• Acta Optica Sinica
  • Vol. 40, Issue 15, 1528003 (2020)
Ruifei Zhu1、2, Jingyu Ma1, Zhuqiang Li1、*, Dong Wang1、2, Yuan An1、2, Xing Zhong1、2, Fang Gao1, and Xiangyu Meng3
Author Affiliations
  • 1Jilin Key Laboratory of Satellite Remote Sensing Application Technology, Chang Guang Satellite Technology Co., Ltd., Changchun, Jilin 130012, China
  • 2Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun, Jilin 130033, China
  • 3Jilin Institute of Land Survey & Planning, Changchun, Jilin 130061, China;
  • show less
    DOI: 10.3788/AOS202040.1528003 Cite this Article Set citation alerts
    Ruifei Zhu, Jingyu Ma, Zhuqiang Li, Dong Wang, Yuan An, Xing Zhong, Fang Gao, Xiangyu Meng. Domestic Multispectral Image Classification Based on Multilayer Perception Convolutional Neural Network[J]. Acta Optica Sinica, 2020, 40(15): 1528003 Copy Citation Text show less
    Flow chart of MPCNet classification algorithm
    Fig. 1. Flow chart of MPCNet classification algorithm
    Multispectral remote sensing image classification model based on MPCNet
    Fig. 2. Multispectral remote sensing image classification model based on MPCNet
    Introduction of experimental data. (a) True color image of Jilin-1GP01; (b) true color image of Landsat8; (c) true color image of Sentinel-2A; (d) true color image of HJ-1A; (e) high resolution image of Google Earth; (f) land cover product of FROM-GLC-2017
    Fig. 3. Introduction of experimental data. (a) True color image of Jilin-1GP01; (b) true color image of Landsat8; (c) true color image of Sentinel-2A; (d) true color image of HJ-1A; (e) high resolution image of Google Earth; (f) land cover product of FROM-GLC-2017
    Reflectivity curves of samples in Nashik research area of Jilin-1GP01. (a) Uncultivated land; (b) cultivated land; (c) building; (d) grassland; (e) bare land; (f) road; (g) forest land; (h) water; (i) cloud; (j) shadow
    Fig. 4. Reflectivity curves of samples in Nashik research area of Jilin-1GP01. (a) Uncultivated land; (b) cultivated land; (c) building; (d) grassland; (e) bare land; (f) road; (g) forest land; (h) water; (i) cloud; (j) shadow
    Classification results using MPCNet algorithm. (a) Jilin-1GP01 image; (b) Landsat8 image; (c) Sentinel-2A image; (d) HJ-1A image
    Fig. 5. Classification results using MPCNet algorithm. (a) Jilin-1GP01 image; (b) Landsat8 image; (c) Sentinel-2A image; (d) HJ-1A image
    Classification detail results of surface features at local Nashik area. (a) Pseudo-color composite image; (b) local enlargement image; (c) classification result by Jilin-1GP01; (d) classification result by Sentinel-2A; (e) remote sensing image superimposed vector data
    Fig. 6. Classification detail results of surface features at local Nashik area. (a) Pseudo-color composite image; (b) local enlargement image; (c) classification result by Jilin-1GP01; (d) classification result by Sentinel-2A; (e) remote sensing image superimposed vector data
    Classification results of different algorithms on Jilin-1GP01 images in Nashik. (a) Local image; (b) SVM; (c) LGBM-GBDT; (d) shallow CNN; (e) MPCNet
    Fig. 7. Classification results of different algorithms on Jilin-1GP01 images in Nashik. (a) Local image; (b) SVM; (c) LGBM-GBDT; (d) shallow CNN; (e) MPCNet
    Classification results of different algorithms on Jilin-1GP02 images in Xintai city. (a) Local image; (b) SVM; (c) LGBM-GBDT; (d) shallow CNN; (e) MPCNet
    Fig. 8. Classification results of different algorithms on Jilin-1GP02 images in Xintai city. (a) Local image; (b) SVM; (c) LGBM-GBDT; (d) shallow CNN; (e) MPCNet
    Satellite typeAvailable/selective band numberWavelength range /nmSpatial resolution /mShooting time
    Jilin-1GP0126/10400-1350052019-01-22
    Landsat811/11430-12510302019-01-13
    Sentinel-2A13/13443-2190102019-01-20
    HJ-1A4/4430-900302019-01-14
    Table 1. Band selection, spatial resolution, and shooting time of multispectral satellite
    Satellite typeSensorScene IDAvailable/selective band numberWavelength range /nmSpatial resolution /mShooting time
    Jilin-1GP02PMS1PMS1PMS2PMS2000100020001000226/10400-1350052019-03-15
    Table 2. Band selection, spatial resolution, and shooting time of Jilin-1GP02
    Area typeJilin-1GP01Landsat8Sentinel-2AHJ-1A
    P/RF1P/RF1P/RF1P/RF1
    Uncultivated land1.00/0.930.960.91/0.830.870.96/0.890.920.76/0.740.75
    Cultivated land0.94/0.890.910.81/0.880.840.83/0.980.900.52/0.700.60
    Building0.96/0.990.970.94/0.790.860.97/0.940.960.85/0.610.71
    Grassland1.00/0.880.941.00/0.730.851.00/0.820.900.92/0.690.79
    Bare land0.96/1.000.980.96/0.930.940.94/0.990.960.75/0.860.80
    Road0.95/0.800.870.43/0.800.560.85/0.830.840.14/0.230.18
    Forest land0.71/1.000.830.62/1.000.760.69/1.000.820.27/0.680.38
    Water1.00/1.001.001.00/0.970.981.00/0.930.960.91/0.600.72
    K0.9480.8280.9200.595
    POA0.9580.8590.9350.667
    Table 3. Classification accuracy evaluation index of different satellite data based on MPCNet algorithm
    Area typeSVMLGBM-GBDTShallow CNNMPCNet
    P/RF1P/RF1P/RF1P/RF1
    Uncultivated land0.91/0.700.780.90/0.830.860.91/0.850.881.00/0.930.96
    Cultivated land0.90/0.890.890.91/0.900.910.98/0.900.940.94/0.890.91
    Building0.86/1.000.920.93/0.990.960.91/1.000.950.96/0.990.97
    Grassland0.99/0.790.881.00/0.690.821.00/0.890.941.00/0.880.94
    Bare land0.91/0.990.950.90/0.990.940.92/1.000.960.96/1.000.98
    Road0.64/0.280.390.88/0.620.730.76/0.400.530.95/0.800.87
    Forest land0.60/0.850.710.65/1.000.790.67/1.000.800.71/1.000.83
    Water1.00/0.960.981.00/0.970.991.00/0.960.981.00/1.001.00
    K0.8570.8990.9010.948
    POA0.8860.9190.9210.958
    Table 4. Classification accuracy evaluation index of different algorithms on Jilin-1GP01 images
    Area typeSVMLGBM-GBDTShallow CNNMPCNet
    P/RF1P/RF1P/RF1P/RF1
    Uncultivated land0.56/0.660.610.99/0.920.950.96/0.990.980.99/0.980.99
    Cultivated land0.97/0.980.970.71/0.670.690.88/0.850.860.96/0.940.95
    Forest land0.95/0.870.900.85/0.850.850.97/0.970.970.97/0.980.97
    Shrub0.56/0.850.670.44/0.870.590.75/0.980.850.88/0.970.92
    Water1.00/0.780.880.75/0.680.710.92/0.780.840.95/0.930.94
    Building0.62/0.690.650.80/0.760.780.83/0.970.900.92/0.940.93
    Bare land0.57/0.490.530.70/0.700.700.85/0.780.810.83/0.890.86
    K0.7310.7240.8770.932
    POA0.7630.7560.8910.940
    Table 5. Classification accuracy evaluation index of different algorithms on Jilin-1GP02 images
    AlgorithmImage sizeImage storage /MbitFeature extraction time /sModel training time /minInference time /minTotal process time /min
    SVMLGBM-GBDTShallow CNNMPCNet5368×4565888.04713.424.338.037.547.451.3414.9210.8063.427.422.963.56122.769.1718.5114.98
    Table 6. Processing efficiency of different algorithms on Jilin-1GP01 images
    Ruifei Zhu, Jingyu Ma, Zhuqiang Li, Dong Wang, Yuan An, Xing Zhong, Fang Gao, Xiangyu Meng. Domestic Multispectral Image Classification Based on Multilayer Perception Convolutional Neural Network[J]. Acta Optica Sinica, 2020, 40(15): 1528003
    Download Citation