Journal of Applied Optics, Volume. 44, Issue 6, 1332(2023)

Bounding box initialization and optimization for ground target tracking in UAV Videos

Chuwei LI... Zhilong ZHANG* and Ping ZHONG |Show fewer author(s)
Author Affiliations
  • National Key Laboratory of Science and Technology on Automatic Target Recognition, College of Electronic Science and Technology, National University of Defense Technology, Changsha 410073, China
  • show less
    Figures & Tables(17)
    Typical initial tracking box in physical systems
    Tracking results of different initialization methods
    Effect of square initial boxes with different deviations on tracking success value
    Effect of square initial boxes with different deviations on tracking precision value
    Strategy of tracking box initialization and optimization
    Global saliency map of UAV ground images
    Local saliency map of UAV ground images
    Comparison of image segmentation and salient region segmentation
    Flow chart of tracking box optimization algorithm
    Overall success plot and precision plot on VisDrone2018-SOT-test-dev dataset
    Overall success plot and precision plot on UAVDT dataset
    Limitations of tracking box optimization algorithms
    Role of distance normalization threshold
    Role of three-tier image pyramid structure
    • Table 1. Comparison of object tracking initialization methods

      View table
      View in Article

      Table 1. Comparison of object tracking initialization methods

      类别方法是否基于先验速度特点
      自动模板匹配[5]准确性低
      自动目标识别[6, 9]不具备通用性
      自动差分法[7-8]不适合复杂场景
      自动视觉显著性[10-11]不适合复杂场景
      手动任意长宽比矩形[12]不适合运动场景
      手动正方形框不适合狭长目标
      半自动正方形框+自适应优化(本文)适合复杂场景,兼具通用性
    • Table 2. Algorithm pseudocode

      View table
      View in Article

      Table 2. Algorithm pseudocode

      算法1:基于视觉显著性和显著区域分割的跟踪框自适应优化算法
      输入:原始图像I,初始正方形跟踪框B
      输出:优化后的跟踪框B'
      1.根据初始正方形跟踪框B,从原始图像I中裁剪3个图像切片C1C2C3(分别为正方形跟踪框的1.25、1.0和0.8倍大小);
      2.使用视觉显著性算法(BMS)对C1C2C3进行显著性检测,得到3幅显著图S1S2S3
      3.基于S1S2S3,分别使用二倍阈值分割和Saliency Cut对原始图像进行分割,得到6幅二值化掩模图像{Mi}i=16
      4.{Mi}i=16映射到I,计算对应的最小外接矩形位置{Bi}i=16
      5.计算{Bi}i=16的形心与B中心的归一化距离{Di}i=16
      6.计算{Bi}i=16的最长边与B边长的比值{Ri}i=16
      7. if{DiTdist&&RiTratio}i=16 then
      8.i*=argmaxi(Ri(TdistDi))
      9.B'Bi*
      10.else
      11.B'B
    • Table 3. Parameters of bounding rectangle

      View table
      View in Article

      Table 3. Parameters of bounding rectangle

      外接矩形在图像中的位置形心与B中心的归一化距离最长边与B 边长的比值综合得分
      B1[622, 544, 680, 625]0.03850.69230.1118
      B2[615, 530, 680, 638]0.00960.92310.1758
      B3[616, 530, 679, 626]0.05620.82050.1180
      B4[622, 544, 680, 625]0.03850.69230.1118
      B5[632, 556, 668, 611]0.03110.47010.0794
      B6[617, 530, 678, 624]0.06470.80340.1087
    Tools

    Get Citation

    Copy Citation Text

    Chuwei LI, Zhilong ZHANG, Ping ZHONG. Bounding box initialization and optimization for ground target tracking in UAV Videos[J]. Journal of Applied Optics, 2023, 44(6): 1332

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Jan. 9, 2023

    Accepted: --

    Published Online: Mar. 12, 2024

    The Author Email: ZHANG Zhilong (张志龙)

    DOI:10.5768/JAO202344.0602003

    Topics