• Laser & Optoelectronics Progress
  • Vol. 58, Issue 20, 2001001 (2021)
Zhongkai Chen1, Xiaorun Li1、*, and Liaoying Zhao2
Author Affiliations
  • 1College of Electrical Engineering, Zhejiang University, Hangzhou, Zhejiang 310027, China
  • 2Institute of Computer Application Technology, Hangzhou Dianzi University, Hangzhou, Zhejiang 310018, China
  • show less
    DOI: 10.3788/LOP202158.2001001 Cite this Article Set citation alerts
    Zhongkai Chen, Xiaorun Li, Liaoying Zhao. Inversion of Suspended Particulate Matter Concentration in Maozhou River Based on Band Selection of Hyperspectral Data[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2001001 Copy Citation Text show less
    Structural diagram of multilayer perceptron with 2 hidden layers
    Fig. 1. Structural diagram of multilayer perceptron with 2 hidden layers
    Correlation matrices of remote sensing reflectance data. (a) March 11, 2019; (b) August 5, 2019
    Fig. 2. Correlation matrices of remote sensing reflectance data. (a) March 11, 2019; (b) August 5, 2019
    Neural network structure in PNN algorithm
    Fig. 3. Neural network structure in PNN algorithm
    Sum of entropy of different algorithms when M=12. (a) March 11, 2019; (b) August 5, 2019
    Fig. 4. Sum of entropy of different algorithms when M=12. (a) March 11, 2019; (b) August 5, 2019
    DateCminCmaxCmeanCstd
    2019-03-114.0018.006.898.10
    2019-08-054.0014.008.0010.32
    Table 1. Statistical results of SPM concentration mg·L-1
    DateSpectral resolutionMinimum wavelengthMaximum wavelength
    2019-03-111.98399.87997.58
    2019-08-051.00325.001075.00
    Table 2. Statistical results of hyperspectral data measured by handheld spectrometersnm
    Number of bandsMSEMAE
    DTBSSVM-RFEPNN-L1PNN-L2PNN-ReLUDTBSSVM-RFEPNN-L1PNN-L2PNN-ReLU
    620.1311.696.757.105.582.782.101.431.041.15
    912.7711.534.778.405.812.031.461.101.271.14
    126.2612.785.586.315.891.242.221.171.231.18
    1512.996.959.045.116.402.031.601.301.231.18
    1816.2616.964.354.417.002.422.050.951.021.02
    2119.328.853.328.535.092.881.540.801.381.21
    2410.115.1210.884.045.191.811.081.640.851.19
    2713.1820.844.655.797.622.042.741.001.031.13
    3011.9113.725.815.167.421.902.391.270.931.39
    3311.7417.737.113.884.831.931.991.470.791.04
    3612.8417.027.333.985.811.892.101.340.911.22
    3912.5418.993.403.556.422.082.251.060.811.34
    4210.7910.808.164.695.581.871.721.311.151.13
    4514.8315.414.805.197.852.042.571.051.061.28
    487.6015.678.924.234.951.442.421.820.921.16
    Table 3. Experimental result comparison of SPM concentration inversion using neural network (March 11, 2019)
    Number of bandsMSEMAE
    DTBSSVM-RFEPNN-L1PNN-L2PNN-ReLUDTBSSVM-RFEPNN-L1PNN-L2PNN-ReLU
    613.2430.333.926.625.991.913.481.251.401.47
    915.7029.178.285.225.272.173.471.691.381.31
    1211.0711.835.717.349.551.612.121.401.381.70
    1525.2114.883.188.107.163.021.831.001.581.44
    187.9420.713.268.757.201.782.841.011.401.76
    2120.9417.038.858.547.272.702.491.841.881.51
    2412.067.654.088.306.511.791.671.181.641.41
    2716.3514.835.726.294.532.292.321.451.420.97
    3018.4915.193.085.1913.172.462.461.121.222.27
    3318.7217.896.157.928.612.552.341.281.531.57
    363.7317.487.457.763.601.062.541.641.661.10
    3916.4812.577.235.926.592.391.811.451.411.28
    4215.6040.474.837.544.312.154.211.351.681.14
    4514.5320.304.026.905.912.362.71.071.501.53
    4820.0115.7810.005.916.802.562.521.771.431.28
    Table 4. Experimental result comparison of SPM concentration inversion using neural network model (August 5, 2019)
    Number of bandsMSEMAE
    DTBSSVM-RFEPNN-L1PNN-L2PNN-ReLUDTBSSVM-RFEPNN-L1PNN-L2PNN-ReLU
    611.3610.8912.8912.5810.612.402.492.302.362.32
    911.4512.0510.5412.2011.502.292.632.272.272.59
    1210.8311.3613.6312.6412.362.172.512.342.332.54
    1512.2011.2111.3411.7313.782.372.472.742.272.56
    1812.2011.9212.1912.1712.512.352.592.382.462.61
    2111.0611.3811.4611.7312.902.352.572.432.332.77
    2410.4511.5514.8710.4716.362.242.572.282.313.06
    2711.5411.5611.6011.5413.772.452.402.492.452.76
    3011.5910.4910.7616.7511.412.252.442.392.882.52
    3311.3410.8513.3612.7310.472.292.462.402.592.25
    3612.0911.7813.7311.7711.992.342.562.482.422.42
    3910.7510.7612.4612.2211.182.342.442.622.442.51
    4211.4012.7313.6413.0612.792.362.642.412.622.69
    4510.9811.2211.6211.3010.922.292.512.932.322.37
    4812.7711.3311.4111.0611.712.412.412.362.422.41
    Table 5. Experimental result comparison of SPM concentration inversion using random forest model (March 11, 2019)
    Number of bandsMSEMAE
    DTBSSVM-RFEPNN-L1PNN-L2PNN-ReLUDTBSSVM-RFEPNN-L1PNN-L2PNN-ReLU
    68.918.459.468.4210.112.402.212.372.182.45
    99.058.8910.118.158.372.292.322.402.192.29
    128.468.978.688.698.562.172.302.302.242.25
    158.208.388.858.697.992.372.292.182.192.30
    188.088.849.728.098.702.352.282.432.202.45
    218.638.808.588.427.342.352.362.192.192.19
    248.118.057.809.697.952.242.272.082.452.23
    278.038.439.228.338.012.452.202.292.142.23
    308.468.598.818.239.362.252.282.332.172.59
    338.548.9510.487.649.602.292.302.472.062.54
    369.368.188.158.787.732.342.242.172.302.14
    399.018.9412.379.036.952.342.262.692.292.04
    428.428.638.777.017.032.362.282.302.032.18
    458.158.4311.387.987.182.292.222.672.122.14
    487.808.188.367.487.192.412.202.282.002.09
    Table 6. Experimental result comparison of SPM concentration inversion using random forest model (August 5, 2019)
    Zhongkai Chen, Xiaorun Li, Liaoying Zhao. Inversion of Suspended Particulate Matter Concentration in Maozhou River Based on Band Selection of Hyperspectral Data[J]. Laser & Optoelectronics Progress, 2021, 58(20): 2001001
    Download Citation