地下隧道单向双车道双洞交通事故准确检测
|
摘要
针对当前检测方法通过地下隧道监控视频检测交通事件,难以准确获取运动目标轨迹参数,存在耗时长、交通事故检测判别不准确等缺点,提出基于运动目标轨迹跟踪的地下隧道单向双车道双洞交通事故准确检测方法。利用背景差法获取地下隧道监控视频运动目标前景。在使用基于边缘检测和HSV颜色空间相融合算法对存在遮挡的运动目标去除阴影的基础上,采用卡尔曼滤波算法对运动目标进行跟踪,得到待检测事件序列的运动轨迹,将复杂的车辆轨迹划分为前行、反行、停滞、斜行四类轨迹元素,依据轨迹元素对待测交通事件的行驶行为进行数学建模,通过该模型准确检测交通事故的发生。实验结果表明,所提方法能够快速准确获取运动目标轨迹参数,在一定程度上提高了检测效率,且地下隧道单向双车道双洞交通事故的检测精度较高。
In the current detection method, it is difficult to accurately obtain the trajectory parameter of moving target. Meanwhile, the detection and discrimination for traffic accident is not accurate. This article puts forward a method for accurately detecting traffic accident in one-way double-lane of double-hole underground tunnel based on trajectory tracking of moving target. At first, the background subtraction method was used to obtain the foreground of monitor video moving target in underground tunnel. Secondly, the fusion algorithm based on edge detection and HSV color space was used to remove the shadow of moving object with shelter, and then Kalman filter algorithm was used to track the moving object, so as to obtain the moving track of event sequence which needed to be detected. Moreover, the complex vehicle trajectory was divided into four kinds of trajectory elements: ahead, backward, stop, and left or right. According to the trajectory elements, the mathematical modeling was performed on the driving behavior of traffic event to be detected. Thus, the traffic accident was detected accurately by this model. Simulation results show that the proposed method can quickly and accurately obtain the trajectory parameters of moving target, which improves the detection efficiency. Meanwhile, the detection accuracy of traffic accident in one-way double-lane of double-hole underground tunnel is higher.
In the current detection method, it is difficult to accurately obtain the trajectory parameter of moving target. Meanwhile, the detection and discrimination for traffic accident is not accurate. This article puts forward a method for accurately detecting traffic accident in one-way double-lane of double-hole underground tunnel based on trajectory tracking of moving target. At first, the background subtraction method was used to obtain the foreground of monitor video moving target in underground tunnel. Secondly, the fusion algorithm based on edge detection and HSV color space was used to remove the shadow of moving object with shelter, and then Kalman filter algorithm was used to track the moving object, so as to obtain the moving track of event sequence which needed to be detected. Moreover, the complex vehicle trajectory was divided into four kinds of trajectory elements: ahead, backward, stop, and left or right. According to the trajectory elements, the mathematical modeling was performed on the driving behavior of traffic event to be detected. Thus, the traffic accident was detected accurately by this model. Simulation results show that the proposed method can quickly and accurately obtain the trajectory parameters of moving target, which improves the detection efficiency. Meanwhile, the detection accuracy of traffic accident in one-way double-lane of double-hole underground tunnel is higher.
引文
[1] 邵长桥,李敏.基于流量与占有率模型的交通事故检测器布设研究[J].北京工业大学学报,2016,42(9):1392-1397.
[2] 朱文兴,刘晓亮,赵成龙.基于跟踪算法和模糊推理的交通事件检测方法[J].数据采集与处理,2016,31(6):1115-1126.
[3] 刘翔,等.基于现场信息的交通事故视频验伪算法[J].计算机工程,2017,43(2):38-42.
[4] 陈卫卫,王卫星.车道偏离预警线图像边缘优化检测仿真[J].计算机仿真,2017,34(6):394-397.
[5] 严亚丹,等.高速公路交通事故黑点鉴别及影响因素分析[J].公路工程,2016,41(6):59-62.
[6] 朱燕,等.基于聚类的出租车异常轨迹检测[J].计算机工程,2017,43(2):16-20.
[7] 魏赟,杨梦灵.基于小波和模糊逻辑理论的交通事件检测[J].计算机工程与设计,2017,38(7):1924-1928.
[8] 韩功,王彦,何楚.使用物体交互模型的车辆违停事件检测[J].电视技术,2017,41(4-5):228-233.
[9] 张密科,胡选儒.基于图像纹理分析的动态车辆识别方法研究[J].公路交通科技,2017,34(10):122-127.
[10] 邹铁方,等.基于Pc-Crash的多车碰撞事故再现仿真分步方法[J].中国安全科学学报,2016,26(5):59-63.
[2] 朱文兴,刘晓亮,赵成龙.基于跟踪算法和模糊推理的交通事件检测方法[J].数据采集与处理,2016,31(6):1115-1126.
[3] 刘翔,等.基于现场信息的交通事故视频验伪算法[J].计算机工程,2017,43(2):38-42.
[4] 陈卫卫,王卫星.车道偏离预警线图像边缘优化检测仿真[J].计算机仿真,2017,34(6):394-397.
[5] 严亚丹,等.高速公路交通事故黑点鉴别及影响因素分析[J].公路工程,2016,41(6):59-62.
[6] 朱燕,等.基于聚类的出租车异常轨迹检测[J].计算机工程,2017,43(2):16-20.
[7] 魏赟,杨梦灵.基于小波和模糊逻辑理论的交通事件检测[J].计算机工程与设计,2017,38(7):1924-1928.
[8] 韩功,王彦,何楚.使用物体交互模型的车辆违停事件检测[J].电视技术,2017,41(4-5):228-233.
[9] 张密科,胡选儒.基于图像纹理分析的动态车辆识别方法研究[J].公路交通科技,2017,34(10):122-127.
[10] 邹铁方,等.基于Pc-Crash的多车碰撞事故再现仿真分步方法[J].中国安全科学学报,2016,26(5):59-63.