Fig. 1. Diagram of fringe projection 3D imaging
Fig. 2. Flowchart of phase calculation from a single fringe image using deep neural network
[40] Fig. 3. Comparison of 3D reconstruction results
[40]. (a) Fourier transform profilometry, (b) windowed Fourier transform profilometry, (c) fringe analysis based on deep learning, and (d) 12-step phase-shifting profilometry
Fig. 4. Flowchart of label enhanced and patch based deep learning fringe analysis for phase retrieval
[41] Fig. 5. Phase measurement of hand movement at six different moments by FT and DNN methods
[41] Fig. 6. Diagram of fringe image denoising using deep learning
[42] Fig. 7. Test results
[42]. (a1), (a2) Simulation fringe pattern with noise; (b1), (b2) fringe pattern without noise; (c1), (c2) denoised results with deep learning
Fig. 8. Schematic of phase unwrapping using PhaseNet
[43] Fig. 9. Results of different wrapped shapes using PhaseNet
[43]. (a) Wrapped phase; (b) unwrapped phase; (c) fringe order with PhaseNet
Fig. 10. Schematics of the training and testing of the neural network
[44]. (a) training; (b) testing
Fig. 11. Comparison of results of phase unwrapping of dynamic candle flame
[44]. Wrap represents the wrapped phase; CNN represents the phase unwrapped by this method; LS represents the phase unwrapped by the least square method; Diff represents the difference between the results of CNN and LS methods
Fig. 12. Schematic of temporal phase unwrapping using deep learning
[45] Fig. 13. Comparison between traditional MF-TPU and the deep learning based method for high-frequency phase unwrapping (for example, the frequencies are 8, 16, 32, 48 and 64 respectively)
[45] Fig. 14. Neural network structure diagram of height estimation from a single fringe image
[46] Fig. 15. Experimental results of spherical, triangular bevel and face image grating
[46]. The first column is the fringe image of the input neural network; the second column is the true simulated height distribution; the third column is the height distribution of the output of the neural network; the last column is the error distribution map based on the second column and the third column
Fig. 16. Flowchart for projector distortion correction with deep learning
[47] Fig. 17. Test results
[47]. (a) 3D shape of the original data; (b) error distribution of the original data; (c) 3D shape of the corrected data; (d) error distribution of the corrected data
Fig. 18. Diagram of micro deep learning profilometry
[49] Fig. 19. High speed 3D imaging of a falling table tennis and static plaster at speed of 20 000 frame/s
[49]