• Laser & Optoelectronics Progress
  • Vol. 56, Issue 16, 162804 (2019)
Jingfeng Hu2, Xiuzai Zhang1、2, and Changjun Yang3、*
Author Affiliations
  • 1 Jiangsu Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, Nanjing University of Information Science and Technology, Nanjing, Jiangsu 210044, China
  • 2 School of Electronic and Information Engineering, Nanjing University of Information Science and Technology, Nanjing, Jiangsu 210044, China
  • 3 National Satellite Meteorological Center, Key Laboratory of Radiometric Calibration and Validation for Environmental Satellites, China Meteorological Administration, Beijing 100081, China
  • show less
    DOI: 10.3788/LOP56.162804 Cite this Article Set citation alerts
    Jingfeng Hu, Xiuzai Zhang, Changjun Yang. Cloud Detection of RGB Color Remote Sensing Images Based on Improved M-Net[J]. Laser & Optoelectronics Progress, 2019, 56(16): 162804 Copy Citation Text show less
    Convolution unit and residual unit. (a) Forward propagation convolution unit; (b) residual unit
    Fig. 1. Convolution unit and residual unit. (a) Forward propagation convolution unit; (b) residual unit
    RM-Net network structure
    Fig. 2. RM-Net network structure
    DDCN network structure
    Fig. 3. DDCN network structure
    Dataset enhancement. (a) Original image; (b) horizontal flip; (c) vertical flip; (d) horizontal and vertical flips; (e) saturation adjustment; (f) brightness adjustment; (g) color adjustment; (h) add noise
    Fig. 4. Dataset enhancement. (a) Original image; (b) horizontal flip; (c) vertical flip; (d) horizontal and vertical flips; (e) saturation adjustment; (f) brightness adjustment; (g) color adjustment; (h) add noise
    Overall accuracy curves
    Fig. 5. Overall accuracy curves
    Visual comparison of Landsat8 image cloud detection results obtained by six methods
    Fig. 6. Visual comparison of Landsat8 image cloud detection results obtained by six methods
    Visual comparison of GaoFen-1 WFV image cloud detection results obtained by six methods
    Fig. 7. Visual comparison of GaoFen-1 WFV image cloud detection results obtained by six methods
    Detection results of cloud and cloud shadow
    Fig. 8. Detection results of cloud and cloud shadow
    MethodPPrecisionRRecallAAccuracyF1score
    k-means0.83660.65850.83960.7369
    CNN+SP0.86050.90250.90640.8704
    FCN2s0.92930.87340.92430.9005
    M-Net0.94320.90910.96730.9258
    DDCN0.93220.92830.97280.9302
    RM-Net0.93340.95090.98160.9418
    Table 1. Quantitative comparison of Landsat8 image cloud detection results obtained by six methods
    MethodPPrecisionRRecallAAccuracyF1score
    k-means0.74990.71540.83560.7322
    CNN+SP0. 84130.86350.89140.8523
    FCN2s0.90190.89350.92380.8976
    M-Net0.93070.90390.95900.9132
    DDCN0.93160.92730.96540.9294
    RM-Net0.92650.93530.97620.9309
    Table 2. Quantitative comparison of GaoFen-1 WFV image cloud detection results obtained by six methods
    MethodP'PrecisionR'RecallA'AccuracyF'1score
    DDCN0.77340.71160.93560.7412
    RM-Net0.86570.79420.97030.8284
    Table 3. Quantitative comparison of detection results of cloud and cloud shadow
    Jingfeng Hu, Xiuzai Zhang, Changjun Yang. Cloud Detection of RGB Color Remote Sensing Images Based on Improved M-Net[J]. Laser & Optoelectronics Progress, 2019, 56(16): 162804
    Download Citation