Due to the recent progress in computer vision to interpret images and sequence of images, the video camera is a promising sensor for traffic monitoring and traffic surveillance at low cost. This paper focuses on the detection and tracking of multiple vehicles present in the field of view of a camera. Until now, the vehicle detection has been mainly performed by the widely used technique called background subtraction, which is based on detecting changes in an image sequence. While there has been long research on this technique, it still faces many challenges. We present in this paper a new framework to detect vehicles, based on a hierarchy of features detection and fusion. The first layer of the hierarchy extract image features. The next layer fuses image features to detect vehicle features such as headlights or windshields. A last layer fuses the vehicle features to detect a vehicle with more confidence. This approach is thus road illumination agnostic and allows vehicles to be detected day and night. The vehicle features are tracked over frames. We use a constant acceleration tracking model augmented with traffic-domain rules to handle the occlusions challenges.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Video-based traffic monitoring at day and night vehicle features detection tracking


    Contributors:


    Publication date :

    2009-10-01


    Size :

    525778 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English





    Rear-Lamp Vehicle Detection and Tracking in Low-Exposure Color Video for Night Conditions

    O'Malley, Ronan / Jones, Edward / Glavin, Martin | IEEE | 2010


    Traffic Parameter Extraction using Video-based Vehicle Tracking

    Jung, Y. K. / Ho, Y. S. / IEEE et al. | British Library Conference Proceedings | 1999


    Research on highway night traffic event detection method based on video processing

    Bao, Lixia / Wang, Qiulan / Zuo, Shuxia et al. | IEEE | 2021