Automatic Traffic Surveillance System for Vehicle Tracking and Classification

In this paper, the authors propose a novel automatic traffic surveillance system designed to detect, track, and recognize vehicles from various video sequences. Using only one camera, the proposed method is able to categorize vehicles into more specific classes through a new linearity feature in vehicle representation. The proposed system can also handle the problem of vehicle occlusions caused by shadows using a novel line-based shadow elimination algorithm. Finally, the proposed system also features an automatic method for detecting lane-dividing lines, which can ultimately enhance the accuracy of vehicle classification.

  • Availability:
  • Authors:
    • Hsieh, Jun-Wei
    • Yu, Shih-Hao
    • Chen, Yung-Sheng
    • Hu, Wen-Fong
  • Publication Date: 2006-6


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01055794
  • Record Type: Publication
  • Source Agency: UC Berkeley Transportation Library
  • Files: BTRIS, TRIS
  • Created Date: Aug 20 2007 5:42PM