Laser & Optoelectronics Progress, Volume. 62, Issue 6, 0637011(2025)

Image Super-Resolution Reconstruction with Spatial/High-Frequency Dual-Domain Feature Saliency

Yue Hou*, Ziwei Hao, Zhihao Zhang, and Jie Yin
Author Affiliations
  • School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730000, Gansu , China
  • show less
    Figures & Tables(10)
    Overall network framework of STDHF
    FDR structure
    Structure of the SBDF module
    Comparison with the visualization results of other algorithms at ×2 magnification
    Comparison with the visualization results of other algorithms at ×4 magnification
    Comparison of feature concerns between STDHF model and BSRN model
    • Table 1. Performance of different algorithms on each dataset

      View table

      Table 1. Performance of different algorithms on each dataset

      MethodScalePSNR /dB(SSIM)
      Set5Set14B100Urban100
      SRCNN×236.66(0.9542)32.43(0.9067)31.36(0.8876)29.48(0.8936)
      IMDN38.00(0.9605)33.63(0.9177)32.19(0.8996)32.17(0.8283)
      CARN37.76(0.9590)33.52(0.9166)32.09(0.8978)31.92(0.9256)
      MRFN37.98(0.9611)33.41(0.9159)32.14(0.8997)31.45(0.9221)
      EDSR37.99(0.9604)33.57(0.9175)32.16(0.8994)31.98(0.9272)
      BSRN38.10(0.9610)33.74(0.9193)32.24(0.9006)32.34(0.9303)
      SMSR37.80(0.9591)33.56(0.9175)32.12(0.8994)32.21(0.9292)
      VapSR38.08(0.9612)33.77(0.9195)32.27(0.9011)32.45(0.9316)
      SwinIR-light38.14(0.9611)33.86(0.9206)32.31(0.9012)32.76(0.9340)
      STDHF38.16(0.9612)33.87(0.9210)32.33(0.9013)32.78(0.9340)
      SRCNN×332.75(0.9090)29.28(0.8209)28.41(0.7863)26.24(0.7989)
      IMDN34.36(0.9270)30.32(0.8417)29.09(0.8046)28.17(0.8519)
      CARN34.29(0.9255)30.29(0.8407)29.06(0.8034)28.06(0.8493)
      MRFN34.21(0.9267)30.03(0.8363)28.99(0.8029)27.53(0.8589)
      EDSR34.37(0.9270)30.28(0.8417)29.09(0.8052)28.15(0.8527)
      BSRN34.46(0.9277)30.47(0.8449)29.18(0.8068)28.39(0.8567)
      SMSR34.40(0.9270)30.33(0.8412)29.10(0.8050)28.25(0.8536)
      VapSR34.52(0.9284)30.53(0.8452)29.19(0.8077)28.43(0.8583)
      SwinIR-light34.62(0.9289)30.54(0.8463)29.20(0.8082)28.66(0.8624)
      STDHF34.62(0.9290)30.55(0.8466)29.24(0.8088)28.66(0.8620)
      SRCNN×430.48(0.8628)27.49(0.7503)26.90(0.7101)24.52(0.7221)
      IMDN32.21(0.8948)28.58(0.7811)27.56(0.7353)26.04(0.7838)
      CARN32.13(0.8937)28.60(0.7806)27.58(0.7349)26.07(0.7837)
      MRFN31.90(0.8916)28.31(0.7746)27.43(0.7309)25.46(0.7654)
      EDSR32.09(0.8938)28.58(0.7813)27.57(0.7357)26.04(0.7849)
      BSRN32.35(0.8966)28.73(0.7848)27.65(0.7387)26.27(0.7908)
      SMSR32.12(0.8932)28.55(0.7808)27.55(0.7351)26.11(0.7868)
      VapSR32.38(0.8978)28.77/0.7852)27.68(0.7398)26.35(0.7941)
      SwinIR-light32.44(0.8976)28.77(0.7858)27.69(0.7406)26.47(0.7980)
      STDHF32.45(0.8976)28.78(0.7858)27.71(0.7410)26.48(0.7981)
    • Table 2. Evaluation metrics of various lightweight SISR models at scale factor of ×2

      View table

      Table 2. Evaluation metrics of various lightweight SISR models at scale factor of ×2

      MethodParamsPSNR /dB (SSIM)
      Set5Set14B100Urban100
      Swin-light0.93×10638.14(0.9611)33.86(0.9206)32.31(0.9012)32.76(0.9340)
      ESRT677×10338.03(0.9600)33.75(0.9184)32.25(0.9005)32.58(0.9318)
      NGswin0.99×10638.05(0.9610)33.63(0.9177)32.19(0.8996)32.17(0.8283)
      CARN1592×10337.76(0.9590)33.52(0.9166)32.09(0.8978)31.92(0.9256)
      EDSR13.7×10637.99(0.9604)33.57(0.9175)32.16(0.8994)31.98(0.9272)
      BSRN0.32×10638.10(0.9610)33.74(0.9193)32.24(0.9006)32.34(0.9303)
      SMSR985×10337.80(0.9591)33.56(0.9175)32.12(0.8994)32.21(0.9292)
      VapSR431×10338.08(0.9612)33.77(0.9195)32.27(0.9011)32.45(0.9316)
      SwinIR11.5×10638.35(0.9620)34.14(0.9206)32.44(0.9012)33.40(0.9340)
      LatticeNet756×10338.15(0.9610)33.78(0.9193)32.25(0.9005)32.43(0.9302)
      STDHF0.70×10638.16(0.9610)33.81(0.9120)32.33(0.9013)32.78(0.9340)
    • Table 3. Influence of BSConv on the performance of the model in STDHF

      View table

      Table 3. Influence of BSConv on the performance of the model in STDHF

      MethodParams /103Flops /109PSNR /dB(SSIM)
      Set5Set14B100Urban100
      Basic43323.932.04(0.9611)28.52(0.7799)27.53(0.7344)25.92(0.7810)
      Basic-B1236.931.95(0.8910)28.40(0.9184)27.45(0.7318)25.64(0.7726)
    • Table 4. Influence of different components in STDHF on model performance

      View table

      Table 4. Influence of different components in STDHF on model performance

      MethodPSNR /dB(SSIM)
      Set5Set14B100Urban100
      Basic38.10(0.9610)33.74(0.9193)32.24(0.9006)32.34(0.9303)
      Basic-M38.15(0.9611)33.79(0.9197)32.27(0.9011)32.71(0.9337)
      Basic-S38.16(0.9612)33.81(0.9120)32.30(0.9013)32.74(0.9340)
    Tools

    Get Citation

    Copy Citation Text

    Yue Hou, Ziwei Hao, Zhihao Zhang, Jie Yin. Image Super-Resolution Reconstruction with Spatial/High-Frequency Dual-Domain Feature Saliency[J]. Laser & Optoelectronics Progress, 2025, 62(6): 0637011

    Download Citation

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

    Category: Digital Image Processing

    Received: Jul. 1, 2024

    Accepted: Aug. 29, 2024

    Published Online: Mar. 10, 2025

    The Author Email:

    DOI:10.3788/LOP241598

    CSTR:32186.14.LOP241598

    Topics