交叉口入口路段车流量及车辆换道率检测研究

发布时间：2019-06-08 15:41

【摘要】：针对交叉口入口路段交通视频中存在车辆遮挡严重、频繁走停以及频繁换道插队等特点,提出了一种获取交叉口停车线后大视野范围内车流量以及车辆换道率的检测方法。首先采用多级虚拟特征线生成多级时空图,对车辆进行快速检测和分割,获得车辆候选区域;然后对车辆候选区域内局部特征点进行初步分组和跟踪,并根据相同组内特征点运动趋势相似性来修正分组,解决车辆遮挡问题,用于检测车辆换道率;最后将多级时空图与特征点跟踪相结合,进行相互反馈,实现对车辆准确分割和鲁棒跟踪,避免车辆行驶中走停的影响。实验结果表明,通过该方法能实时准确地获取大视野范围内交叉口入口路段车流量和车辆换道率的交通参数。
[Abstract]:In view of the characteristics of serious traffic occlusion, frequent walking and stopping and frequent lane switching in the traffic video of intersection entrance section, a detection method for traffic flow and vehicle change rate in the wide field of vision after the intersection stop line is proposed. Firstly, the multi-level virtual feature line is used to generate the multi-level space-time map, and the vehicle is detected and segmented quickly to obtain the vehicle candidate region. Then the local feature points in the vehicle candidate region are preliminarily grouped and followed, and the grouping is modified according to the similarity of the movement trend of the feature points in the same group to solve the vehicle occlusion problem, which can be used to detect the vehicle lane change rate. Finally, the multistage space-time map and feature point tracking are combined to feedback each other to realize the accurate segmentation and robust tracking of vehicles, so as to avoid the influence of walking and parking on vehicles. The experimental results show that the traffic parameters of traffic flow and vehicle exchange rate can be obtained in real time and accurately by this method.
【作者单位】：北京工业职业技术学院机电工程系;清华大学自动化系;
【基金】：国家自然科学基金资助项目(61273006) 中国博士后科学基金资助项目(2015M571051) 北京工业技术学院科研课题资助项目(bgzyky201407)
【分类号】：U491;TP391.41

【共引文献】