• Journal of Infrared and Millimeter Waves
  • Vol. 42, Issue 6, 906 (2023)
Meng DING1、*, Song GUAN2, Shuai LI1, Kuai-Kuai YU2, and Yi-Ming XU1
Author Affiliations
  • 1College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China
  • 2Science and Technology on Electro-Optical Information Security Control Laboratory,Tianjin 300308,China
  • show less
    DOI: 10.11972/j.issn.1001-9014.2023.06.024 Cite this Article
    Meng DING, Song GUAN, Shuai LI, Kuai-Kuai YU, Yi-Ming XU. Depth estimation of thermal infrared images based on self-supervised learning[J]. Journal of Infrared and Millimeter Waves, 2023, 42(6): 906 Copy Citation Text show less
    The framework of depth estimation using thermal infrared image sequences
    Fig. 1. The framework of depth estimation using thermal infrared image sequences
    The depth estimation network of this paper ( The number below Conv indicates the number of convolutional kernels)
    Fig. 2. The depth estimation network of this paper ( The number below Conv indicates the number of convolutional kernels)
    The structure of ECANet
    Fig. 3. The structure of ECANet
    Test images from the FLIR dataset and corresponding depth maps
    Fig. 4. Test images from the FLIR dataset and corresponding depth maps
    Test images from the FLIR A35 TIR camera and corresponding depth maps
    Fig. 5. Test images from the FLIR A35 TIR camera and corresponding depth maps
    Input image and distance estimation results,(a) the input image and the ground truth,(b) the result of distance estimation by the proposed method,(c) the result of distance estimation by HR-Depth,(d) the result of distance estimation by Monodepth2
    Fig. 6. Input image and distance estimation results,(a) the input image and the ground truth,(b) the result of distance estimation by the proposed method,(c) the result of distance estimation by HR-Depth,(d) the result of distance estimation by Monodepth2
    参数FLIR-Tau2FLIR-A35
    图像分辨率640×512320×256
    相机参数

    HFOV 45°

    VFOV 37°

    13 mm f/1.0

    HFOV 48°

    VFOV 39°

    9 mm f/1.0

    相机内参数矩阵0.66900.5000.8280.50001000010.640300.5000.80030.5000100001
    图像采样率30 Hz30 Hz
    Table 1. Related parameters of thermal infrared cameras
    参数数值
    ResNet层数18
    学习率0.000 1
    迭代次数20
    Table 2. Training parameters
    方法ProposedHR-Depthmonodepth2
    E19.58%20.09%21.68%
    Table 3. Error Rates of depth estimation for different networks
    方法E
    <10%<20%<30%>30%
    Proposed41.67%66.67%90.00%10.00%
    HR-Depth36.67%63.33%86.67%13.33%
    monodepth225.00%58.33%85.00%15.00%
    Table 4. Proportions of different network error distribution intervals (%)
    Meng DING, Song GUAN, Shuai LI, Kuai-Kuai YU, Yi-Ming XU. Depth estimation of thermal infrared images based on self-supervised learning[J]. Journal of Infrared and Millimeter Waves, 2023, 42(6): 906
    Download Citation