Optimization of initial guess phase for multiple intensity phase retrieval

Abstract

Phase retrieval is significantly hampered in the case for smooth test objects due to a lack of intensity variation leading to stagnation in the iterative reconstruction process. While there have been numerous studies to improve the method such as optimizations of the number of measurements used, step distance and other parameters, the initial guess phase has not been put into much consideration as it is usually set at a constant value or random value. In this study, we investigate the effects of the initial guess phase on the rate of reconstruction using the multiple-intensity phase retrieval technique, as well as the robustness of phase retrieval through testing different curvatures of test object waves with various guess phases.