Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 61, Issue 10, 1037002(2024)

Image Super-Resolution Reconstruction Algorithm Based on Adaptive Two-Branch Block

Yan Zhang, Minglei Sun, Yemei Sun*, and Fujie Xu
Author Affiliations
  • School of Computer and Information Engineering, Tianjin Chengjian University, Tianjin 300384, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (14)
    Equations (16)
    References (31)
    Cited By (1)
    Tools
    Paper Information
    Topics
    Figures & Tables(14)
    Network of image super-resolution reconstruction based on adaptive two-branch

    Fig. 1. Network of image super-resolution reconstruction based on adaptive two-branch

    Download full size
    Adaptive two-branch block

    Fig. 2. Adaptive two-branch block

    Download full size
    Decomposition process of the large kernel convolution

    Fig. 3. Decomposition process of the large kernel convolution

    Download full size
    Channel shuffle coordinate attention block

    Fig. 4. Channel shuffle coordinate attention block

    Download full size
    Double-layer residual aggregation block

    Fig. 5. Double-layer residual aggregation block

    Download full size
    Effect comparison of output feature maps

    Fig. 6. Effect comparison of output feature maps

    Download full size
    Reconstruction effect comparison of Urban100_012 at ×2 magnification

    Fig. 7. Reconstruction effect comparison of Urban100_012 at ×2 magnification

    Download full size
    Reconstruction effect comparison of Urban100_092 at ×3 magnification

    Fig. 8. Reconstruction effect comparison of Urban100_092 at ×3 magnification

    Download full size
    Reconstruction effect comparison of Urban100_033 at ×4 magnification

    Fig. 9. Reconstruction effect comparison of Urban100_033 at ×4 magnification

    Download full size

    • Table 1. Results of ablation study

      View table

      Table 1. Results of ablation study

      MethodATBCSCABDRABPSNR /dB（SSIM）
      Model 1×××30.87（0.9150）
      Model 2√××32.08（0.9278）
      Model 3×√×32.03（0.9274）
      Model 4××√31.98（0.9269）
      Model 5√√×32.28（0.9299）
      Model 6（ATSR）√√√32.44（0.9308）

    • Table 2. Performance of the proposed algorithm before and after large kernel convolutional decomposition

      View table

      Table 2. Performance of the proposed algorithm before and after large kernel convolutional decomposition

      MethodParams /103PSNR /dB（SSIM）
      Set14Urban100
      Before decomposition1108933.75（0.9191）32.34（0.9302）
      After decomposition130833.82（0.9193）32.44（0.9308）

    • Table 3. Influence of number of RB on the performance of the proposed algorithm

      View table

      Table 3. Influence of number of RB on the performance of the proposed algorithm

      Res module No.Params /103PSNR /dB（SSIM）
      Set14Urban100
      296733.63（0.9179）32.16（0.9286）
      3112133.66（0.9181）32.17（0.9287）
      4130833.82（0.9193）32.44（0.9308）
      5161233.80（0.9188）32.48（0.9312）

    • Table 4. Peformance of different algorithms on each dataset

      View table

      Table 4. Peformance of different algorithms on each dataset

      MethodScalePSNR /dB（SSIM）
      Set5Set14Urban100BSD100Manga109
      VDSR［6］×237.53（0.9590）33.05（0.9130）30.77（0.9140）31.90（0.8960）37.22（0.9750）
      CARN［7］37.76（0.9590）33.52（0.9166）31.92（0.9256）32.09（0.8978）38.36（0.9765）
      IMDN［25］38.00（0.9605）33.63（0.9177）32.17（0.9283）32.19（0.8996）38.88（0.9774）
      SRMDNF［26］37.79（0.9601）33.32（0.9159）31.33（0.9204）32.05（0.8985）38.07（0.9761）
      SMSR［27］38.00（0.9601）33.64（0.9179）32.19（0.9284）32.17（0.8990）38.76（0.9771）
      PAN［13］38.00（0.9605）33.59（0.9181）32.01（0.9273）32.18（0.8997）38.70（0.9773）
      ARRFN［28］38.01（0.9606）33.66（0.9179）32.27（0.9295）32.20（0.8999）
      PDAN［29］38.05（0.9607）33.65（0.9182）32.36（0.9300）32.20（0.8998）38.71（0.9771）
      A2N［14］38.06（0.9608）33.75（0.9194）32.43（0.9311）32.22（0.9002）38.87（0.9769）
      DRSDN［30］38.06（0.9607）33.65（0.9189）32.40（0.9308）32.23（0.9003）
      LESR［31］38.07（0.9606）33.80（0.9194）32.42（0.9308）32.24（0.9000）38.96（0.9776）
      ATSR38.11（0.9608）33.82（0.9193）32.44（0.9308）32.24（0.8996）38.94（0.9775）
      VDSR［6］×333.67（0.9210）29.78（0.832027.14（0.8290）28.83（0.7990）32.01（0.9340）
      CARN［7］34.29（0.9255）30.29（0.8407）28.06（0.8493）29.06（0.8034）33.50（0.9440）
      IMDN［25］34.36（0.9270）30.32（0.8417）28.17（0.8519）29.09（0.8046）33.61（0.9445）
      SRMDNF［26］34.12（0.9254）30.04（0.8382）27.57（0.8398）28.97（0.8025）33.00（0.9403）
      SMSR［27］34.40（0.9270）30.33（0.8412）28.25（0.8536）29.10（0.8050）33.68（0.9445）
      PAN［13］34.40（0.9271）30.36（0.8423）28.11（0.8511）29.11（0.8050）33.61（0.9448）
      ARRFN［28］34.38（0.9272）30.36（0.8422）28.22（0.8533）29.09（0.8050）
      PDAN［29］34.44（0.9276）30.39（0.8437）28.34（0.8563）29.11（0.8063）33.63（0.9448）
      A2N［14］34.47（0.9279）30.44（0.8437）28.41（0.8570）29.14（0.8059）33.78（0.9458）
      DRSDN［30］34.48（0.9282）30.41（0.8445）28.45（0.8589）29.17（0.8072）
      LESR［31］34.49（0.9278）30.42（0.8431）28.39（0.8567）29.13（0.8059）33.76（0.9455）
      ATSR34.53（0.9283）30.50（0.8441）28.54（0.8593）29.18（0.8060）33.93（0.9463）
      VDSR［6］×431.35（0.8830）28.02（0.7680）25.18（0.7540）27.29（0.7260）28.83（0.8870）
      CARN［7］32.13（0.8937）28.60（0.7806）26.07（0.7837）27.58（0.7349）30.47（0.9084）
      IMDN［25］32.21（0.8948）28.58（0.7811）26.04（0.7838）27.56（0.7353）30.45（0.9075）
      SRMDNF［26］31.96（0.8925）28.35（0.7787）25.68（0.7731）27.49（0.7337）30.09（0.9024）
      SMSR［27］32.12（0.8932）28.55（0.7808）26.11（0.7868）27.55（0.7351）30.54（0.9085）
      PAN［13］32.13（0.8948）28.61（0.7822）26.11（0.7854）27.59（0.7363）30.51（0.9095）
      ARRFN［28］32.22（0.8952）28.60（0.7817）26.09（0.7858）27.57（0.7355）
      PDAN［29］32.28（0.8957）28.66（0.7831）26.27（0.7922）27.62（0.7378）30.64（0.9098）
      A2N［14］32.30（0.8966）28.71（0.7842）26.27（0.7920）27.61（0.7374）30.67（0.9110）
      DRSDN［30］32.28（0.8962）28.64（0.7836）26.30（0.7933）27.64（0.7388）
      LESR［31］32.28（0.8952）28.65（0.7827）26.25（0.7908）27.61（0.7365）30.68（0.9100）
      ATSR32.34（0.8973）28.75（0.7832）26.37（0.7934）27.66（0.7370）30.69（0.9089）

    • Table 5. Comparison of different parameters of different algorithms

      View table

      Table 5. Comparison of different parameters of different algorithms

      MethodParams /103PSNR /dBSSIM
      SRMDNF［26］155231.960.8925
      CARN［7］159232.130.8937
      ARRFN［28］100832.220.8952
      PDAN［29］158732.280.8957
      DRSDN［30］109532.280.8962
      LESR［31］102032.280.8952
      ATSR140332.340.8973
    Tools

    Get Citation

    Copy Citation Text

    Yan Zhang, Minglei Sun, Yemei Sun, Fujie Xu. Image Super-Resolution Reconstruction Algorithm Based on Adaptive Two-Branch Block[J]. Laser & Optoelectronics Progress, 2024, 61(10): 1037002

    Download Citation

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

    Category: Digital Image Processing

    Received: Aug. 29, 2023

    Accepted: Oct. 30, 2023

    Published Online: May. 6, 2024

    The Author Email: Yemei Sun (sunyemei1216@163.com)

    DOI:10.3788/LOP232007

    CSTR:32186.14.LOP232007

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology