Journal of South China University of Technology (Natural Science Edition) ›› 2014, Vol. 42 ›› Issue (3): 52-58.doi: 10.3969/j.issn.1000-565X.2014.03.009

• Traffic & Transportation Engineering • Previous Articles     Next Articles

Multi- Sensor Signature Fusion Algorithm for Vehicle Type Classification

Tian Yin1 Dong Hong- hui1 Jia Li- min1 Wang Lei- yun2 Li Si- yu3   

  1. 1.School of Traffic and Transportation,Beijing Jiaotong University,Beijing 100044,China; 2.School of Civil Engineering andTransportation,South China University of Technology, Guangzhou 510640,Guangdong,China; 3.Shenzhen Graduate School,Peking University,Shenzhen 518055,Guangdong,China
  • Received:2013-06-06 Revised:2013-11-19 Online:2014-03-25 Published:2014-02-19
  • Contact: 贾利民(1963-)，男，教授，博士生导师，主要从事智能运输系统、交通安全工程研究． E-mail:jialm@vip.sina.com
  • About author:田寅(1986-)，男，博士生，主要从事交通系统工程、交通传感器网络研究．E-mail:10114241@bjtu．edu．cn
  • Supported by:

    国家 “863” 计划项目(2012AA112401);国家自然科学基金资助项目(61104164)

PDF

231

Abstract:

In order to avoid the error of vehicle type classification with single- geomagnetic sensor,a geomagneticsensor network is constructed and is used to create a multi- sensor signature fusion algorithm for vehicle type classifi-cation.In this algorithm,the statuses of moving vehicles are determined on the basis of correlations of different ve-hicle signatures,and vehicle data are fused via the maximum likelihood in combination with the Pearson correlationcoefficient,so that more accurate vehicle signatures are acquired and more precise inputs for vehicle type classifica-tion are provided.Practical road experiments show that,in comparison with the existing single- sensor classificationalgorithms,this newly proposed algorithm helps the classification accuracy for medium- or large- size vehicles im-prove by 17.5%

Key words: transportation system engineering, vehicle type classification, traffic sensor network, correlationanalysis, maximum likelihood estimation