摘要
设计了一套基于视频的机动车跨道违章监测系统,系统的硬件平台由工控机,摄像机组成.首先由平均算法得到良好的动态背景图像,并由Hough变换法提取车道线;在车辆检测过程中提出了背景差法与边缘提取法相结合,车道线辅助识别的两次目标轮廓提取算法,从而获得完整准确的运动目标轮廓;在违章判别中提出了简易准确的跨道违章判别算法;最后使用均值漂移算法对运动车辆进行跟踪,并结合卡尔曼滤波器预测车辆在下一帧的位置.通过实际道路测试,该系统具有一定的实时性、准确性和智能性,可应用于智能交通监控领域.
A kind of crossing road monitoring system based on traffic video was presented.The hardware system was composed of industrial computers and cameras.The background image of road was obtained by using average method,and the lane lines were extracted by Hough transform.In order to obtain complete and accurate outline of the moving vehicle,a method of obtaining vehicle profile combined with background subtraction and edge detection was proposed.A simple method for illegal crossing road detection was also put forward.Mean-shift algorithm combined with Kalman filter was used to track and predict the location of vehicle.Experimental results show that this system satisfies the need of real-time computing and accuracy,which can be used in intelligent traffic supervising.
出处
《应用基础与工程科学学报》
EI
CSCD
2011年第3期517-525,共9页
Journal of Basic Science and Engineering
基金
国家自然科学基金(60772080)
天津市自然科学基金资助项目(08JCYBJC13800)
天津市科技支撑计划重点项目基金资助项目(10ZCKFSF01100)
天津市公安交通局科研基金资助项目(2005[16])
关键词
智能交通
跨道违章
车道识别
HOUGH变换
运动跟踪
intelligent traffic
illegal crossing road
lane line detection
Hough transformation
motion tracking
