• Journal of Infrared and Millimeter Waves
  • Vol. 39, Issue 3, 339 (2020)
Wei FANG1、2, Yan-Li QIAO1, Dong-Ying ZHANG1, Li-Li DU1、*, and Wei-Ning YI1
Author Affiliations
  • 1Key Laboratory of Optical Calibration and Characterization of Chinese Academy of Sciences, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei23003, China
  • 2University of Science and Technology of China, Hefei3006, China
  • show less
    DOI: 10.11972/j.issn.1001-9014.2020.03.012 Cite this Article
    Wei FANG, Yan-Li QIAO, Dong-Ying ZHANG, Li-Li DU, Wei-Ning YI. Angle selection research based on multi-objectives optimized detection of clouds[J]. Journal of Infrared and Millimeter Waves, 2020, 39(3): 339 Copy Citation Text show less
    (a) The surface reflectance radiance (clear),(b) cloud reflectance & scattering (thin cloud),(c) cloud molecule scattering & radiance(thick cloud)
    Fig. 1. (a) The surface reflectance radiance (clear),(b) cloud reflectance & scattering (thin cloud),(c) cloud molecule scattering & radiance(thick cloud)
    The relationship of scattering angles with 865nm polarized reflectance for liquid droplets and ice crystals respectly in cloud
    Fig. 2. The relationship of scattering angles with 865nm polarized reflectance for liquid droplets and ice crystals respectly in cloud
    Three geometrical factors may affect cloud fraction changes with viewing zenith angle
    Fig. 3. Three geometrical factors may affect cloud fraction changes with viewing zenith angle
    The principal of multi-directional viewing by polarization spectral radiometer
    Fig. 4. The principal of multi-directional viewing by polarization spectral radiometer
    Pareto optimal front in objective space
    Fig. 5. Pareto optimal front in objective space
    The Knee on Pareto optimal front
    Fig. 6. The Knee on Pareto optimal front
    Pareto optimal front of R670 dataset 1.A-knee (ideal solution)
    Fig. 7. Pareto optimal front of R670 dataset 1.A-knee (ideal solution)
    The results of cloud detection on dataset 1 for POLDER product (up),and for the proposed method (down)
    Fig. 8. The results of cloud detection on dataset 1 for POLDER product (up),and for the proposed method (down)
    The results of cloud detction on dataset 2 for POLDER product (up),and for the proposed method (down)
    Fig. 9. The results of cloud detction on dataset 2 for POLDER product (up),and for the proposed method (down)
    Synthetic image from original reflectance at 443,565,670 band (left), the binary image of cloud detection by the proposed method(right)
    Fig. 10. Synthetic image from original reflectance at 443,565,670 band (left), the binary image of cloud detection by the proposed method(right)
    Multi-spectral contribution to cloud detection through the proposed method on DPC
    Fig. 11. Multi-spectral contribution to cloud detection through the proposed method on DPC
    组合角度层ang1组合角度层ang2信息散度DKL平均信息熵H(x)¯
    4121.08054.9934
    4130.91055.0158
    4140.87545.0232
    2130.60275.0305
    2140.58475.0379
    3130.57615.0390
    3140.52975.0464
    13140.35605.0473
    Table 1. 数据集1的R6702角度组合Pareto前沿最优解
    数据集波段组合角度层ang1组合角度层ang2
    1B670L2L14
    1B490L11L14
    1B865L2L6
    2B670L2L13
    2B490L12L14
    2B865L8L9
    Table 2. 三波段二角度层理想组合结果
    数据集采用方法云像素(误检数)晴像素(漏检数)未确定符合像素
    1POLDER1067(0)1433(0)02500
    2POLDER1188(0)1312(0)02500
    1该方法1133(106)1364(173)32218
    2该方法1231(90)1261(163)82239
    Table 3. POLDER与该方法检测结果
    数据集误检率(FP)漏检率(FN)总体精度(ACC)Kappa系数
    10.09930.12070.88840.7771
    20.07570.12420.89880.7918
    1+2平均0.08750.12240.89360.7845
    Table 4. 该方法与POLDER方法的评价指标比对
    Wei FANG, Yan-Li QIAO, Dong-Ying ZHANG, Li-Li DU, Wei-Ning YI. Angle selection research based on multi-objectives optimized detection of clouds[J]. Journal of Infrared and Millimeter Waves, 2020, 39(3): 339
    Download Citation