• Acta Optica Sinica
  • Vol. 41, Issue 18, 1815001 (2021)
Zongda Liu, Liquan Dong*, Yuejin Zhao, Lingqin Kong, and Ming Liu
Author Affiliations
  • School of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, China
  • show less
    DOI: 10.3788/AOS202141.1815001 Cite this Article Set citation alerts
    Zongda Liu, Liquan Dong, Yuejin Zhao, Lingqin Kong, Ming Liu. Adaptive Model Tracking Algorithm for Fast-Moving Targets in Video[J]. Acta Optica Sinica, 2021, 41(18): 1815001 Copy Citation Text show less
    Application of correlation filtering algorithm in visual tracking
    Fig. 1. Application of correlation filtering algorithm in visual tracking
    Flow chart of the correlation filtering target tracking algorithm
    Fig. 2. Flow chart of the correlation filtering target tracking algorithm
    Tracking effect with a learning rate of 0.01
    Fig. 3. Tracking effect with a learning rate of 0.01
    Tracking effect with a learning rate of 0.05
    Fig. 4. Tracking effect with a learning rate of 0.05
    Response graph and maximum response value when tracking is good. (a) Response graph; (b) tracking effect graph
    Fig. 5. Response graph and maximum response value when tracking is good. (a) Response graph; (b) tracking effect graph
    Response graph and maximum response value when tracking is difficult. (a) Response graph; (b) tracking effect graph
    Fig. 6. Response graph and maximum response value when tracking is difficult. (a) Response graph; (b) tracking effect graph
    Response graph and APCE value when tracking is good. (a) Response graph; (b) tracking effect graph
    Fig. 7. Response graph and APCE value when tracking is good. (a) Response graph; (b) tracking effect graph
    Response graph and APCE value when tracking is difficult. (a) Response graph; (b) tracking effect graph
    Fig. 8. Response graph and APCE value when tracking is difficult. (a) Response graph; (b) tracking effect graph
    Tracking results of fast motion objects by different algorithms. (a) Fixed model; (b) adaptive model
    Fig. 9. Tracking results of fast motion objects by different algorithms. (a) Fixed model; (b) adaptive model
    OPE curves of different algorithms. (a) Average precision; (b) success rate
    Fig. 10. OPE curves of different algorithms. (a) Average precision; (b) success rate
    OPE curves of different algorithms in fast motion target sequences. (a) Average precision; (b) success rate
    Fig. 11. OPE curves of different algorithms in fast motion target sequences. (a) Average precision; (b) success rate
    OPE curves of different algorithms in the motion blur target sequence. (a) Average precision; (b) success rate
    Fig. 12. OPE curves of different algorithms in the motion blur target sequence. (a) Average precision; (b) success rate
    OPE curves of different algorithms in the rapid deformation target sequences. (a) Average precision; (b) success rate
    Fig. 13. OPE curves of different algorithms in the rapid deformation target sequences. (a) Average precision; (b) success rate
    Test results of our algorithm on the skier data set
    Fig. 14. Test results of our algorithm on the skier data set
    Verification result of the recheck mechanism. (a) Target is blurred; (b) target is deformed; (c) poor continuity; (d) dislocation tracking
    Fig. 15. Verification result of the recheck mechanism. (a) Target is blurred; (b) target is deformed; (c) poor continuity; (d) dislocation tracking
    OPE curve of our algorithm on the self-made data set. (a) Average precision; (b) success rate
    Fig. 16. OPE curve of our algorithm on the self-made data set. (a) Average precision; (b) success rate
    AlgorithmOursLCTDSSTKCFCSKTLD
    Fast motionprecision /%80.875.653.263.440.063.3
    success rate /%73.375.954.361.439.954.3
    Motion blueprecision /%81.173.248.762.129.353.0
    success rate /%71.078.050.364.531.748.7
    Rapid deformationprecision /%87.784.056.867.543.549.1
    success rate /%74.782.565.356.535.943.2
    Precision /%87.685.369.371.953.260.3
    Success rate /%76.776.961.563.743.349.6
    Speed /FPS37.530.440.548.126.921.7
    Table 1. Performance analysis of different algorithms
    Zongda Liu, Liquan Dong, Yuejin Zhao, Lingqin Kong, Ming Liu. Adaptive Model Tracking Algorithm for Fast-Moving Targets in Video[J]. Acta Optica Sinica, 2021, 41(18): 1815001
    Download Citation