Lensless diffraction imaging using a rotated slit modulation

Abstract

Translate Article

Iterative phase retrieval is a computational imaging technique to recover the lost phase information from intensity-only images for lensless diffraction imaging. However, the conventional single-shot phase retrieval is always constrained by the precise selection of support region or sparse assumption. To address it, we propose a multi-rotation phase retrieval (MRPR) algorithm for lensless imaging. In our method, a support-based phase retrieval with total variation is first designed to calculate all illumination patterns, which removes the requirement of rotation angle estimation. Then, the illumination modulation and total variation regularization are combined to restrain the multi-rotation optimization function to iteratively reconstruct the complex-valued image of the sample. In comparison with support-based phase retrieval, our method does not rely on the selection of support region and shows high-fidelity image reconstruction. The simulation and experimental results are given to demonstrate the effectiveness of our method.