• Infrared and Laser Engineering
  • Vol. 53, Issue 3, 20230538 (2024)
Pengyue Li1,2,*, Xinying Xu1, Yandong Tang3,4, Zhaoxia Zhang1..., Xiaoxia Han1 and Haifeng Yue2|Show fewer author(s)
Author Affiliations
  • 1College of Electrical and Power Engineering, Taiyuan University of Technology, Taiyuan 030024, China
  • 2Taiyuan Heavy Machinery (Group) Company, Taiyuan 030027, China
  • 3State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China
  • 4Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences, Shenyang 110016, China
  • show less
    DOI: 10.3788/IRLA20230538 Cite this Article
    Pengyue Li, Xinying Xu, Yandong Tang, Zhaoxia Zhang, Xiaoxia Han, Haifeng Yue. Image highlight removal method based on parallel multi-axis self-attention[J]. Infrared and Laser Engineering, 2024, 53(3): 20230538 Copy Citation Text show less
    The overall structure of the proposed network
    Fig. 1. The overall structure of the proposed network
    The visual results of different methods on SD1 dataset
    Fig. 2. The visual results of different methods on SD1 dataset
    The visual results of different methods on RD dataset
    Fig. 3. The visual results of different methods on RD dataset
    The visual results of different methods on SHIQ dataset
    Fig. 4. The visual results of different methods on SHIQ dataset
    Comparison of visual results by ablation models for image highlight removal
    Fig. 5. Comparison of visual results by ablation models for image highlight removal
    DatasetRDSD1SHIQ
    Training dataset1800120009825
    Testing dataset2252 0001000
    Table 1. Data distribution of three public highlight datasets
    DatasetSD1RDSHIQ
    PSNRSSIMPSNRSSIMPSNRSSIM
    DCSR18.480.8918.600.8027.640.92
    SHRBF8.010.319.660.3521.660.75
    NMF18.410.5816.110.5922.820.62
    SCS9.770.3012.960.2813.470.49
    SHRRI12.920.6612.350.6116.340.69
    Ours22.580.9325.690.8934.220.95
    Table 2. Comparison of quantitative results of different methods on three public datasets
    ModelM0M1M2M3M4M5
    PSNR30.1530.7831.4633.6733.1534.22
    SSIM0.910.920.930.940.940.95
    Table 3. Ablation experiments of the contribution of the threshold fusion structure and parallel multi-axis self-attention mechanism
    Pengyue Li, Xinying Xu, Yandong Tang, Zhaoxia Zhang, Xiaoxia Han, Haifeng Yue. Image highlight removal method based on parallel multi-axis self-attention[J]. Infrared and Laser Engineering, 2024, 53(3): 20230538
    Download Citation