交通视频的质心算法车辆跟踪计数系统  

Centroid algorithm of vehicle tracking and counting systemin traffic video

作　　者：雷柏超 马国亮 雷帮军[1,2] 邹耀斌[1,2] 任强[1,2] 

机构地区：[1]三峡大学水电工程智能视觉监测湖北省重点实验室 [2]三峡大学计算机与信息学院 [3]宜昌市公安局交警支队,湖北宜昌443002

出　　处：《信息通信》2018年第12期45-48,共4页Information & Communications

基　　金：湖北省创新群体项目(2015CFA025);赛尔网络下一代互联网技术创新项目(NGII20170704)

摘　　要：由于基于虚拟检测线的车辆计数误差较大,因此,研究一种实时性、鲁棒性好的车辆跟踪计数的算法成为研究的重点。文中提出了一种基于质心算法的多目标跟踪模型计数算法。首先提取视频图像中的前景运动图像,然后在检测位置设定虚拟线框,并设置区域的质心位置坐标,以巴氏系数判断当前目标和参考目标的匹配程度。通过实验分析,该算法可实现快速、有效的车辆跟踪计数。Due to the large vehicle counting error based on the virtual detection line. Therefore, researching a real-time, robust algorithm for vehicle tracking and counting is a top priority in this field. In this paper, a multi-target tracking and counting algorithm based on algorithm is proposed. First, the foreground moving image in the video image is extracted, then the virtual wireframe is set at the detection position, and the centroid position coordinates of the region are set, then the degree of matching between the current target and the reference target is determined by the Barthel coefficient. Through experimental analysis, the algorithm can achieve vehicle tracking and counting fast and effectively.

关 键 词：质心算法 车辆跟踪 车辆计数 虚拟线框 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]