Fig. 1. Typical NLoS imaging modalities

Fig. 2. Classification of reconstruction algorithms of NLoS imaging

Fig. 3. Schematic diagrams of (a) three reflected light trajectories of NLoS imaging and (b) hidden object reconstruction principle (adapted from Ref. [8])

Fig. 4. Schematic diagram of the principle of the inverse ellipsoidal projection reconstruction algorithm^[11]

Fig. 5. Schematic diagram of the hardware configuration for confocal NLoS imaging^[29]

Fig. 6. Schematic diagram of the passive NLoS imaging system based on occlusion-enhanced imaging^[49]

Fig. 7. Schematic diagram of NLoS imaging through intermediary wall corner penumbra

Fig. 8. Schematic diagram of the active NLoS imaging implementing process based on deep learning (adapted from Ref. [62])

Fig. 9. Schematic diagram of the NLoS imaging based on deep learning and optical transport matrices decomposition (adapted from Ref. [74])

Fig. 10. (a) Passive NLoS imaging and (b) implementing process of popular embedding and optimal transmission generated step by step based on deep learning (adapted from Ref. [84])

Fig. 11. Comparison of reconstruction results of 6 hidden scenes using ShapeNet and LCT algorithms^[62]

Fig. 12. Comparison of reconstruction results of 3 hidden scenes using CNN and PR algorithms (HIO and Alt-Min)^[72]

Fig. 13. An illustration of the NeTF network architecture ^[75]

Fig. 14. Comparison of reconstruction results of 6 hidden scenes using NeTF and traditional NLoS reconstruction algorithms (Phasor Field, F-K and DLCT)^[75]

Fig. 15. Comparison of reconstruction results of 4 type of hidden scenes using NLOS-OT and U-Net, C-GAN^[80]