• Acta Optica Sinica
  • Vol. 38, Issue 11, 1115002 (2018)
Baoyi Ge*, Xianzhang Zuo**, and Yongjiang Hu
Author Affiliations
  • Department of UAV Engineering, Army Engineering University, Shijiazhuang, Hebei 050003, China
  • show less
    DOI: 10.3788/AOS201838.1115002 Cite this Article Set citation alerts
    Baoyi Ge, Xianzhang Zuo, Yongjiang Hu. Long-Term Object Tracking Based On Feature Fusion[J]. Acta Optica Sinica, 2018, 38(11): 1115002 Copy Citation Text show less
    Framework of proposed tracking algorithm
    Fig. 1. Framework of proposed tracking algorithm
    Object feature visualization. (a) Image; (b) HOG; (c) LSH; (d)-(f) CN
    Fig. 2. Object feature visualization. (a) Image; (b) HOG; (c) LSH; (d)-(f) CN
    Region proposal
    Fig. 3. Region proposal
    Flow chart of proposed algorithm
    Fig. 4. Flow chart of proposed algorithm
    Tracking results of different target detection threshold
    Fig. 5. Tracking results of different target detection threshold
    Tracking results evaluation plots on OTB100
    Fig. 6. Tracking results evaluation plots on OTB100
    Partial object tracking results and overlap rates on OTB100. (a) Panda; (b) Lemming; (c) Blur Owl; (d) Human5; (e) Soccer; (f) Wake board1; (g) Jogging-2
    Fig. 7. Partial object tracking results and overlap rates on OTB100. (a) Panda; (b) Lemming; (c) Blur Owl; (d) Human5; (e) Soccer; (f) Wake board1; (g) Jogging-2
    OPE evaluation on UAV123. (a) Precision; (b) success rate
    Fig. 8. OPE evaluation on UAV123. (a) Precision; (b) success rate
    Tracking results in practical applications. (a) Bicycle; (b) child; (c) girl
    Fig. 9. Tracking results in practical applications. (a) Bicycle; (b) child; (c) girl
    AlgorithmSuccess rateTracking speed /(frame·s-1)
    Proposed0.77828.2
    BACF0.75821.2
    SRDCF0.7174.2
    LCT0.62119.7
    MUSTER0.6762.2
    DSST0.54818.6
    KCF0.536124.1
    Table 1. Success rate of object tracking
    SequenceCharacteristicFrame number
    PandaSV,OCC,DEF,IPR,OPR,OV,LR1000
    LemmingIV,SV,OCC,FM,OPR,OV1336
    Blur OwlSV,MB,FM,IPR631
    Human5SV,OCC,DEF713
    Jogging-2OCC,DEF,OPR307
    SoccerIV,SV,OCC,MB,FM,IPR,OPR,BC392
    Wake board1SV,IPR,DEF,FM141
    Table 2. Characteristics of image sequences in experiment
    AlgorithmSuccess rateTracking speed /(frame·s-1)
    Proposed0.53729.8
    BACF0.50622.3
    SRDCF0.5115.6
    LCT0.32923.9
    MUSTER0.4360.9
    DSST0.31144.7
    KCF0.294307.6
    Table 3. Success rate of object tracking