• Laser & Optoelectronics Progress
  • Vol. 60, Issue 10, 1030002 (2023)
Xueli Lin1,2, Zilin Wang1,2, Yanxia Zou1,2, Hao Liu1,2..., Ran Hao1,2 and Shangzhong Jin1,2,*|Show fewer author(s)
Author Affiliations
  • 1College of Optical and Electronic Technology, China Jiliang University, Hangzhou 310018, Zhejiang, China
  • 2Key Laboratory of Zhejiang Province on Modern Measurement Technology and Instruments, Hangzhou 310018, Zhejiang, China
  • show less
    DOI: 10.3788/LOP220984 Cite this Article Set citation alerts
    Xueli Lin, Zilin Wang, Yanxia Zou, Hao Liu, Ran Hao, Shangzhong Jin. Filtering Hyperspectral Imaging Technology Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2023, 60(10): 1030002 Copy Citation Text show less
    Spectral imaging model. (a) Schematic of principle; (b) schematic of network
    Fig. 1. Spectral imaging model. (a) Schematic of principle; (b) schematic of network
    Simulation of light modulation by filter and camera based on LN layer
    Fig. 2. Simulation of light modulation by filter and camera based on LN layer
    Imaging principle of PS Kappa DX4 hyperspectral camera
    Fig. 3. Imaging principle of PS Kappa DX4 hyperspectral camera
    Comparison of the spectral reconstruction performance of the optimized model with two structural filter schemes. (a) RGB image, tf MRAE and nl MRAE images, tf MRAE hist and nl MRAE hist images; (b) reconstructed irradiance of some pixels in sample No. 3
    Fig. 4. Comparison of the spectral reconstruction performance of the optimized model with two structural filter schemes. (a) RGB image, tf MRAE and nl MRAE images, tf MRAE hist and nl MRAE hist images; (b) reconstructed irradiance of some pixels in sample No. 3
    Comparison of the optimized filter's spectral response function. (a) Thin film interference filter scheme; (b) no-limited structure filter scheme
    Fig. 5. Comparison of the optimized filter's spectral response function. (a) Thin film interference filter scheme; (b) no-limited structure filter scheme
    Impact of various array sizes on the spectral imaging performance of no-limited structure filters. (a) RGB image, tf MRAE and nl MRAE images, tf MRAE hist and nl MRAE hist images; (b) reconstructed irradiance of some pixels in sample No. 3
    Fig. 6. Impact of various array sizes on the spectral imaging performance of no-limited structure filters. (a) RGB image, tf MRAE and nl MRAE images, tf MRAE hist and nl MRAE hist images; (b) reconstructed irradiance of some pixels in sample No. 3
    LayerInputOuput
    BNn2n2
    PReLUn2n2
    LNn2600
    BN600600
    PReLU600600
    LN600300
    BN300300
    PReLU300300
    LN300100
    BN100100
    PReLU100100
    LN10030
    BN3030
    PReLU3030
    Table 1. Structure of NET2
    Xueli Lin, Zilin Wang, Yanxia Zou, Hao Liu, Ran Hao, Shangzhong Jin. Filtering Hyperspectral Imaging Technology Based on Deep Learning[J]. Laser & Optoelectronics Progress, 2023, 60(10): 1030002
    Download Citation