• Laser & Optoelectronics Progress
  • Vol. 57, Issue 22, 221009 (2020)
Ziqi Shao1, Haihong Jin1、2, Lijin Qian1, and Zhiguo Fan1、*
Author Affiliations
  • 1School of Computer Science and Information Engineering, Hefei University of Technology, Hefei, Anhui 230601, China
  • 2School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China
  • show less
    DOI: 10.3788/LOP57.221009 Cite this Article Set citation alerts
    Ziqi Shao, Haihong Jin, Lijin Qian, Zhiguo Fan. Method of Haze Image Reconstruction Based on PolarizationLayering and Analysis of Airlight[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221009 Copy Citation Text show less
    Diagram of haze image reconstruction method
    Fig. 1. Diagram of haze image reconstruction method
    Gradient characteristics of nature image. (a) Nature scene image; (b) gradient histogram; (c) Laplace distribution
    Fig. 2. Gradient characteristics of nature image. (a) Nature scene image; (b) gradient histogram; (c) Laplace distribution
    Gradient characteristics of airlight under different dense haze
    Fig. 3. Gradient characteristics of airlight under different dense haze
    Airlight gradient prior. (a) Gradient histogram of airlight; (b) model of airlight gradient prior
    Fig. 4. Airlight gradient prior. (a) Gradient histogram of airlight; (b) model of airlight gradient prior
    Influence of ξ for image layering
    Fig. 5. Influence of ξ for image layering
    Layering results for polarized airlight images under 0° channel. (a) Polarized haze image; (b) object texture layer; (c) airlight layer; (d) estimated polarized airlight image by proposed method
    Fig. 6. Layering results for polarized airlight images under 0° channel. (a) Polarized haze image; (b) object texture layer; (c) airlight layer; (d) estimated polarized airlight image by proposed method
    Estimated results for polarized airlight images. (a) Polarized haze image; (b) object texture layer; (c) airlight layer; (d) estimated polarized airlight image using the proposed method
    Fig. 7. Estimated results for polarized airlight images. (a) Polarized haze image; (b) object texture layer; (c) airlight layer; (d) estimated polarized airlight image using the proposed method
    Results of haze image reconstruction. (a) Haze intensity image; (b) estimated airlight image by proposed method; c) reconstructed image using the proposed method
    Fig. 8. Results of haze image reconstruction. (a) Haze intensity image; (b) estimated airlight image by proposed method; c) reconstructed image using the proposed method
    Comparison results of gray histogram and gradient histogram. (a) Haze image and reconstructed image; b) gray histogram; (c) gradient histogram
    Fig. 9. Comparison results of gray histogram and gradient histogram. (a) Haze image and reconstructed image; b) gray histogram; (c) gradient histogram
    Reconstructed results for haze image with different dense haze. (a) Haze image; (b) dehazing results of Schechner method; (c) dehazing results of proposed method
    Fig. 10. Reconstructed results for haze image with different dense haze. (a) Haze image; (b) dehazing results of Schechner method; (c) dehazing results of proposed method
    Degree of polarization(Dop) image and its gray histogram of haze image and airlight image. (a) Haze image; (b) Dop image of haze image; (c) gray histogram of haze image; (d) estimated airlight image by proposed method; (e) Dop image of airlight image; (f) gray histogram of airlight image
    Fig. 11. Degree of polarization(Dop) image and its gray histogram of haze image and airlight image. (a) Haze image; (b) Dop image of haze image; (c) gray histogram of haze image; (d) estimated airlight image by proposed method; (e) Dop image of airlight image; (f) gray histogram of airlight image
    Haze densityHaze imageSchechner's methodProposed method
    Degree ofcontrastAveragegradientDegree ofcontrastAveragegradientDegree ofcontrastAveragegradient
    Mist haze8.316.4915.2410.8323.4146.57
    Middle haze5.353.258.325.2112.6327.47
    Big haze4.222.966.164.009.7622.93
    Dense haze3.202.354.542.666.7317.38
    Table 1. Objective evaluation results for different dehazing methods
    Ziqi Shao, Haihong Jin, Lijin Qian, Zhiguo Fan. Method of Haze Image Reconstruction Based on PolarizationLayering and Analysis of Airlight[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221009
    Download Citation