基于驾驶人视认性的交通标线反光性检测方法

Detection Method of Traffic Marking Reflectance Based on Driver’s Visual Recognition

交通标线是保障交通安全的生命线,标线夜间的视认性差是致使道路交通事故频发的主要原因之一.本文结合标线材料反光性原理以及标线质量检测要求,基于驾驶人视认特性,提出一种利用交通标线亮度值评价夜间标线可视性的方法,该方法运用Yolov4算法以及Deeplabv3等算法对标线图像进行目标检测、图像分割以及亮度值计算.为实验提出方法的效用性,开展了标线逆反射系数和亮度值检测实验,收集到24组标线逆反射系数和亮度值的样本数据,并利用斯皮尔曼相关系数法对逆反射系数和亮度值进行相关性分析.研究结果表明标线的亮度值与逆反射系数具有显著相关性,亮度值可代替逆反射系数表征标线的反光特征.

Traffic marking is the lifeline of traffic safety. The poor visual recognition of marking at night is one of the main causes of frequent road traffic accidents. In this paper, combined with the reflective principle of line marking materials and the requirements of line marking quality detection, based on drivers’ visual recognition characteristics, a method is proposed to use Yolov4 algorithm and Deeplabv3 algorithm to perform target detection, image segmentation and brightness value calculation on the collected line marking images, and use the brightness value of traffic line to evaluate the visibility of night line marking. In order to verify the effectiveness of the proposed method, 24 groups of sample data of the retroreflectance coefficient and luminance value were collected, and the correlation between the retroreflectance coefficient and luminance value was analyzed by spearman correlation coefficient method. The results show that there is a significant correlation between the brightness value of the line and the retroreflection coefficient, and the brightness value can represent the reflective characteristics of the line instead of the retroreflection coefficient.