ABSTRACT
In the single-plane phase retrieval method, the use of a fixed object support is not efficient and could lead to inaccurate reconstructions. While there have been adaptive support strategies for the single-plane method, numerical processing is slow because such strategies are based in the space domain. Here, an adaptive support strategy based in the Fourier domain in conjunction with the multiple-plane phase retrieval method is presented and demonstrated through simulations and experiments. Optimizations of Fourier filter size and mask threshold parameters result in 3× faster convergence compared to the conventional multiple-plane method for the test object waves used. The proposed strategy offers fast, automated determination of the object support and affords the use of fewer intensity patterns.
