Enhanced multi-distance phase retrieval using alternating amplitude constraint

Abstract

Multi-distance computational imaging has garnered widespread attention due to its simple configuration and ability to achieve quantitative phase imaging. Phase retrieval is the core of this imaging technology, and the single-beam multiple-intensity reconstruction (SBMIR) is one of the most widely recognized multiple-distance phase retrieval algorithms. However, its strong constraint can easily cause the solution to be trapped in local minima during the reconstruction process. In this paper, a novel strategy for single-beam multiple-intensity reconstruction based on the alternating amplitude constraint (SBMIR-AC) is introduced. Instead of completely replacing the calculated amplitude with the measured amplitude, our approach preserves a portion of the calculated amplitude information during the iterations. This method weakens the stringency of the amplitude constraint and enables the retrieval algorithm to search in a larger space, thereby avoiding stagnation at local minima. The reconstruction performance is initially verified through simulation and then tested on three different samples. The results show that the suggested SBMIR-AC strategy can effectively improve the accuracy and stability of the retrieval algorithm while maintaining a satisfactory convergence speed. Its excellent capability and low computational complexity make it a promising technique for multi-distance phase retrieval.