ShrinkWrap-Optimized Phase Retrieval for High-Resolution Coherent Diffractive Imaging: Diffraction Applications and Image Encryption

Coherent diffractive imaging (CDI) has become a critical technique across multiple disciplines, owing to its phase retrieval capability. While successfully implemented in various scientific domains, inherent limitations persist when processing specimens with complex structures, suggesting potential algorithmic enhancement. To explore potential improvements, we present a dedicated phase retrieval framework incorporating ShrinkWrap regularization. The study tested two representative specimen types: standardized resolution targets (USAF1951) and non-periodic plant tissue sections. Diffraction patterns were acquired via a modular optical setup and processed using a MATLAB algorithm integrating modulus constraints and adaptive support updates. Preliminary comparisons with conventional error reduction (ER) methods demonstrated noticeable resolution improvements under equivalent noise conditions. Additionally, in biological specimens, key structural features were successfully reconstructed despite residual noise. Based on the algorithms, a novel image encryption scheme was developed by encoding plain text into phase-space ambiguities, achieving Shannon entropy of 7.98 against brute-force attacks. These results demonstrate the potential of physics-informed regularization in advancing CDI for applications ranging from nanoscale metrology to secure data transmission.