Fig. 1. Neural network structure of ASPP-CNN

Fig. 2. Structure of ASPP

Fig. 3. Training and testing process of neural networks

Fig. 4. Loss function curve

Fig. 5. Unwrapping results of sphere

Fig. 6. Unwrapping results of random surface data

Fig. 7. Optimized unwrapping results and errors of spherical and random surfaces

Fig. 8. Wrapped phase data with simulated environmental noise and unwrapping result

Fig. 9. Point diffraction interferometry system

Fig. 10. Seven-step phase-shift interferograms collected by the point diffraction interferometry system

Fig. 11. Wrapped phase and unwrapped phase by ASPP-CNN method and 2D surface shape

Fig. 12. Surface shape， PV and RMS values obtained by ESDI interferogram analysis software

Fig. 13. Unwrapping results， 3D surface fitting results， and errors under different methods