Predicting ptychography probe positions using single-shot phase retrieval neural network

• DOI: 10.48550/arxiv.2405.20910
• Publication Date: 2024-05-31
• Authors: Ming Du; Tao Zhou; Junjing Deng; Daniel J. Ching; Steven Henke; Mathew J. Cherukara
• Keywords: Ptychography; Single shot; One shot

Ptychography is a powerful imaging technique that used in variety of fields, including materials science, biology, and nanotechnology. However, the accuracy reconstructed ptychography image highly dependent on recorded probe positions which often contain errors. These errors are typically corrected jointly with phase retrieval through numerical optimization approaches. When error accumulates along scan path or when magnitude large, these approaches may not converge satisfactory result. We propose fundamentally new approach for position prediction data large errors, where neural network to make single-shot individual diffraction patterns, yielding object at each point. The pairwise offsets among images then found using robust registration method, results combined yield complete by constructing solving linear equation. show our method can achieve good accumulating order $10^2$ pixels, makes optimization-based algorithms fail converge. For instruments without sophisticated control equipment such as interferometers, significant practical potential.