基于局部图块目标匹配的交通车流跟踪与统计  

Traffic Car Flow Tracking and Statistics Based on Local Block-Graphs Targets Matching

作　　者：刘剑[1,2] 龚志恒[1] 林璐瑶[3] 吴成东[2] 高恩阳[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,沈阳110168 [2]东北大学信息科学与工程学院,沈阳110004 [3]沈阳建筑大学管理学院,沈阳110168

出　　处：《控制工程》2014年第3期436-440,445,共6页Control Engineering of China

基　　金：国家自然科学基金项目(61272253);国家住建部科技项目(2010-K9-22)

摘　　要：针对传统目标跟踪算法在实现过程中的局限,提出一种基于局部图块目标匹配(local block-graphs targets matching,-LBTM)的跟踪算法,将其应用于车流跟踪,并验证其有效性。首先,采集视频帧画面中含有目标的多幅图像,并对其进行局部图块分割,得到目标图块;其次,对图块进行目标匹配,通过匹配完成对目标的检测过程,并计算得到最优的目标集合;最后,针对目标集合进行全部帧画面的最小偏差的预测,实现目标的跟踪过程。选取某路口的交通监控视频进行对比验证实验,实验结果表明:所提出的算法可以有效地跟踪车辆,比传统算法有更好的目标检测率和跟踪准确率,并能有效地完成车流统计。The traditional targets tracking algorithms have been limitations in the realization process. A novel tracking method based on local block-graphs targets matching is proposed. In order to prove the effectiveness of the method, it is applied to car flow tracking. Firstly, some images are captured by the video frames containing the targets, and the targets block-graphs can be obtained. Secondly, the block-graphs can be used to achieve targets matching, and the detection process can be achieved by target matching. Thereafter, targets set can be obtained. Finally, the targets set can be used to achieve the minimum deviation forecasting of all frames, and the tar- get tracking can be achieved. A junction video is selected as the experimental data source. The results show that the proposed method not only can track cars effectively, but also has better detection rate and tracking accuracy. The car flow statistics can be completed ef- fectively.

关 键 词：目标跟踪 匹配 车流 局部图块 最小偏差 预测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]