessment in the spatial domain [J]. IEEE Trans Image Process, 21: 4695.
PAL T, HALDER M , BARUA S, et al, 2023. A deep learning model to detect foggy images for vision enhancement[J]. Imag Sci J,71(6):484-498.
RAJEVENCELTHA J, GAIDHANE V H, 2024. A no-reference image quality assessment model based on neighborhood component analysis and Gaussian process[J]. J Vis Commun Image Represent,98:104041.
SHENG D, DENG J, XIANG J W, 2021. Automatic smoke detection based on Slic-Dbscan enhanced convolutional neural network[J] IEEE Access,9(19): 63933-63942.
WICHITAKSORN N, KANG Y, ZHANG F, 2023. Random feature selection using random subspace logistic regression [J]. Expert Syst Appl,217(5):119535.
WU W, HUANG D, YAO Y, et al, 2024. Feature rectification and enhancement for no-reference image quality assessment[J]. J Visual Commun Image Represent,98(9):104030.
XIAO P, ZHANG Z, LUO X, et al,2023. Highway visibility estimation in foggy weather via multi-scale fusion network[J]. Sensors,23(24):9739.
YANG W C,ZHAO Y T, LI Q, et al, 2023. Multi visual feature fusion based fog visibility estimation for expressway surveillance using deep learning network[J]. Expert Syst Appl, 234(9): 121151.
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构