摘要
以道路实际行车图像为基础,通过Yolop识别算法对交通标线进行识别分割,对分割后的图像采集其HSV颜色模型指标(明度、色调、饱和度)图像特征参数,利用SPSS软件对通过人工方式采集到的标线逆反射亮度系数数据和图像特征参数进行了相关性分析。结果表明,标线图像的亮度值和饱和度值与标线的逆反射亮度系数值的斯皮尔曼相关系数分别为0.601和0.765,在0.01下相关性显著,并建立对应的多元回归方程,检测结果为后续基于机器视觉技术的车载式逆反射亮度系数检测仪标准制定和发展提供基础。
The correlation is key to realizing this technology.Based on the actual driving images of the road,this paper uses the Yolop recognition algorithm to identify and segment the traffic markings.The segmented images collect their HSV color model index(brightness,hue,saturation)image feature parameters,and use SPSS software to manually Correlation analysis was conducted on the marking line retroreflection coefficient data and image feature parameters collected by this method.The results show that the Spearman correlation coefficients of the brightness value and saturation value of the reticle image and the retroreflection brightness coefficient value of the reticle are 0.601 and 0.765 respectively.The correlation is significant at 0.01,and the corresponding multiple regression equation is established.The test results provide a basis for the subsequent formulation and development of vehicle-mounted retroreflective brightness coefficient detector standards based on machine vision technology.
作者
姚岩松
王益
罗强
尚永毅
何廷全
YAO Yansong;WANG Yi;LUO Qiang;SHANG Yongyi;HE Tingquan(Guangzhou University School of Civil Engineering and Transportation,Guangzhou 510006;Guangxi Xinfazhan Commnications Group Co.,Ltd,Nanning 530029,China)
出处
《交通工程》
2024年第9期9-16,共8页
Journal of Transportation Engineering
基金
广西壮族自治区交通运输厅科技示范工程(广西高速公路交通拥堵防治智能决策系统工程),广西重点研发计划(2021AB22078)
广州市基础研究计划基础与应用基础研究专题青年博士“启航”项目(SL2024A04J00335)。
关键词
标线
逆反射系数
图像特征
目标检测
相关性分析
marking
retroreflection coefficient
image features
target detection
correlation analysis
