Single-shot coherent diffraction imaging with weighted complex structure tensor

Phase retrieval is a critical process in coherent imaging systems, designed to reconstruct both amplitude and phase from only intensity recording. A significant challenge in single-shot phase retrieval arises from the reduction in the number of measurements, which impedes the accurate resolving of fine details. In this Letter, we propose an unsupervised noise-robust single-shot phase retrieval algorithm that integrates alternating direction optimization with weighted complex structure tensor total variation regularization. This approach effectively balances data fitting with the preservation of texture details. Both simulation results and experimental validation demonstrate that the proposed method is applicable to complex-valued objects, enabling the reconstruction of the complex amplitude of texture-rich objects with enhanced fidelity and fine detail.