• Acta Optica Sinica
  • Vol. 44, Issue 16, 1610001 (2024)
Xuan Liu1, Bingzhen Li1, Li Li1,*, Weiqi Jin1, and Hongchang Cheng2
Author Affiliations
  • 1Key Laboratory of Photoelectronic Imaging Technology and System, Ministry of Education, Beijing Institute of Technology, Beijing 100081, China
  • 2Science and Technology on Low-Light-Level Night Vision Laboratory, Xi’an 710065, Shaanxi , China
  • show less
    DOI: 10.3788/AOS240702 Cite this Article Set citation alerts
    Xuan Liu, Bingzhen Li, Li Li, Weiqi Jin, Hongchang Cheng. Adaptive Wavelet Threshold Denoising Based on Pixel Dark Noise of EBAPS[J]. Acta Optica Sinica, 2024, 44(16): 1610001 Copy Citation Text show less
    Schematic of EBAPS imaging[20]
    Fig. 1. Schematic of EBAPS imaging[20]
    Ideal total noise PTC curve for CMOS/CCD
    Fig. 2. Ideal total noise PTC curve for CMOS/CCD
    Schematic of experimental setup
    Fig. 3. Schematic of experimental setup
    Schematic of physical object and structural particularity of EBAPS
    Fig. 4. Schematic of physical object and structural particularity of EBAPS
    Different regions of EBAPS image (solid green line box indicates the dark pixel area, and solid red line box indicates the effective signal area)
    Fig. 5. Different regions of EBAPS image (solid green line box indicates the dark pixel area, and solid red line box indicates the effective signal area)
    Signal mean versus standard deviation for difference image
    Fig. 6. Signal mean versus standard deviation for difference image
    Histogram distribution of image with different mean values. (a) Mean value is 93; (b) mean value is 166; (c) mean value is 466; (d) mean value is 749; (e) mean value is 2223; (f) mean value is 3283
    Fig. 7. Histogram distribution of image with different mean values. (a) Mean value is 93; (b) mean value is 166; (c) mean value is 466; (d) mean value is 749; (e) mean value is 2223; (f) mean value is 3283
    Difference image of adjacent frames (enhanced image for valid show)
    Fig. 8. Difference image of adjacent frames (enhanced image for valid show)
    Flow chart of AWT-PDN algorithm
    Fig. 9. Flow chart of AWT-PDN algorithm
    Comparison of single frame images and mean values of 50 images under different illuminances. (a) Single frame images; (b) 50 images
    Fig. 10. Comparison of single frame images and mean values of 50 images under different illuminances. (a) Single frame images; (b) 50 images
    Experiment results of denoising methods for scene 1 under different illuminances. (a) 1×10-1 lx; (b) 5×10-2 lx; (c) 1×10-2 lx; (d) 5×10-3 lx
    Fig. 11. Experiment results of denoising methods for scene 1 under different illuminances. (a) 1×10-1 lx; (b) 5×10-2 lx; (c) 1×10-2 lx; (d) 5×10-3 lx
    Experiment results of denoising methods for scene 2 under different illuminances. (a) 1×10-1 lx; (b) 5×10-2 lx; (c) 1×10-2 lx; (d) 5×10-3 lx
    Fig. 12. Experiment results of denoising methods for scene 2 under different illuminances. (a) 1×10-1 lx; (b) 5×10-2 lx; (c) 1×10-2 lx; (d) 5×10-3 lx
    Type of noiseSourceCharacteristic
    Signal-dependent noise σP2Photon shot noise, dark electron emission noise, and dark current shot noiseRandom distribution in space and time, dependent on signal strength
    Signal-independent noise σN2Transverse electron diffusion and read noiseRandom distribution in space and time, independent on signal strength
    Fixed pattern noise N02Silicon material and processing technologyFixed distribution
    Table 1. Sources of noise in EBAPS images
    Illuminance /lxMethodScene 1Scene 2
    PSNRSSIMAFDPSNRSSIMAFD
    1×10-1Oringinal28.67860.69618.290728.70490.69658.4357
    UT31.38070.83795.181430.69570.80456.1660
    Rigrsure29.51550.74337.307329.45230.73687.5533
    Min-max31.63420.84764.931931.76100.84955.0776
    WTDA30.65270.80645.954130.73110.80636.1212
    Improved-WDM31.66810.84864.949431.76460.84925.1215
    AWT-PDN30.82720.84014.190630.93090.83324.5949
    5×10-2Oringinal28.00760.68798.869227.60410.66129.5644
    UT30.39670.81805.773429.45510.77036.9302
    Rigrsure28.77920.73367.843730.51120.70758.4860
    Min-max30.63770.82635.533627.60410.80995.7640
    WTDA29.78310.79336.463129.58950.77616.7664
    Improved-WDM30.73940.82925.496030.64960.81375.6692
    AWT-PDN31.24510.84074.471731.34210.82874.6236
    1×10-2Oringinal32.00670.78765.272331.23810.75905.9366
    UT34.20140.86403.539632.88640.82594.3909
    Rigrsure32.51980.80864.827531.76420.85195.4190
    Min-max34.42750.86963.400733.81730.7593.7056
    WTDA33.65270.84923.901833.03760.83044.2740
    Improved-WDM34.48970.87143.377533.93200.85503.6452
    AWT-PDN36.34600.90462.522635.81480.89222.9423
    5×10-3Oringinal34.52900.83923.883935.49490.85263.6433
    UT36.77870.89692.565337.30570.89682.5696
    Rigrsure34.97100.85343.565736.07250.86953.3022
    Min-max37.00340.90122.466238.22990.91282.1258
    WTDA36.21720.88552.834737.4810.89992.4854
    Improved-WDM37.01770.90182.46338.29550.91412.0905
    AWT-PDN39.16650.93371.789340.06850.93941.8437
    Table 2. Evaluation indices of image processed by different methods