Uncertainty-aware Fourier ptychography

Area of Science:

• Optics and Imaging Science
• Computational Imaging
• Holography

Background:

• Fourier ptychography (FP) provides wide field-of-view and high-resolution imaging but is sensitive to system uncertainties.
• Current methods address uncertainties like misalignment and aberrations separately, failing to tackle interconnected degradations.
• Challenges include precise numerical modeling, optical aberrations, and data quality limitations in practical FP implementations.

Purpose of the Study:

• To introduce a comprehensive framework, Uncertainty-Aware FP (UA-FP), for simultaneously addressing multiple system uncertainties in FP.
• To develop a differentiable forward model incorporating deterministic and stochastic uncertainties as optimizable parameters.
• To enable robust FP performance without extensive calibration or data collection.

Main Methods:

• Developed a fully differentiable forward imaging model for FP.
• Incorporated deterministic uncertainties (misalignment, aberrations) as optimizable parameters.
• Utilized differentiable optimization with domain-specific priors for stochastic uncertainties (noise, data quality).

Main Results:

• UA-FP achieved superior reconstruction quality under challenging conditions.
• Demonstrated robust performance with reduced sub-spectrum overlap requirements.
• Maintained high-quality reconstructions even with low bit sensor data.

Conclusions:

• UA-FP offers a unified approach to mitigate interconnected uncertainties in FP.
• The framework enhances system reconfigurability and broadens FP's applicability in uncontrolled environments.
• This method advances FP as a robust measurement tool for practical, real-world applications.