Phase-Retrieval as a Regularization Problem
For researchers in phase-retrieval imaging, it offers a method to estimate difficult-to-determine physical parameters, improving downstream image processing.
The paper proposes a connection between phase-retrieval algorithms and optimization strategies to numerically determine unknown physical parameters in phase-retrieval imaging, avoiding errors from blind parameter choices.
It was recently shown that the phase retrieval imaging of a sample can be modeled as a simple convolution process. Sometimes, such a convolution depends on physical parameters of the sample which are difficult to estimate a priori. In this case, a blind choice for those parameters usually lead to wrong results, e.g., in posterior image segmentation processing. In this manuscript, we propose a simple connection between phase-retrieval algorithms and optimization strategies, which lead us to ways of numerically determining the physical parameters