交通流视频检测中车辆的检测与跟踪  

Vehicles detecting and tracking in traffic flow video

作　　者：胡峰俊[1] 沈国珍[1] 杨海波[1] 

机构地区：[1]浙江树人大学现代教育技术中心,浙江杭州310005

出　　处：《计算机工程与设计》2008年第6期1582-1584,共3页Computer Engineering and Design

基　　金：浙江省科技计划基金项目(2006C31011)

摘　　要：交通流视频检测技术是实现城市交通现代化管理的核心技术之一,也是智能交通系统的重要组成部分。使用改进混合高斯分布模型来表征图像帧中每一个像素点的亮度特征,使用绝对值差分提取出运动车辆的轮廓,采集目标的轮廓特征等数学特征,采用相似性算子跟踪多运动目标,采用面积阈值排除非车因素的干扰。最后在城市交通流实况检测中验证算法。The traffic flow video detecting technology is one of the key technologies of realizing modernized management of the city traf-fic,which is also the important component of the intellectual traffic system,using Gaussian mixture model to signify every pixel＇s lumi-nance characteristic of the picture,drawing the contour of the moving vehicles by absolute value difference and gathering the mathematic characteristic of the object,and using comparability operator for tracking multi-targets,eliminating other disturbance factor by using area threshold,confirming algorithm in the live city transportation detecting.

关 键 词：交通流 混合高斯分布模型 车辆跟踪 多目标 相似性算子 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]