Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161012(2020)

Mural Image Super Resolution Reconstruction Based on Multi-Scale Residual Attention Network

Zhigang Xu, Juanjuan Yan, and Honglei Zhu*
Author Affiliations
  • School of Computer and Communication, Lanzhou University of Technology, Lanzhou, Gansu 730050, China
  • show less
    Figures & Tables(9)
    Structure of residual block
    Schematic diagram of channel attention mechanism
    Structure of our network
    Structure of residual channel attention block
    Reconstruction results of different algorithms. (a) Original images; (b) bicubic interpolation algorithm; (c) SRCNN algorithm; (d) ESPCN algorithm; (e) VDSR algorithm; (f) our algorithm
    PSNR curves of reconstructed mural image. (a) Buddha mural; (b) caisson mural
    Parameter amounts and training time of different reconstruction algorithms. (a) Parameter amount; (b) training time
    • Table 1. Reconstructed image quality at different network depths

      View table

      Table 1. Reconstructed image quality at different network depths

      Test setIndexVDSRRCAB
      5710
      Buddha muralPSNR /dB33.4330.9232.5534.33
      SSIM0.94780.93130.94240.9656
      Flying apsaras muralPSNR /dB30.1228.7230.4430.89
      SSIM0.95840.94560.96410.9884
      Caisson muralPSNR /dB31.3528.1930.6432.23
      SSIM0.96040.93510.94780.9712
    • Table 2. Test results of mural images by different algorithms

      View table

      Table 2. Test results of mural images by different algorithms

      DatesetUpscalingfactorBicubicSRCNNESCPNVDSROurs
      PSNR /dBSSIMPSNR /dBSSIMPSNR /dBSSIMPSNR /dBSSIMPSNR /dBSSIM
      231.570.930131.960.945132.210.948333.870.951734.660.9634
      Test(a)330.370.861231.650.874131.780.879533.560.883233.920.9187
      427.420.830128.110.841428.690.845229.350.867730.240.8792
      230.640.942132.150.945132.870.947233.710.952733.880.9563
      Test(b)328.350.823229.740.830429.870.834530.560.865731.420.8779
      426.130.801426.890.811627.140.813528.430.835429.190.8426
      229.330.924229.700.925529.940.928732.130.941232.990.9488
      Test(c)328.270.872330.340.874730.950.878731.330.885231.84/0.8943
      424.590.837825.230.842325.880.846226.790.852327.36/0.8589
    Tools

    Get Citation

    Copy Citation Text

    Zhigang Xu, Juanjuan Yan, Honglei Zhu. Mural Image Super Resolution Reconstruction Based on Multi-Scale Residual Attention Network[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161012

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Image Processing

    Received: Dec. 10, 2019

    Accepted: Jan. 14, 2020

    Published Online: Aug. 5, 2020

    The Author Email: Honglei Zhu (xzg_cn@163.com)

    DOI:10.3788/LOP57.161012

    Topics