Random Positional Deviations Correction for Each LED via ePIE in Fourier Ptychographic Microscopy

Fourier ptychography microscopy (FPM) is a lately developed technique, which achieves wide field, high resolution, and phase imaging, simultaneously. FPM stitches together the captured low-resolution images corresponding to angular varying illuminations in Fourier domain utilizing the concept of synthetic aperture and phase retrieval algorithms, which can surpass the space-bandwidth product limit of the objective lens and reconstruct a high-resolution complex image. In general FPM system, the LED source is important for the reconstructed quality and it is sensitive to the positions of each LED element. We find that the random positional deviations of each LED element can bring errors in reconstructed results, which is relative to a feedback parameter. To improve the reconstruction rate and correct random deviations, we combine an initial phase guess and a feedback parameter based on differential phase contrast and extended ptychographical iterative engine to propose an optimized iteration process for FPM. The simulated and experimental results indicate that the proposed method shows the reliability and validity towards the random deviations yet accelerates the convergence. More importantly, it is verified that this method can accelerate the convergence, reduce the requirement of LED array accuracy, and improve the quality of the reconstructed results.