Toward robust macroscopic Fourier ptychography: Feature-domain optimization for coupled error suppression

Fourier ptychography (FP) is a powerful computational imaging technique that enables both large field-of-view and high-resolution (HR) imaging. The combination of synthetic aperture and phase retrieval techniques in FP offers a new paradigm for HR long-distance imaging, stimulating the development of macroscopic FP. By scanning the camera aperture, macroscopic FP achieves HR imaging beyond the physical aperture limitations of the camera. However, high-fidelity macroscopic FP reconstruction requires sufficient physical prior knowledge of parameters such as the scanning aperture position, noise type, and fluctuation in illumination intensity, posing extremely high demands on system error calibration. In this work, we present a robust macroscopic FP based on feature-domain optimization algorithm, which eliminates the high sensitivity of reconstruction algorithms to coupled system errors. Our method transforms the reconstruction constraint from the pixel-level grayscale space to the structural feature level, naturally endowing it with high tolerance to system errors. Simulation results have quantitatively proved the better reconstruction performance of our method compared with traditional methods in the case of coupled system errors. Experimental results also validate the effective of our method to achieve high-quality reconstruction with a 2.24 times coherent diffraction limit resolution under coupled system errors, in which the results of traditional methods exhibit remarkable artifacts and even fail to converge.