Journal of South China University of Technology (Natural Science Edition) ›› 2005, Vol. 33 ›› Issue (10): 19-23.

Previous Articles     Next Articles

Tracking Method for Moving Vehicles Based on Feature-Point Optical Flow and Kalman Filtration

Sun ji-feng  Wang Cheng-qing   

  1. College of Electronic and Information Engineering,South China Univ.of Tech.,Guangzhou 510640,Guangdong,China
  • Received:2004-12-08 Online:2005-10-25 Published:2005-10-25
  • Contact: 孙季丰(1962-)，男，博士，教授，主要从事图像和数字视频处理方面的研究 E-mail:ecjfsun@scut.edu.cn
  • About author:孙季丰(1962-)，男，博士，教授，主要从事图像和数字视频处理方面的研究
  • Supported by:

    广东省工业科技攻关计划资助项目(B2320821)

PDF

331

Abstract:

A tracking method for motion objects is proposed based on the feature-point optical flow,which is then applied to the real-time tracking of moving vehicles in the intelligent traffic system.In the proposed method,the Kalman filtration and the polygonal tracking strategy are adopted,thus simplifying and stabilizing the tracking.Si-mulated results indicate that the proposed method can accurately determine the maneuvering target,extract the fea-ture points,implement the computing and clustering of the feature-point optical flow and recognize moving objects.Moreover,it is of less computation and is easy to implement.

Key words: motion object, real-time tracking, feature-point optical flow, Kalman filtration