• High Power Laser and Particle Beams
  • Vol. 34, Issue 12, 129001 (2022)
Luyao Liang1, Xiaoyun Zhao2、*, and Jinquan Zhao3
Author Affiliations
  • 1College of Mathmatics and Physics, Chengdu University of Technology, Chengdu 610000, China
  • 2The college of Nuclear Technology and Automation Engineering, Chengdu University of Technology, Chengdu 610000, China
  • 3School of Geophysics, Chengdu University of Technology, Chengdu 610000, China
  • show less
    DOI: 10.11884/HPLPB202234.220086 Cite this Article
    Luyao Liang, Xiaoyun Zhao, Jinquan Zhao. An automatic focusing algorithm based on U-Net for target location in multiple depth-of-field scene[J]. High Power Laser and Particle Beams, 2022, 34(12): 129001 Copy Citation Text show less
    Multi-depth-of-field images of the object in the foreground and in the background
    Fig. 1. Multi-depth-of-field images of the object in the foreground and in the background
    Focus evaluation function curve in multi-depth-of-field scene
    Fig. 2. Focus evaluation function curve in multi-depth-of-field scene
    The target objects are roughly distributed in the image
    Fig. 3. The target objects are roughly distributed in the image
    Flow chart of target location using U-net
    Fig. 4. Flow chart of target location using U-net
    Flowchart of focusing window algorithm
    Fig. 5. Flowchart of focusing window algorithm
    Flow to locate the target and set the corresponding window
    Fig. 6. Flow to locate the target and set the corresponding window
    Partial images of five sets of images
    Fig. 7. Partial images of five sets of images
    For a group of images, the prediction map obtained by U-net and the images to be evaluated
    Fig. 8. For a group of images, the prediction map obtained by U-net and the images to be evaluated
    The first group of image focus evaluation function curves
    Fig. 9. The first group of image focus evaluation function curves
    Focus evaluation function curves of the second group of images off and with windows
    Fig. 10. Focus evaluation function curves of the second group of images off and with windows
    Focus evaluation function curves of the third group of images off and with windows
    Fig. 11. Focus evaluation function curves of the third group of images off and with windows
    Focus evaluation function curves of the fourth group of images before and after adding windows
    Fig. 12. Focus evaluation function curves of the fourth group of images before and after adding windows
    Focus evaluation function curves of the fifth group of images before and after
    Fig. 13. Focus evaluation function curves of the fifth group of images before and after
    Image obtained by the central window method and obtained by the proposed method
    Fig. 14. Image obtained by the central window method and obtained by the proposed method
    Focus evaluation function of central window method and window method proposed in this paper
    Fig. 15. Focus evaluation function of central window method and window method proposed in this paper
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD246.88750.04117.258201.61371
    Roberts247.34310.04106.864801.64561
    Sobel241.46230.03891.243903.10411
    Brenner238.12750.04205.842601.29901
    SML221.31830.04090.976112.15981
    SMD-Roberts231404.70.0434121.6902.29951
    Table 1. Evaluation indexes of focus evaluation functions of group one
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD/SMD-W29/252.1698/2.51570.0325/0.02831.9085/0.8393/02.0453/2.16111/1
    Roberts/Roberts-W29/252.2958/2.56870.0324/0.02781.7515/0.76712/02.2270/2.09751/1
    Sobel/Sobel-W29/281.5113/1.72460.0307/0.03151.8000/2.90750/03.6727/3.71811/1
    Brenner/Brenner-W30/302.8219/3.77610.0311/0.03141.5077/4.08072/01.6520/1.59941/1
    SML/SML-W28/271.1721/1.27260.0344/0.03351.7768/3.74762/22.5761/2.66161/1
    Table 2. Evaluation indexes of focus evaluation functions of group two
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD/SMD-W6/1118.6244/10.05820.1372/0.08913.1678/21.5561/01.6038/1.63481/1
    Roberts/Roberts-W6/1114.2448/7.63390.1328/0.08862.2774/15.9691/01.8546/1.67421/1
    Sobel/Sobel-W8/113.0674/3.76500.1005/0.08711.9598/9.00971/03.2885/3.30391/1
    Brenner/Brenner-W6/1010.2044/11.61040.1269/0.09422.0341/10.8801/01.2648/1.26601/1
    SML/SML-W7/111.43591/9.37510.0734/0.08900.3356/19.3141/02.4652/1.92950.86/1
    Table 3. Evaluation indexes of focus evaluation functions of group three
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD/SMD-W22/2112.374/3.56260.0448/0.04593.8276/5.03361/01.5648/1.55510.90/1
    Roberts/Roberts-W22/2113.184/3.44880.0449/0.04583.6610/4.71001/01.5518/1.67370.90/1
    Sobel/Sobel-W22/211.9829/1.67980.0421/0.04401.0151/1.76242/12.9408/2.97970.90/1
    Brenner/Brenner-W22/2212.778/10.2170.0448/0.04504.4590/6.58562/11.2274/1.22550.90/1
    SML/SML-W22/211.4090/1.28700.0409/0.03970.7853/0.80941/12.5329/1.93050.90/1
    Table 4. Evaluation indexes of focus evaluation functions of group four
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD/SMD-W29/263.5789/2.30220.0335/0.03770.0334/2.95701/02.1879/2.10250.94/1
    Roberts/Roberts-W29/263.8800/2.30540.0334/0.03761.9252/2.73581/12.2622/2.09140.94/1
    Sobel/Sobel-W27/281.5170/1.47990.0324/0.03470.8593/0.82981/14.4030/3.92671/1
    Brenner/Brenner-W27/284.3384/4.48700.0352/0.03532.1226/1.84851/11.7418/1.69391/1
    SML/SML-W28/281.3850/1.35310.0345/0.03531.2580/0.87972/12.7556/2.65461/1
    Table 5. Evaluation indexes of focus evaluation functions of group five
    functionw/pix R/pix S/pix δSE/pix ɑ/pcs Τ/s ϛ/pcs
    SMD-C/SMD-W10/113.5506/10.05820.0861/0.08910.0861/21.5563/01.6262/1.59071/1
    Roberts-C/Roberts-W10/112.8268/7.63390.08214/0.08861.9627/15.9693/01.6108/1.67421/1
    Sobel-C/Sobel-W8/113.0189/3.76500.1045/0.08712.3171/9.00971/02.9433/3.30391/1
    Brenner-C/Brenner-W10/103.0372/11.6100.0804/0.09421.9952/10.8804/01.2488/1.26601/1
    SML-C/SML-W11/116.4727/1.60530.0827/0.07464.3409/1.88730/01.8998/2.18361/1
    Table 6. Evaluation indexes of focus evaluation functions of center window method and proposed method
    Luyao Liang, Xiaoyun Zhao, Jinquan Zhao. An automatic focusing algorithm based on U-Net for target location in multiple depth-of-field scene[J]. High Power Laser and Particle Beams, 2022, 34(12): 129001
    Download Citation