Maximum entropy method for diffractive imaging

Area of Science:

• Optics and Photonics
• Image Processing
• Statistical Physics

Background:

• Diffractive imaging is susceptible to noise, particularly quantum noise, which complicates accurate object reconstruction.
• Traditional phase retrieval methods can struggle with noisy data, leading to artifacts and reduced resolution.
• The maximum entropy principle offers a robust framework for solving inverse problems with limited or noisy data.

Purpose of the Study:

• To derive a maximum entropy method (MEM) specifically for diffractive imaging affected by quantum noise.
• To develop and present an iterative phase retrieval algorithm, Hybrid Input-Output MEM (HIO-MEM), for noisy diffractive imaging.
• To investigate the relationship between the new MEM formulation for imaging and conventional MEM in crystallography.

Main Methods:

• Minimization of a Lagrange formula incorporating two information measures to derive the MEM for diffractive imaging.
• Development of an iterative phase retrieval algorithm, HIO-MEM, utilizing the derived MEM.
• Numerical simulations applying HIO-MEM to Fourier intensity data with added Poisson noise.

Main Results:

• The derived MEM provides a principled approach to object reconstruction in the presence of quantum noise.
• The HIO-MEM algorithm demonstrates effectiveness in retrieving phase information from noisy Fourier intensity data.
• The study elucidates the connection between MEM for diffractive imaging and its established use in crystallography.

Conclusions:

• The maximum entropy method is a viable and effective tool for addressing quantum noise in diffractive imaging.
• The HIO-MEM algorithm offers an improved solution for phase retrieval in noisy diffractive imaging applications.
• This work bridges the application of MEM across different scientific domains, from imaging to structural analysis.