This is torch demo implementation of paper including ipynb (Fourier phase retrieval using physics-enhanced deep learning). You may get more information in the original paper:
Zike Zhang, Fei Wang, Qixuan Min, Ying Jin, and Guohai Situ, "Fourier phase retrieval using physics-enhanced deep learning," Opt. Lett. 49, 6129-6132 (2024)
Paper Link: https://doi.org/10.1364/OL.537792
Our method includes two steps: First, we leverage the image formation model of the FPR to guide the generation of data for network training in a self-supervised manner. Second, we exploit the physical model to fine-tune the pre-trained model to impose the physics-consistency constraint on the network prediction.
Our method provides the best image quality, more results can be found in paper and supplement.
