Fig. 1. Components of light on transparent object’s surface

Fig. 2. Simulation results. (a) Reflectivity and polarization degree of reflected light; (b) transmissivity and polarization degree of transmitted light

Fig. 3. Diagram of transmission of polarization state of light waves on surface of transparent object

Fig. 4. Process of solving azimuth angle of incident plane at central point. (a) Image of polarization angles; (b) Histogram of central pixel block in image of polarization angles

Fig. 5. Reflected light scenes. (a) Transmitted light image; (b) reflected light image; (c) polarization image in vertical direction; (d) polarization image in parallel direction

Fig. 6. Reflected light separation results when χ=0.2 and γ=0.2, 0.4, 0.8. (a) γ=0.2; (b) γ=0.4; (c) γ=0.8

Fig. 7. NCC curves in different γ. (a) NCC curve of point P₁; (b) NCC curve of point P₂

Fig. 8. Extraction of zero-crossing pixels and variation curves of correlation value with viewing angle. (a) NCC between over-separated transmitted light image and under-separated transmitted light image; (b) extraction result of zero-crossing pixels; (c) variation curve of mutual information with viewing angle; (d) f_R(δ,φ) versus viewing angle

Fig. 9. Flow diagram of proposed algorithm

Fig. 10. Schematic diagram of reflected light data collection

Fig. 11. Polarization images of indoor scene. (a) True transmitted light image; (b) 0° polarization image; (c) 45° polarization image; (d) 90° polarization image; (e) 135° polarization image; (f) polarization image in parallel direction; (g) polarization image in vertical direction; (h) viewing angle image; (i) azimuth angle image

Fig. 12. Polarization images of outdoor scene. (a) True transmitted light image; (b) 0° polarization image; (c) 45° polarization image; (d) 90° polarization image; (e) 135° polarization image; (f) polarization image in parallel direction; (g) polarization image in vertical direction; (h) viewing angle image; (i) azimuth angle image

Fig. 13. Reflected light separation results of indoor scene. (a1)(a2) Proposed algorithm; (b1)(b2) algorithm in Ref. [2]; (c1)(c2) algorithm in Ref. [10]; (d1)(d2) algorithm in Ref. [11]

Fig. 14. Reflected light separation results of outdoor scene. (a1)(a2) Proposed algorithm; (b1)(b2) algorithm in Ref. [2]; (c1)(c2) algorithm in Ref. [10]; (d1)(d2) algorithm in Ref. [11]

Table 1. Quantitative comparison of separation effects of reflected light under different methods