In satellite video vehicle tracking, due to the tracking failure and tracking loss caused by similar characteristics of the target and obstacle occlusion, respectively, the traffic flow extraction accuracy is reduced. To address these issues, an improved traffic flow extraction method for satellite video based on kernelized correlation filter (KCF) was proposed. First, we introduced a multifeature fusion strategy into the KCF based on the discrete Fourier transform (DFT) framework to enhance vehicle tracking accuracy and reduce tracking drift and jumps. Second, we utilized the Kalman filter for trajectory prediction to reduce the loss of target during vehicle tracking. Compared with other mainstream algorithms on the satellite video dataset, the results showed that the tracking accuracy and success rate of the proposed method reached 86.74% and 79.96%, respectively. Finally, the virtual detection line method was used to extract the traffic flow. The experimental results showed that compared with the real traffic flow data obtained by visual method, the accuracy of satellite video traffic flow extraction by virtual detection line was 98.48% under noncongestion condition and 90.18% under congestion condition.


    Access

    Download


    Export, share and cite



    Title :

    An Improved Kernelized Correlation Filter for Extracting Traffic Flow in Satellite Videos


    Contributors:
    Dudu Guo (author) / Hongbo Shuai (author) / Jie Zhang (author) / Yang Wang (author) / Miao Sun (author)


    Publication date :

    2025




    Type of media :

    Article (Journal)


    Type of material :

    Electronic Resource


    Language :

    Unknown




    A Visual Target Tracking Algorithm Based on Improved Kernelized Correlation Filters

    Zhang, Yanghong / Zeng, Chunnian / Liang, Hong et al. | British Library Conference Proceedings | 2016





    Real-Time Scalable Visual Tracking via Quadrangle Kernelized Correlation Filters

    Ding, Guiguang / Chen, Wenshuo / Zhao, Sicheng et al. | IEEE | 2018