基于视频检测信息的不连续路面标线优化方法研究  

Optimization Method of Discontinuous Road Marking Based on Video Detection

作　　者：庄俊杰 吴鼎新[1] 张盼 陈渤文 吴涛 ZHUANG Jun-jie;WU Ding-xin;ZHANG Pan;CHEN Bo-wen;WU Tao(Faculty of Transportation Engineering,Huaiyin Institute of Technology,Huai'an Jiangsu 223001,China)

机构地区：[1]淮阴工学院交通工程学院,江苏淮安223001

出　　处：《淮阴工学院学报》2022年第5期8-15,共8页Journal of Huaiyin Institute of Technology

基　　金：国家自然科学基金(51308246,51408253);江苏省产学研项目(BY2021007)。

摘　　要：城市道路中的不良设计,导致了路段标线间断现象广泛存在,交叉口进口处渠化不规范导致进出车道数量不匹配,并严重制约道路通行能力。因为路面标线中断,导致车辆交织混行现象频发,严重影响城市通行效率。首先将YOLOv5检测程序进行算法优化,使其满足交通流检测需求。优化后视频检测程序能动态可视化显示交通视频中的车辆数目、车辆变道行为、车辆行驶轨迹及区间车速等信息。其次根据可视化结果得出交通流信息特征,分析不连续路面标线处车辆驾驶习惯,指出不连续标线处的标线方法。接着,提出针对不连续路面标线段的标线绘制方法,通过对不连续路面标线段进行特征性优化,进一步提高道路通行能力。在VISSIM仿真下可知,其优化方案对比当前状态有显著提升,通行量提升21%。The design defects in urban roads have led to the widespread existence of discontinuous phenomenon of road sections.The channelization at the entrance of the intersection is not standardized,which leads to the mismatch of the entrance and exit,and seriously restricts the capacity of the road.Because of the interruption of road marking,the frequent phenomenon of vehicles interweaving and crossing seriously affects the efficiency of urban traffic.Firstly,the algorithm of machine vision is optimized to meet the traffic flow detection requirements.The optimized video detection program can dynamically and visually display the number of vehicles,lane change behavior of vehicles,vehicle track and interval speed in the traffic video.Secondly,the results show the traffic flow information characteristics,then the driving habits of vehicles at the discontinuous road marking are analyzed,and the marking method at the discontinuous road marking is pointed out.Then,a marking method for the discontinuous road marking is proposed,which can improve the road capacity through the characteristic optimization of drawing marking in discontinuous road.According to the micro simulation results,the optimization scheme is significantly improved compared with the current state,and the traffic volume is increased by 21%.

关 键 词：城市交通 YOLOv5视频检测 不连续路面标线 VISSIM仿真 车辆指引方法 

分 类 号：U491.523[交通运输工程—交通运输规划与管理]