Batch-based alternating direction methods of multipliers for Fourier ptychography

Area of Science:

• Optics and Photonics
• Computational Imaging
• Image Reconstruction

Background:

• Fourier ptychography (FP) is a powerful imaging technique but suffers from large datasets and high computational costs due to data redundancy requirements for stable recovery.
• Existing methods like embedded pupil function recovery (EPRY) have limitations in efficiency.

Purpose of the Study:

• To develop a more efficient and computationally tractable method for Fourier ptychography reconstruction.
• To reduce the dataset size and computational burden associated with FP imaging.

Main Methods:

• Implementation of batch-based alternating direction methods of multipliers (ADMM) for FP reconstruction.
• Utilizing the additive property of optical pupils in FP recovery.
• Performing partial updates in sub-problems of the standard ADMM algorithm.

Main Results:

• The proposed batch-based ADMM algorithms demonstrate faster convergence compared to the EPRY algorithm.
• The new methods yield higher-quality reconstructed images.
• Validation performed using both simulated and experimental measurement data.

Conclusions:

• Batch-based ADMM offers a significant improvement in efficiency and image quality for Fourier ptychography.
• The developed scheme effectively addresses the computational challenges of FP reconstruction.
• This advancement enables broader applications of FP imaging by reducing resource demands.