Laser & Optoelectronics Progress, Volume. 58, Issue 6, 610003(2021)

Kernel Correlation Filtering Tracking Algorithm Based on Multi-Target Video Images Edge Feature

Zhang Bo* and Liu Hongping
Author Affiliations
  • College of Information Science and Engineering, Changsha Normal University, Changsha, Hunan 410100, China
  • show less
    Figures & Tables(6)
    Multi-target tracking results of our algorithm. (a) Frame100; (b) frame500; (c) frame1000
    • Table 1. Multi-target tracking success rates of different algorithms unit: %

      View table

      Table 1. Multi-target tracking success rates of different algorithms unit: %

      Frame No.OursRef. [3]Ref. [5]Ref. [6]Ref. [8]Ref. [11]Ref. [12]
      10099.8598.5298.1898.2698.0598.3498.28
      20099.7698.1698.3298.4698.4698.2898.76
      30099.6598.2498.2898.3598.2698.1698.64
      40099.5898.6198.0698.1697.8598.5198.28
      50099.5198.7598.4698.2897.9898.7498.64
      60099.4698.6298.3298.1697.5898.6598.85
      70099.4198.2798.4898.0598.1698.1697.85
      80099.3898.1698.6498.3498.2598.2597.64
      90099.2798.2798.2898.4698.4397.8597.58
      100099.1898.6798.3497.8998.2597.7697.28
    • Table 2. Central errors of different algorithms unit: m

      View table

      Table 2. Central errors of different algorithms unit: m

      Frame No.OursRef. [3]Ref. [5]Ref. [6]Ref. [8]Ref. [11]Ref. [12]
      1000.0250.2510.0270.3520.0290.1520.185
      2000.0860.3450.0880.2850.0890.1520.165
      3000.0750.3520.0780.1950.0790.0950.195
      4000.0350.2680.2560.1370.1650.1250.128
      5000.0560.3450.3540.2320.1280.1350.194
      6000.0350.4250.4510.2850.1690.1520.098
      7000.0480.4850.3520.3740.1850.1640.125
      8000.0680.4690.2880.2850.2680.1750.197
      9000.0750.4250.2670.1950.3490.2540.254
      10000.0810.3850.3540.1350.3890.2320.236
    • Table 3. Central errors of different algorithms in complex environment unit: m

      View table

      Table 3. Central errors of different algorithms in complex environment unit: m

      EnvironmentOursRef. [3]Ref. [5]Ref. [6]Ref. [8]Ref. [11]Ref. [12]
      Slight occlusion0.0670.2850.2640.3250.2160.2510.185
      Severe occlusion0.0750.3120.2120.3150.1850.1360.164
      Shield leaves0.0850.3520.2340.2850.1650.1520.132
      Shade is similar in color0.0580.3450.2840.2740.1740.1340.152
      Color of the shelter is obviously different0.0340.3690.2360.2650.1980.1290.147
      Size change0.0420.3450.2480.2480.2640.1850.196
      Illumination change0.0640.2850.2130.2360.2340.1850.254
      Angle change0.0720.2490.3130.2710.3150.2410.264
    • Table 4. Tracking overlap rates of different algorithms unit: %

      View table

      Table 4. Tracking overlap rates of different algorithms unit: %

      Frame No.OursRef. [3]Ref. [5]Ref. [6]Ref. [8]Ref. [11]Ref. [12]
      10098.7697.5897.4897.8597.8597.5897.52
      20098.5897.5297.2897.8697.9497.4697.51
      30098.6197.1697.6497.4897.9597.2697.26
      40098.6496.2597.5897.3497.1897.5497.54
      50099.2896.4896.2597.1896.2597.6397.67
      60099.3496.3496.4897.2696.4397.5497.79
      70098.6496.8196.1297.4996.1696.8596.85
      80098.1696.4896.3496.5896.8596.7196.79
      90098.2797.2596.8596.7696.2396.8597.84
      100098.6497.4697.1196.2496.7196.1897.97
    • Table 5. Number of frames transmitted per second by different algorithms unit: frame

      View table

      Table 5. Number of frames transmitted per second by different algorithms unit: frame

      EnvironmentOursRef. [3]Ref. [5]Ref. [6]Ref. [8]Ref. [11]Ref. [12]
      Slight occlusion86.5137.5851.6426.5210.529.8518.64
      Severe occlusion79.8441.5267.8523.5416.8510.5425.46
      Shield leaves66.8559.5243.5228.6418.6411.2621.34
      Shade is similar in color71.5858.6141.2539.5219.5212.4623.52
      Color of the shelter is obviously different88.6448.6165.8537.5125.6413.5219.52
      Size change89.3139.5261.2541.5834.5215.6731.54
      Illumination change71.5247.6258.7642.6428.6416.8527.54
      Angle change75.8746.2551.3232.4639.5414.5216.52
    Tools

    Get Citation

    Copy Citation Text

    Zhang Bo, Liu Hongping. Kernel Correlation Filtering Tracking Algorithm Based on Multi-Target Video Images Edge Feature[J]. Laser & Optoelectronics Progress, 2021, 58(6): 610003

    Download Citation

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

    Category: Image Processing

    Received: Jun. 17, 2020

    Accepted: --

    Published Online: Mar. 6, 2021

    The Author Email: Bo Zhang (zb801121@126.com)

    DOI:10.3788/LOP202158.0610003

    Topics