Phase retrieval from random phase-shifting interferograms using neural network and least squares method

This paper proposes a neural network and least squares method to retrieve phase from three-frame random phase-shifting interferograms. The phase retrieval method involves two processes. Firstly, a neural network is utilized to predict phase shifts of the three-frame random phase-shifting interferograms. After the phase shifts are determined, the phase is retrieved using the least squares method. The method is simple, and does not require iterative calculation. The accuracy of the proposed method is verified by comparing the advanced iterative algorithm. Through the analysis of the simulated interferograms, the root mean square (RMS) of phase error can approach 0.1 rad. The interferograms recorded in the interferometer verifies the feasibility.