Optics and Precision Engineering, Volume. 32, Issue 6, 930(2024)

Multi-threshold image segmentation method of QFN chip based on improved grey wolf optimization

Yuan CHAO1,*... Wei XU1, Wenhui LIU1, Zhen CAO1 and Min ZHANG2 |Show fewer author(s)
Author Affiliations
  • 1College of Mechanical Engineering, Jiangsu University of Technology, Changzhou2300, China
  • 2Changzhou Xiangming Intelligent Drive System Corporation, Changzhou13011, China
  • show less
    Figures & Tables(14)
    Flowchart of the IGWO algorithm
    Iterative curve of decay factor
    Original images , the corresponding histograms and segmentation results
    Original images and segmentation results
    Convergence curves of Kapur entropy by five-threshold segmentation
    • Table 1. Initialization parameter settings for each comparative algorithm

      View table
      View in Article

      Table 1. Initialization parameter settings for each comparative algorithm

      算法参数取值
      GWOr1r2r3[0,1]
      DSF-GWOω2.5
      LSSAC2C3[0,1]
      莱维参数1.5
      INGOn1.02e+04
      IGWOr1r2r3[0,1]
      ω11+t/Tmax
      ω2ω31
      k10.8
      k20.2
      k3a
      k4a/2
    • Table 2. Population size and number of iterations settings of algorithms

      View table
      View in Article

      Table 2. Population size and number of iterations settings of algorithms

      组号实验内容种群数量最大迭代次数
      1消融实验50100
      2基准函数实验30500
      3图像分割实验50100
    • Table 3. Results of standard test functions

      View table
      View in Article

      Table 3. Results of standard test functions

      序号基准函数评价指标GWODSF-GWOLSSAINGOIGWO
      1F1mean9.46×10-2806.68×10-501.11e×10-292
      std1.04×10-2701.62×10-400
      2F2mean8.55×10-171.13×10-1950.2503.79×10-148
      std7.25×10-1700.4601.57×10-148
      3F3mean4.54×10-602.04×10-302.37×10-285
      std6.35×10-601.17×10-300
      4F4mean6.85×10-72.51×10-18512.235.67×10-1435.68×10-143
      std6.93×10-702.762.18×10-1432.18×10-143
      5F5mean1.98×10-21.97×1027.29×1061.99×1021.97×102
      std0.410.41.76×10600.05
      6F7mean0.841.047.05×10-57.53.77
      std0.490.492.82×10-400.86
      7F9mean2.49060.4300
      std2.96017.2800
      8F10mean1.03×10-138.89×10-162.028.89×10-168.88×10-16
      std2.06×10-1400.7400
      9F11mean4.28×10-301.39×10-200
      std8.83×10-301.08×10-200
    • Table 4. Description of different improved GWO algorithms

      View table
      View in Article

      Table 4. Description of different improved GWO algorithms

      算法名称改进点
      IGWO1仅改进衰减因子a
      IGWO2仅改进位置更新公式(包括反向学习)
      IGWO3仅引入头狼靠拢策略
      IGWO4仅交替引入头狼靠拢策略和种群变异策略
      IGWO引入上述所有改进点
    • Table 5. Results of ablation experiments for IGWO

      View table
      View in Article

      Table 5. Results of ablation experiments for IGWO

      测试图像评价指标GWOIGWO1IGWO2IGWO3IGWO4IGWO
      Lenamean20.748 820.746 220.758 120.760 720.760 620.760 7
      std0.026 306 110.036 528 490.008 967 500.000 074 660.000 134 870.000 042 90
      best20.760 820.760 820.760 820.760 820.760 820.760 8
      worst20.652 220.616 620.714 720.760 620.760 620.760 6
      t(s)0.731 10.732 20.734 30.733 20.734 40.731 5
      NS42220261128
      Peppermean21.529 621.523 021.525 021.530 821.530 821.530 9
      std0.003 872 040.026 674 060.017 662 540.000 123 850.000 150 910
      best21.530 921.530 921.530 921.530 921.530 921.530 9
      worst21.514 421.409 121.457 321.530 321.530 321.530 9
      t(s)0.727 10.727 10.728 40.751 60.726 30.726 8
      NS102412242030
      Baboonmean20.920 220.913 720.914 020.922 220.922 220.922 2
      std0.010 723 540.026 162 420.017 034 050.000 051 820.000 081 550.000 040 91
      best20.922 220.922 220.922 220.922 220.922 220.922 2
      worst20.863 420.822 320.853 920.922 120.922 020.922 1
      t(s)0.721 30.724 40.724 70.722 50.723 50.721 6
      NS152318272228
      Buildingmean21.388 121.386 821.383 721.392 821.398 221.398 4
      std0.026 430 530.034 148 290.030 042 490.021 259 510.000 308 420.000 264 70
      best21.398 521.398 521.398 521.398 521.398 521.398 5
      worst21.291 821.265 221.287 221.314 621.397 621.397 1
      t(s)0.730 70.731 30.733 50.734 10.733 30.730 1
      NS42017231627
    • Table 6. Segmentation results of comparative algorithms

      View table
      View in Article

      Table 6. Segmentation results of comparative algorithms

      测试

      图像

      分割算法阈值个数
      2345
      阈值熵值阈值熵值阈值熵值阈值熵值
      QFN3GWO(54,117)10.372 0(54,108,143)13.254 3(48,79,111,147)15.936 1(54,112,162,209,255)18.574 5
      DSF-GWO(54,117)10.372 0(54,108,143)13.254 3(51,81,112,147)15.944 9(55,112,162,209,255)18.573 7
      LSSA(54,117)10.372 0(54,108,143)13.254 3(51,81,112,146)15.945 0(55,112,160,207,255)18.573 6
      INGO(54,117)10.372 0(54,108,143)13.254 3(51,81,112,146)15.945 0(54,112,162,209,255)18.574 5
      IGWO(54,117)10.372 0(54,108,143)13.254 3(51,81,112,146)15.945 0(54,112,162,209,255)18.574 5
      QFN4GWO(74,124)10.380 1(61,93,128)13.061 4(61,90,120,148)15.709 6(74,121,169,212,255)18.495 3
      DSF-GWO(74,124)10.380 1(61,90,124)13.064 7(61,88,120,147)15.709 2(74,121,171,212,255)18.494 4
      LSSA(74,124)10.380 1(61,91,128)13.063 1(61,88,120,147)15.709 2(74,121,169,211,255)18.495 8
      INGO(74,124)10.380 1(61,90,124)13.064 7(61,90,120,147)15.709 8(74,121,169,211,255)18.495 8
      IGWO(74,124)10.380 1(61,90,124)13.064 7(61,90,120,147)15.709 8(74,121,169,211,255)18.495 8
    • Table 6. Segmentation results of comparative algorithms

      View table
      View in Article

      Table 6. Segmentation results of comparative algorithms

      测试

      图像

      分割算法阈值个数
      2345
      阈值熵值阈值熵值阈值熵值阈值熵值
      QFN1GWO(60,117)10.347 9(60,109,139)13.112 5(54,81,113,141)15.751 2(60,117,171,210,255)18.379 5
      DSF-GWO(60,117)10.347 9(60,109,139)13.112 5(54,81,113,140)15.751 3(60,115,171,211,255)18.376 8
      LSSA(60,117)10.347 9(60,109,139)13.112 5(54,81,113,140)15.751 3(60,117,171,211,255)18.380 5
      INGO(60,117)10.347 9(60,109,139)13.112 5(54,81,113,140)15.751 3(60,117,171,211,255)18.380 5
      IGWO(60,117)10.347 9(60,109,139)13.112 5(54,81,113,140)15.751 3(60,117,171,211,255)18.380 5
      QFN2GWO(62,119)10.190 7(62,102,131)12.857 3(61,87,115,140)15.418 3(62,119,174,210,255)21.376 7
      DSF-GWO(62,119)10.190 7(65,106,132)12.848 9(61,87,115,140)15.418 3(62,118,175,210,255)21.245 4
      LSSA(62,119)10.190 7(62,102,132)12.857 1(61,87,115,140)15.418 3(62,119,174,210,255)21.376 7
      INGO(62,119)10.190 7(62,102,131)12.857 3(61,87,115,140)15.418 3(62,119,174,210,255)21.376 7
      IGWO(62,119)10.190 7(62,102,131)12.857 3(61,87,115,140)15.418 3(62,119,174,210,255)21.376 7
    • Table 7. PSNR and FSIM of comparative algorithms

      View table
      View in Article

      Table 7. PSNR and FSIM of comparative algorithms

      测试图像评价指标GWODSF-GWOLSSAINGOIGWO
      QFN1PSNR21.203 721.286 320.990 821.308 821.308 8
      FSIM0.875 30.879 40.874 60.880 10.880 1
      QFN2PSNR18.734 818.735 018.983 718.983 718.983 7
      FSIM0.868 00.864 70.868 00.868 00.868 0
      QFN3PSNR19.333 721.568 021.568 021.568 021.639 1
      FSIM0.854 40.897 60.897 60.897 60.898 4
      QFN4PSNR20.876 421.168 520.986 420.986 420.986 4
      FSIM0.864 30.868 00.866 60.866 60.866 6
    • Table 8. Mean values, standard deviation, optimal and worst values of the entropy function and running times of comparative algorithms

      View table
      View in Article

      Table 8. Mean values, standard deviation, optimal and worst values of the entropy function and running times of comparative algorithms

      测试图像评价指标GWODSF-GWOLSSAINGOIGWO
      QFN1mean18.375 118.379 118.370 518.380 118.380 5
      std0.011 379 820.001 213 730.051 017 780.000 849 460.000 330 28
      best18.380 518.380 518.380 518.380 518.380 5
      worst18.338 518.376 118.100 418.376 618.379 2
      t(s)0.359 4830.709 2100.490 4471.416 0690.361 398
      QFN2mean18.151 518.163 318.166 118.167 118.168 0
      std0.034 509 400.003 249 930.002 869 510.002 132 610.000 000 00
      best18.168 018.168 018.168 018.168 018.168 0
      worst18.023 018.151 618.161 518.158 718.168 0
      t(s)0.359 6630.707 9490.492 3871.450 7190.356 361
      QFN3mean18.570 218.573 518.555 518.574 418.574 5
      std0.011 987 280.000 796 100.049 052 140.000 230 290.000 031 12
      best18.574 518.574 518.574 518.574 518.574 5
      worst18.519 118.571 518.431 118.573 618.574 4
      t(s)0.389 9050.752 0730.510 5481.532 2620.382 314
      QFN4mean18.490 818.492 418.495 518.495 618.495 8
      std0.010 392 500.002 819 840.000 679 200.000 532 060
      best18.495 818.495 318.495 818.495 818.495 8
      worst18.448 518.480 818.493 718.493 718.495 8
      t(s)0.366 2490.702 4040.495 7431.417 4330.352 128
    Tools

    Get Citation

    Copy Citation Text

    Yuan CHAO, Wei XU, Wenhui LIU, Zhen CAO, Min ZHANG. Multi-threshold image segmentation method of QFN chip based on improved grey wolf optimization[J]. Optics and Precision Engineering, 2024, 32(6): 930

    Download Citation

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

    Category:

    Received: Aug. 22, 2023

    Accepted: --

    Published Online: Apr. 19, 2024

    The Author Email:

    DOI:10.37188/OPE.20243206.0930

    Topics