Acta Photonica Sinica, Volume. 53, Issue 9, 0910003(2024)

Infrared and Visible Image Fusion Method Based on Information Enhancement and Mask Loss

Xiaodong ZHANG... Shuo WANG*, Shaoshu GAO, Xinrui WANG and Long ZHANG |Show fewer author(s)
Author Affiliations
  • College of Computer Science and Technology, China University of Petroleum ( East China ),Qingdao 226580, China
  • show less
    Figures & Tables(13)
    Infrared mask effect diagram
    The network architecture of the proposed fusion method
    Comparison diagram before and after source image enhancement
    Interactive enhancement module
    Comparison diagram before and after feature enhancement
    Attention guidance module
    Qualitative comparison results
    Qualitative comparison results of ablation experiments
    • Table 1. Quantitative comparison results of different algorithms on MSRS dataset

      View table
      View in Article

      Table 1. Quantitative comparison results of different algorithms on MSRS dataset

      MethodsENSFAGSDVIF
      DenseFuse5.7254.9971.63520.6710.703
      FusionGAN5.1683.7261.21914.8410.49
      GANMcC6.055.2081.7925.7650.686
      RFN-Nest6.0515.6431.8727.0110.707
      SDNet4.8477.0232.05314.20.486
      PIAFusion6.3649.5492.9738.0521.002
      SeaFusion6.4739.2942.91937.6561.016
      SwinFusion6.4579.3152.85538.5431.03
      IRFS6.3698.4682.58532.3270.8
      Ours6.76411.4093.5444.5541.043
    • Table 2. Quantitative comparison results of different algorithms on TNO dataset

      View table
      View in Article

      Table 2. Quantitative comparison results of different algorithms on TNO dataset

      MethodsENSFAGSDVIF
      DenseFuse6.8198.9853.5634.8250.658
      FusionGAN6.5586.2752.42130.6630.422
      GANMcC6.7366.1612.54433.4370.53
      RFN-Nest6.9635.8742.66936.8970.559
      SDNet6.69511.6434.61233.6690.578
      PIAFusion6.8149.623.82837.1410.74
      SeaFusion7.13312.2534.9844.2440.704
      SwinFusion6.89110.7234.21139.4470.75
      IRFS6.6148.7053.16731.1270.59
      Ours7.18915.5726.32146.9870.68
    • Table 3. Quantitative comparison results of different algorithms on LLVIP dataset

      View table
      View in Article

      Table 3. Quantitative comparison results of different algorithms on LLVIP dataset

      MethodsENSFAGSDVIF
      DenseFuse6.5976.1291.8428.190.662
      FusionGAN6.2885.9381.61724.3410.486
      GANMcC6.6655.7281.75331.5560.625
      RFN-Nest6.8335.2691.77533.7190.689
      SDNet6.6419.3362.5931.270.643
      PIAFusion7.14410.9093.10243.4490.976
      SeaFusion7.1910.2272.87743.6620.917
      SwinFusion7.12410.3552.9143.520.938
      IRFS6.97510.1012.60338.3340.834
      Ours7.2712.5283.50453.4330.998
    • Table 4. Comparison results of running time of different algorithms (units: s)

      View table
      View in Article

      Table 4. Comparison results of running time of different algorithms (units: s)

      MethodsMSRSTNOLLVIP
      DenseFuse0.283±0.1030.607±0.1280.782±0.158
      FusionGAN0.168±0.0190.385±0.0510.462±0.109
      GANMcC0.131±0.0270.671±0.1290.711±0.132
      RFN-Nest0.258±0.0180.214±0.0420.552±0.127
      SDNet0.038±0.0060.049±0.0180.139±0.013
      PIAFusion0.013±0.0020.028±0.0120.015±0.004
      SeaFusion0.012±0.0010.019±0.0060.012±0.003
      SwinFusion2.258±0.5783.324±0.8782.782±0.702
      IRFS0.138±0.0060.552±0.0510.236±0.012
      Ours0.057±0.0050.073±0.0130.187±0.009
    • Table 5. Quantitative comparison results of ablation experiments on MSRS dataset 56 pairs of images

      View table
      View in Article

      Table 5. Quantitative comparison results of ablation experiments on MSRS dataset 56 pairs of images

      StageABCDENSFAGSDVIF
      16.4679.3532.87338.3470.986
      26.72111.1393.46144.0171.041
      36.61810.3973.21741.4211.089
      46.75411.3153.36843.8611.01
      56.76411.4093.5444.5541.043
    Tools

    Get Citation

    Copy Citation Text

    Xiaodong ZHANG, Shuo WANG, Shaoshu GAO, Xinrui WANG, Long ZHANG. Infrared and Visible Image Fusion Method Based on Information Enhancement and Mask Loss[J]. Acta Photonica Sinica, 2024, 53(9): 0910003

    Download Citation

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

    Category:

    Received: Jan. 29, 2024

    Accepted: Apr. 24, 2024

    Published Online: Nov. 13, 2024

    The Author Email: WANG Shuo (S22070043@s.upc.edu.cn)

    DOI:10.3788/gzxb20245309.0910003

    Topics