• Laser & Optoelectronics Progress
  • Vol. 56, Issue 16, 161009 (2019)
Bin Yang1、2、* and Xiang Wang1、2
Author Affiliations
  • 1 School of Electrical Engineering, University of South China, Hengyang, Hunan 421001, China
  • 2 Hunan Provincial Key Laboratory for Ultra-Fast Micro/Nano Technology and Advanced Laser Manufacture, University of South China, Hengyang, Hunan 421001, China
  • show less
    DOI: 10.3788/LOP56.161009 Cite this Article Set citation alerts
    Bin Yang, Xiang Wang. Boosting Quality of Pansharpened Images Using Deep Residual Denoising Network[J]. Laser & Optoelectronics Progress, 2019, 56(16): 161009 Copy Citation Text show less
    Framework of proposed method
    Fig. 1. Framework of proposed method
    Structure of boosting network
    Fig. 2. Structure of boosting network
    Experimental images. (a) MS; (b) PAN
    Fig. 3. Experimental images. (a) MS; (b) PAN
    Results of different fusion methods. (a) ATWT; (b) BT; (c) GIHS; (d) GS; (e) SVT; (f) SWT
    Fig. 4. Results of different fusion methods. (a) ATWT; (b) BT; (c) GIHS; (d) GS; (e) SVT; (f) SWT
    Boosted results of different fusion methods. (a) ATWT; (b) BT; (c) GIHS; (d) GS; (e) SVT; (f) SWT
    Fig. 5. Boosted results of different fusion methods. (a) ATWT; (b) BT; (c) GIHS; (d) GS; (e) SVT; (f) SWT
    Influences of different network parameters on experimental results. (a) ERGAS; (b) SAM; (c) Q4; (d) CC
    Fig. 6. Influences of different network parameters on experimental results. (a) ERGAS; (b) SAM; (c) Q4; (d) CC
    Reference image and results of compared methods and proposed method. (a) Reference image; (b) CS; (c) DRPNN; (d) PNN; (e) proposed method
    Fig. 7. Reference image and results of compared methods and proposed method. (a) Reference image; (b) CS; (c) DRPNN; (d) PNN; (e) proposed method
    Reference image and results of compared methods and proposed method. (a) Reference image; (b) CS; (c) DRPNN; (d) PNN; (e) proposed method
    Fig. 8. Reference image and results of compared methods and proposed method. (a) Reference image; (b) CS; (c) DRPNN; (d) PNN; (e) proposed method
    Different reference images and results of proposed method. (a) Reference images; (b) results of proposed method
    Fig. 9. Different reference images and results of proposed method. (a) Reference images; (b) results of proposed method
    MethodATWTBTGIHSGSSWTSVT
    Fusion0.03270.05420.03230.16531.11150.1728
    Boosting1.01470.94990.95090.94790.95400.9546
    Total1.04741.00420.98321.11322.06551.1274
    Table 1. Running time of different methods and corresponding boosting stages
    IndexATWTBTGIHS
    Fused resultBoosted resultFused resultBoosted resultFused resultBoosted result
    ERGAS5.54133.64647.79433.57885.57663.8262
    SAM8.52815.78926.03265.48216.55395.7464
    Q40.70850.84440.76000.86160.74790.8590
    CC0.84870.93580.78220.93750.84810.9297
    IndexGSSVTSWT
    Fused resultBoosted resultFused resultBoosted resultFused resultBoosted result
    ERGAS4.75653.27724.06252.83334.56733.2977
    SAM5.85875.29916.10184.56296.96495.2726
    Q40.79000.88750.80350.90440.78370.8727
    CC0.91170.94900.92230.96140.89880.9478
    Table 2. Evaluation of fusion results obtained by different methods and corresponding boosted results
    FigureIndexCSDRPNNPNNProposed methodIdeal
    Fig. 7ERGAS3.13823.20302.44092.39340
    SAM4.56103.78843.75993.66270
    Q40.76310.74470.80320.80051
    CC0.95100.95320.97050.96981
    Fig. 8ERGAS2.55212.71642.25302.24120
    SAM3.47943.10213.35273.34130
    Q40.62900.62290.64630.63381
    CC0.96120.95870.97250.96761
    Fig. 9(The 1st column)ERGAS3.73823.61313.00332.55990
    SAM5.83135.00335.02544.28700
    Q40.86930.85650.90950.91691
    CC0.93530.93710.95680.96841
    Fig. 9(The 2nd column)ERGAS3.96564.24783.13953.01940
    SAM6.19535.25925.26174.92110
    Q40.89170.85620.93330.93031
    CC0.92850.91490.95380.95721
    Fig. 9(The 3rd column)ERGAS3.79393.60332.89362.11360
    SAM5.86064.67644.68063.66950
    Q40.86560.84720.91420.93771
    CC0.93290.93500.95850.97771
    Table 3. Evaluation of compared methods and proposed method
    Bin Yang, Xiang Wang. Boosting Quality of Pansharpened Images Using Deep Residual Denoising Network[J]. Laser & Optoelectronics Progress, 2019, 56(16): 161009
    Download Citation