华南理工大学学报(自然科学版) ›› 2006, Vol. 34 ›› Issue (9): 70-75.

• 交通与运输工程 • 上一篇    下一篇

基于视频角点信息特征的交通流参数测算方法

秦钟 徐建闽 刘利频 史胜利   

  1. 华南理工大学 交通学院,广东 广州 510640
  • 收稿日期:2005-09-02 出版日期:2006-09-25 发布日期:2006-09-25
  • 通信作者: 秦钟(1972-).男,博士生,主要从事智能交通和视频检测研究 E-mail:qzscut@yahoo.corn.cn
  • 作者简介:秦钟(1972-).男,博士生,主要从事智能交通和视频检测研究
  • 基金资助:

    国家自然科学基金资助项目(50578064 );广东省科技攻关项目(2005A10101001)

Parameter Detection for Trafi c Flow Based on Characteristics of Video Corner Information

Qin Zhong  Xu Jian-min  Liu Li-pin  Shi Sheng-li   

  1. School ofTrafic and Communications,South China Univ.ofTech.,Guangzhnu 510640,Guangdong,China
  • Received:2005-09-02 Online:2006-09-25 Published:2006-09-25
  • Contact: 秦钟(1972-).男,博士生,主要从事智能交通和视频检测研究 E-mail:qzscut@yahoo.corn.cn
  • About author:秦钟(1972-).男,博士生,主要从事智能交通和视频检测研究
  • Supported by:

    国家自然科学基金资助项目(50578064 );广东省科技攻关项目(2005A10101001)

摘要: 提出了一种基于视频图像中角点特征的交通参数检测方法.首先设置检测区域,并将检测区域分为初始检测区和跟踪检测区,在初始检测区内对运动车辆的存在性进行检测.为了加快计算速度,利用虚拟线状态检测的方法判断车辆是否驶入,如果有运动车辆出现,则对运动车辆的角点进行定位,并在初始检测区内实现状态参数的初始化.在跟踪检测区.基于刚体运动的一致性假设,依据角点的分布特征对车辆进行分割,实现对运动车辆的跟踪.在车辆完全驶离跟踪区域时,统计交通流的流量、速度以及车型等信息.

关键词: 角点检测, 特征匹配, 运动跟踪, 交通参数, 参数测量

Abstract:

This paper proposes a detection method of traffic parameters based on the characteristics of video comer information,in which a detection region is divided into an initial one and a tracing one.In the initiative detection region,the existence of moving vehicles is tested,and a suppositional line state detection is carried out to deter-mine whether the vehicle is moving in,thus speeding up the calculation.W hen a vehicle is moving in, its com ers are firstly located and its state parameters are initialized.In the tracing region,the objections are segmented and traced on the basis of their comer characteristics and according to the kinetic consistency hypothesis of the rigid body.At the end of the tracing,the inform ation,such as the traffic flow,the average velocity and the vehicle type,is computed.

Key words: comer detection, feature matching, motion tracing, traffic parameter, parameter measurement