Phase retrieval: Fienup's algorithms

Python implementation of Fienup's phase retrieval algorithms, which allow reconstruction of greyscale images only from measurements of the magnitudes. The following methods are implemented:

• input-output algorithm
• output-output algorithm
• hybrid input-output algorithm

Example

image = imageio.imread('test.png', as_gray=True)
magnitudes = np.abs(np.fft.fft2(image))
result = fienup_phase_retrieval(magnitudes, steps=500, verbose=False)

Requirements

• Python (>= 3.7.3)
• NumPy (>= 1.17.2)
• matplotlib (tested with 3.1.1)
• imageio (tested with 2.6.0)

References

[1] E. Osherovich, Numerical methods for phase retrieval, arXiv:1203.4756, 2012

[2] J. R. Fienup, Phase retrieval algorithms: a comparison, Applied Optics, 1982

[3] https://github.com/cwg45/Image-Reconstruction