Fig. 1. Schematic of the vertical and parallel components of reflected and refracted light^[18]

Fig. 2. (a) Framework of NSST-based polarization image fusion^[45]; (b) The architecture of the proposed network^[48]

Fig. 3. (a), (b) The relationship between the polarization degree and the incident angle of specular light and diffuse light； (c) The curve of light intensity with the rotation angle of polarizer

Fig. 4. (a)(b) Brewster segmentation^[52]; (c) The relationship between incident angle and polarization in infrared and visible light^[53]; (d) Two degree of polarization curves simulated for two wavelengths^[54]; (e) DoLP and DoCP as a function of the zenith angle^[55]

Fig. 5. Imaging result. (a) 3D reconstruction based on fusion of polarization imaging and binocular stereo vision^[64]; (b) Near-infrared monocular polarization 3D imaging^[66]; (c) Linear depth estimation based on a sparse system of linear equations^[69-70]; (d) Polarization 3D reconstruction based on deep learning^[72]

Fig. 6. (a) Polarization measurement device with the metasurface and the CCD array^[79]; (b) SEM image of the fabricated devices^[79]; (c) Polarimetric imaging^[81]; (d) Full-Stokes polarimetric imaging^[82]