摘要
车辆检测与跟踪是智能交通领域的重要研究课题之一。为了促进平安城市的建设,更好地辅助车辆驾驶,提出了一种基于类Haar特征和Adaboost分类器的实时车辆检测与跟踪算法。采集大量车辆正负样本图像,基于积分图提取图像的类Haar特征;利用Adaboost算法对类Haar特征进行选择及分类器训练;利用得到的分类器进行模式匹配,实现对车辆的检测。在相邻帧中进行车辆的特征匹配,完成车辆的跟踪。在车辆跟踪的基础上,通过场景标定,实现对车辆的测速和车流量的统计。在真实道路场景中的实验结果表明,所提方法能实时并有效地对车辆进行检测与跟踪,在一定程度上缓解了交通压力;能准确地进行车辆测速和车流量统计,可为超速和道路拥挤的判定提供相关依据,具有较好的应用前景。
Vehicle detection and tracking is one of the most important research topics in the field of intelligent transportation. A real-time algorithm of vehicle detection and tracking based on Haar-like features and the Adaboost classifier is proposed to promote the construc- tion of safe city and assist vehicle driving. A large number of positive and negative sample images of vehicle are collected. The Haar-like features of the images are extracted based on the integral map and the Adaboost algorithm is exploited to do Haar-like features selection and classifier training for matching the pattern with the obtained classifier to realize the vehicles detection. The characteristics of the vehi- cles in the adjacent frames are matched to complete vehicles tracking. By calibrating scene, the vehicle speed measurement and traffic sta- tistics have been achieved based on vehicles tracking. Experimental results in real road scene show that it can effectively conduct vehicle detection and tracking in real-time for alleviating the traffic pressure to some extent and can implement vehicle speed measurement and traffic statistics accurately, which has provided the relevant basis for speeding and road congestion with an excellent application prospect.
出处
《计算机技术与发展》
2017年第10期165-168,176,共5页
Computer Technology and Development
基金
安徽省自主创新专项资金计划项目(13Z02005)