• Acta Photonica Sinica
  • Vol. 52, Issue 4, 0410005 (2023)
Dezhen YANG1,2, Jingying HUANG1, Songlin YU1,*, Jinjun FENG2..., Jiangyong LI1 and Tong LIU1|Show fewer author(s)
Author Affiliations
  • 1North China Research Institute of Electro-optics, Beijing 100015, China
  • 2Beijing Vacuum Electronics Research Institute, Beijing 100015, China
  • show less
    DOI: 10.3788/gzxb20235204.0410005 Cite this Article
    Dezhen YANG, Jingying HUANG, Songlin YU, Jinjun FENG, Jiangyong LI, Tong LIU. Single-frame Infrared Dim Target Detection Based on Guided Filter and Segmented Adaptive Thresholds[J]. Acta Photonica Sinica, 2023, 52(4): 0410005 Copy Citation Text show less
    Target characteristics for different backgrounds
    Fig. 1. Target characteristics for different backgrounds
    The overall algorithm flow
    Fig. 2. The overall algorithm flow
    Background suppression results
    Fig. 3. Background suppression results
    Heat maps corresponding to grayscale values for different areas in the image(zone 2 is the target region)
    Fig. 4. Heat maps corresponding to grayscale values for different areas in the image(zone 2 is the target region)
    Schematic of background suppression results(zone 0 represents the target area and zones 1~8 represents its 8 neighborhoods)
    Fig. 5. Schematic of background suppression results(zone 0 represents the target area and zones 1~8 represents its 8 neighborhoods)
    Background suppression results of different regions and their 3-dimensional characteristics
    Fig. 6. Background suppression results of different regions and their 3-dimensional characteristics
    The influence of different parameters m on the value k mapping curve
    Fig. 7. The influence of different parameters m on the value k mapping curve
    Ablation experiment results
    Fig. 8. Ablation experiment results
    Example of different test video figure
    Fig. 9. Example of different test video figure
    Five typical scenario test figure
    Fig. 10. Five typical scenario test figure
    Comparison of visual results of each method in five typical scenarios
    Fig. 11. Comparison of visual results of each method in five typical scenarios
    The hardware implementation effect of the proposed method in different complexity scenarios
    Fig. 12. The hardware implementation effect of the proposed method in different complexity scenarios
    Data setN=1N=2N=4
    Data187.97%87.97%87.97%
    Data582.87%87.73%84.93%
    Data983.46%85.71%86.22%
    Table 1. Ablation experiments test recall comparison results
    Data setRaw imageTop-HatLCMMax-MedOurs
    SNRSNRBSFSNRBSFSNRBSFSNRBSF
    Fig. 10(a)0.474.888.183.641.170.706.8279.58371.45
    Fig. 10(b)0.430.303.051.471.970.053.581.1815.88
    Fig. 10(c)0.400.471.002.411.130.151.332.2116.30
    Fig. 10(d)0.433.595.295.001.221.688.079.4746.99
    Fig. 10(e)0.430.815.155.531.591.1011.913.6338.93
    Table 2. Comparison of SNR and BSF index results of each method
    Data setTop-HatLCMMax-MedOurs
    Data179.45%79.20%1.25%87.97%
    Data583.67%59.87%81.40%84.93%
    Data973.68%47.62%76.69%86.22%
    Table 3. Comparison of recall rates of various methods in different video data detection
    Dezhen YANG, Jingying HUANG, Songlin YU, Jinjun FENG, Jiangyong LI, Tong LIU. Single-frame Infrared Dim Target Detection Based on Guided Filter and Segmented Adaptive Thresholds[J]. Acta Photonica Sinica, 2023, 52(4): 0410005
    Download Citation