Acta Optica Sinica, Volume. 45, Issue 15, 1500001(2025)

Metasurfaces-Enabled Advanced Multispectral Imaging and Image Feature Detection (Invited)

Jiajun Wu1,2, Chen Chen1,3, Xiaoyuan Liu1,4, Binfeng Ju2, and Din Ping Tsai1,4,5,6、*
Author Affiliations
  • 1Department of Electrical Engineering, City University of Hong Kong, Hong Kong 999077, China
  • 2School of Mechanical Engineering, Zhejiang University, Hangzhou 310027, Zhejiang , China
  • 3College of Engineering and Applied Sciences, Nanjing University, Nanjing 210093, Jiangsu , China
  • 4Department of Physics, City University of Hong Kong, Hong Kong 999077, China
  • 5State Key Laboratory of Terahertz and Millimeter Waves, City University of Hong Kong, Hong Kong 999077, China
  • 6Centre for Biosystems, Neuroscience, and Nanotechnology, City University of Hong Kong, Hong Kong 999077, China
  • show less
    Figures & Tables(14)
    Dispersive spectral imaging metasurface. (a) Broadband anomalous reflection metasurface for spectral demultiplexing[58]; (b) dispersive metasurface for off-axis focusing[62]; (c) aberration corrected dispersive metasurface for off-axis focusing[63]; (d) compact folded spectrometer with metasurfaces[64]; (e) optical field imaging by metasurface array with transverse dispersion[57]
    Filter-type spectral imaging metasurface. (a) Plasmonic metasurface with nanoslit structure for spectral filtering[68]; (b) multispectral imaging using metasurface with nanohole array[69]; (c) tunable metasurface based on GST materials[67]; (d) high-Q metasurface array for molecular absorption spectroscopy[65]; (e) cascaded metasurfaces for multispectral imaging[66]; (f) multispectral imaging with multi-resonant metasurface using CODE algorithm[77]
    Photonic crystal metasurface. (a) Photonic crystal spectrometer[93]; (b) single-photon spectrometer with integrated photonic crystal filters[94]; (c) multispectral imaging with on-chip spectral sensors based on photonic crystal slabs[95]
    Spectral imaging metasurface by computationally reconstructed. (a) Tunable graphene metasurface spectrometer[96]; (b) multispectral imaging metasurface with free-form unit cells[97]; (c) anti-spoofing face recognition based on hyperspectral imaging metasurface[98]; (d) broadband encoding stochastic hyperspectral imaging based on deep learning[99]; (e) hyperspectral polarization imaging empowered by machine learning[100]
    Artificial intelligence empowered spectral imaging metasurface. (a) Linear least squares regression algorithm for identifying fingerprints of biomolecules[109]; (b) deep neural network for recognition and detection of biomolecules in metasurface microfluidic devices[108]; (c) multispectral imaging via genetic algorithm combined with compressed sensing[111]; (d) hyperspectral imaging based on deep neural network[112]
    Passive depth sensing metasurface. (a) Binocular meta-lens[49]; (b) monocular meta-lens for polarization-multiplexed imaging[117]; (c) spider-inspired monocular dual-defocus meta-lens[119]; (d) depth sensing metasurface based on PSF encoding[120]; (e) achromatic meta-lens array for light field imaging[122]; (f) meta-lens array for light field imaging and point cloud projection multiplexing[123]
    Active depth sensing metasurface. (a) 3D reconstruction based on metasurface point cloud projection[124]; (b) depth sensing metasurface with holographic point cloud pattern encoding[130]; (c) metasurface with 180° field-of-view point cloud projection[131]; (d) metasurface- and PCSEL-based structured light for facial recognition[132]; (e) beam steering with meta-lens array for distance detection[133]; (f) 3D imaging using super-dispersion metasurface[134]; (g) multi-wavelength structured light projection for 3D imaging[135]
    Metasurface for edge detection. (a) One-dimensional spatial differentiation for edge detection[141]; (b) edge detection based on polarization-entangled photon source[142]; (c) second-order differentiation metasurface for biological cell imaging[145]; (d) metasurface differentiator with tailored polarization responses[146]
    Metasurface for edge detection. (a) Metasurface spatial differentiator with oblique incidence[147]; (b) edge detection metasurface based on quasi-BIC mode[148]; (c) edge detection metasurface under incoherent broadband illumination[150]; (d) long-wave infrared metasurface differentiator optimized by inverse design[151]; (e) broadband high order spatial differentiator based on meta-lens[152]
    Artificial intelligence empowered multi-dimensional feature detection of metasurface. (a) Edge enhanced depth perception with binocular meta-lens[157]; (b) large depth-of-field imaging with multi-scale convolutional neural networks[158]; (c) non-iterative metasurface holography design method based on GAN and VAE[159]
    Recent advances in frontier applications of metasurface. (a) Near-infrared silicon-based meta-lens with diameter of 80 mm[160]; (b) all-glass large-aperture meta-lens with diameter of 100 mm[161]; (c) metasurface solar optical reflector for spacecraft thermal protection layers[166]; (d) metasurface-enabled underwater wireless optical communication[162]
    • Table 1. Performance of spectral imaging metasurface

      View table

      Table 1. Performance of spectral imaging metasurface

      Working

      principle

      SpectroscopyImagingAI assitedRef.
      Spectral rangeSpectral resolutionSpatial resolutionField of view
      Dispersive1.1‒1.6 μm0.27 nm/mradNo[62]
      488‒660 nm~1 nmNo[63]
      760‒860 nm~1.2 nmNo[64]
      450‒650 nm4 nm75 pixel×75 pixel, diffraction-limitedYes[57]
      750‒850 nm~1.5 nm~0.075°(angular)30°No[26]
      Filter-type1350‒1750 cm-1 (~5.7‒7.4 μm)4 cm-1(~17 nm)No[65]
      795‒980 nm20 channels~58000 pixel, diffraction-limited20°No[66]
      500‒650 nm18 channels484 pixel×192 pixelYes[77]

      Computationally

      reconstructed

      550‒750 nm~1 nm60 pixel×60 pixelNo[95]
      450‒750 nm~0.5 nm356 pixel×436 pixelYes[97]
      700‒1150 nm~0.23 nm20 pixel×20 pixelYes[100]
      400‒800 nm~1.99 nm135 pixel×75 pixelYes[81]
    • Table 2. Comparison of performance for depth sensing metasurfaces

      View table

      Table 2. Comparison of performance for depth sensing metasurfaces

      Working principleDetection rangeAccuracySpatial /angular resolutionField of viewAI assitedRef.
      Passive>20 cm~50 μmSmall (<5°)Yes[49]
      >25 cm~1%No[118]
      >25 cmNo[117]
      >10 cm~5%No[119]
      ~50 cm<1 cm~21.65 μmNo[122]
      Active>30 cm~0.25 cm~21.65 μmYes[123]
      >30 mm<0.24 mm~0.627°~90°No[124]
      >10 cm~1.9%1.757°~180°No[131]
      >50 cm~10 mm0.611°158°No[132]
    • Table 3. Performance of edge detection metasurfaces

      View table

      Table 3. Performance of edge detection metasurfaces

      Differentiation orderSpectral rangePolarization independentPolarization state of incident lightDirectionalityRef.
      First430‒670 nmNoLPOne-dimension[141]
      Narrow bandNoLPOne-dimension[142]
      400‒800 nmNoLPTwo-dimension[143]
      480‒630 nmNoCPTwo-dimension[144]
      7.5‒13.5 μmNoLPTwo-dimension[151]
      SecondNarrow bandNoLPTwo-dimension[145]
      Narrow bandYesRandomTwo-dimension[146]
      Narrow bandYesRandomTwo-dimension[148]
      500‒800 nmYesRandomTwo-dimension[150]
      First to fourth490‒633 nmNoCPTwo-dimension[152]
    Tools

    Get Citation

    Copy Citation Text

    Jiajun Wu, Chen Chen, Xiaoyuan Liu, Binfeng Ju, Din Ping Tsai. Metasurfaces-Enabled Advanced Multispectral Imaging and Image Feature Detection (Invited)[J]. Acta Optica Sinica, 2025, 45(15): 1500001

    Download Citation

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

    Category: Reviews

    Received: Mar. 13, 2025

    Accepted: Apr. 27, 2025

    Published Online: Aug. 15, 2025

    The Author Email: Din Ping Tsai (dptsai@cityu.edu.hk)

    DOI:10.3788/AOS250737

    CSTR:32393.14.AOS250737

    Topics