A Measurement Fault Elimination Method in One-dimensional Fourier Phase Retrieval

This paper proposes a measurement fault elimination method applied to Fourier phase retrieval. In the phase retrieval problem, if the diffraction pattern is partially measured incorrectly, the accuracy of reconstruction results will be reduced. This paper considers one-dimensional Fourier phase retrieval and proposes a method that can search the faulty measurements out by comparing Fourier magnitude of different oversampling multiples, thus is possible to filter out pixels that are not working properly in the detector to improve the reconstruction algorithm. Numerical experiments show that the method proposed in this paper can effectively eliminate the influence of faulty measurement and obtain accurate reconstruction.