DF-FPM: Low Complexity Fourier Ptychographic Microscopy via Selective Dark Field Updating

Abstract

Fourier ptychographic microscopy (FPM) attracts growing interest for its capacity to achieve high resolution, large field-of-view quantitative phase imaging. However, noise contamination in abundant dark field images restricts further improvements in reconstruction quality and computational efficiency. This work reported a low-complexity FPM framework (DF-FPM), which integrated selective dark field updating with plug-and-play stochastic gradient descent. The algorithm utilizes mini-batch containing randomly sampled dark field images for iterative updates, effectively preserving high-frequency details while suppressing noise interference. Compared to conventional FPM, the experimental results demonstrate that our approach exhibits a faster computational speed, which can effectively suppress the influence of noise, recover more image details, and enhance image contrast. Our work improves the computational efficiency without increasing hardware requirements and promotes the practical application of FPM in meeting engineering demands.