• Laser & Optoelectronics Progress
  • Vol. 60, Issue 24, 2401001 (2023)
Xiuzai Zhang1、2, Jingxuan Li2, Changjun Yang3、4、*, and Xuan Feng5
Author Affiliations
  • 1Jiangsu Province Atmospheric Environment and Equipment Technology Collaborative Innovation Center, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China
  • 2School of Electronic and Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China
  • 3Key Laboratory of Radiometric Calibration and Validation for Environmental Satellites, National Satellite;Meteorological Center (National Center for Space Weather), China Meteorological Administration, Beijing 100081, China
  • 4Innovation Center for FengYun Meteorological Satellite (FYSIC), Beijing 100081, China
  • 5Technology and Engineering Center for Space Utilization, Chinese Academy of Sciences, Beijing 100094, China
  • show less
    DOI: 10.3788/LOP231059 Cite this Article Set citation alerts
    Xiuzai Zhang, Jingxuan Li, Changjun Yang, Xuan Feng. Feature-Enhanced Cloud Image Prediction Algorithm Based on Spatio-Temporal Attention Gated Recurrent Unit[J]. Laser & Optoelectronics Progress, 2023, 60(24): 2401001 Copy Citation Text show less
    Basic framework of GRU
    Fig. 1. Basic framework of GRU
    Overall framework of CrevNet
    Fig. 2. Overall framework of CrevNet
    Overall framework of SmartCrevNet
    Fig. 3. Overall framework of SmartCrevNet
    Two-way SGE autoencoder
    Fig. 4. Two-way SGE autoencoder
    Lightweight attention model SGE
    Fig. 5. Lightweight attention model SGE
    STA-GRU structure
    Fig. 6. STA-GRU structure
    Reversible STA-GRU module
    Fig. 7. Reversible STA-GRU module
    Two prediction examples from satellite cloud images
    Fig. 8. Two prediction examples from satellite cloud images
    Frame-wise MSE and SSIM comparison of the different models on the satellite cloud map dataset
    Fig. 9. Frame-wise MSE and SSIM comparison of the different models on the satellite cloud map dataset
    Two sets of prediction examples
    Fig. 10. Two sets of prediction examples
    Method(4→4)MSE /10-3MAE /10-3SSIM↑PSNR↑
    ConvLSTM27.41104.190.44815.62
    ConvGRU26.8199.070.44415.72
    PredRNN25.3097.110.45715.97
    PredRNN++25.1195.060.45116.00
    CrevNet19.6472.410.48217.07
    SmartCrevNet18.2167.340.52017.40
    Table 1. Comparative analysis of various prediction algorithms (4 frames→4 frames)
    Model(4→4)MSE /10-3)↓MAE /10-3)↓SSIM↑PSNR↑
    CrevNet19.6472.410.48217.07
    SmartCrevNet w/o SGE19.4769.360.51817.11
    SmartCrevNet w/o HIM19.6469.160.51817.06
    SmartCrevNet+SE+HIM19.1368.080.51117.18
    SmartCrevNet+CBAM+HIM19.3469.680.50417.14
    SmartCrevNet+SGE+HIM18.2167.340.52017.40
    Table 2. Two-way autoencoder and prediction module ablation experiments
    ModelMSE /10-3SSIM↑
    ConvLSTM11.5620.882
    PreRNN9.4960.905
    PredRNN++8.3440.917
    CrevNet5.6030.925
    SmartCrevNet5.2610.936
    Table 3. Quantitative evaluation of different methods on Moving MNIST
    Xiuzai Zhang, Jingxuan Li, Changjun Yang, Xuan Feng. Feature-Enhanced Cloud Image Prediction Algorithm Based on Spatio-Temporal Attention Gated Recurrent Unit[J]. Laser & Optoelectronics Progress, 2023, 60(24): 2401001
    Download Citation