Momentum-Accelerated Incoherent Fourier Ptychographic Photography

Abstract

Incoherent Fourier Ptychography (IFP) is a super-resolution imaging method, which requires lots of iteration time to recover the high-resolution object image from raw images. To improve the convergence rate of the IFP, we propose an improved algorithm, termed momentum-accelerated Incoherent Fourier Ptychography (mIFP). Based on the IFP algorithm, the mIFP adds momentum that is commonly used to accelerate the training of weights in a neural network. During the update process, the current vector is updated based on the downward trend of the objective function during several updates and the previous vector. The updated vector is then added to the objective function. We demonstrate the proposed approach by both simulations and experiments. By reducing the recovery time, mIFP may accelerate the applications of IFP.

Toward robust super-resolution imaging: A low-rank approximation approach for pattern-illuminated Fourier ptychography

“…As such, many algorithmic extensions of FP are also applicable in piFP. 26,27 Notably, because of the ability to blindly recover an unknown illumination pattern, piFP is also considered a type of blind SIM technique. 9,28,29 To date, piFP has found applications in distinct fields, including computational photography, 25 fluorescence imaging, 24,[30][31][32] and optical synthetic aperture imaging.…”