Parameter optimization for enhanced multiple plane phase retrieval with statistical fringe processing and object support

Abstract

Phase retrieval is essential for characterizing transparent objects such as lenses and laser sources. Single-beam multiple-intensity reconstruction (SBMIR) is a widely used iterative phase retrieval technique, but it can suffer from slow convergence and artifacts, especially for weakly scattering objects. To address these issues, we integrate statistical fringe processing (SFP) into SBMIR to generate object support constraints, resulting in the proposed SFP-SBMIR method. We investigate the impact of scanning window size L on the robustness and accuracy of SFP-SBMIR using simulated and experimental data. Results show that an optimal range of L enhances convergence speed and accuracy, while excessively small or large values lead to limited improvement or artifacts due to over-reduction of the object support. The findings highlight the effectiveness of SFP-SBMIR in accelerating phase retrieval and improving reconstruction quality.