Random encoding illuminates Fourier ptychography: achieving uniform phase transfer characteristics for quantitative...

Fourier ptychography (FP) is a powerful computational imaging technique for reconstructing wide-field, high-resolution images. However, conventional FP and other standard quantitative phase methods exhibit non-uniform phase transfer characteristics. Consequently, accurately recovering linear phase ramps is impossible, as the captured images yield only constant values at certain spatial frequencies. To address this, we propose a novel, to our knowledge, FP approach that combines random spatial-domain coded illumination with a deep-learning-based phase recovery algorithm. The randomly encoded illumination guarantees a uniform phase response across the entire synthetic bandwidth, while the deep learning model effectively mitigates the multimodal confusion introduced by multiplexed random illumination. Ultimately, this enables robust and accurate quantitative phase reconstruction. Furthermore, our method reduces the FP data acquisition time and provides the flexibility to customize illumination schemes for specific sample characteristics, broadening its applicability in computational quantitative phase imaging.