Fourier ptychographic microscopy with sparse representation.

Area of Science:

• Computational imaging
• Microscopy techniques
• Optical physics

Background:

• Fourier ptychographic microscopy (FPM) offers wide field-of-view and high-resolution imaging by combining synthetic aperture and phase retrieval.
• FPM reconstructs high-resolution complex images from low-resolution intensity images but is susceptible to noise and local optima during reconstruction.
• Existing FPM methods face challenges in noise suppression and achieving optimal reconstruction quality.

Purpose of the Study:

• To propose and validate a novel FPM method utilizing sparse representation for improved noise depression and reconstruction performance.
• To enhance the quality of high-resolution images reconstructed by FPM.
• To address the limitations of FPM regarding noise vulnerability and local optima.

Main Methods:

• Developed a sparse representation-based FPM approach by formulating the reconstruction as a regularized optimization problem.
• Incorporated a data fidelity term based on maximum likelihood theory.
• Utilized Nash equilibrium to find an approximated solution for the regularized cost function, incorporating sparse priors for amplitude and phase.

Main Results:

• The proposed sparse representation FPM method demonstrated superior performance in noise reduction compared to other approaches.
• Achieved state-of-the-art results in high-resolution image reconstruction using both simulated and experimental data.
• Successfully mitigated issues related to noise and local optima inherent in traditional FPM.

Conclusions:

• Sparse representation is an effective strategy for enhancing Fourier ptychographic microscopy.
• The proposed method significantly improves the quality and reliability of FPM reconstructions.
• This work advances computational microscopy by providing a more robust and high-performing FPM technique.