Efficient near-field ptychography reconstruction using the Hessian operator.

Area of Science:

• Coherent diffractive imaging
• Computational imaging
• Materials science

Background:

• X-ray ptychography is a powerful coherent imaging technique for complex object and probe reconstruction.
• Current reconstruction methods primarily use computationally efficient first-order algorithms.
• Higher-order methods offer potential accuracy gains but are often computationally prohibitive.

Purpose of the Study:

• To develop a computationally efficient mathematical framework for higher-order ptychography reconstruction.
• To enable simultaneous reconstruction of object, probe, and object positions using second-order information.
• To reduce the computational cost associated with advanced ptychography reconstruction.

Main Methods:

• Developed a mathematical framework utilizing second-order information via efficient bilinear Hessian and Hessian operator computation.
• Formulated the approach for Gaussian-based models, facilitating simultaneous reconstruction.
• Integrated derived Hessian formulas into optimization schemes for enhanced reconstruction.

Main Results:

• Demonstrated a ten-fold reduction in computation time compared to traditional first-order methods.
• Validated the approach using both synthetic data and experimental near-field ptychography data.
• Showcased the adaptability of the framework to various ptychography problem formulations.

Conclusions:

• The presented second-order framework offers a significant speedup for X-ray ptychography reconstruction.
• This advancement makes higher-order reconstruction methods more computationally feasible.
• The well-structured and adaptable formulas pave the way for broader adoption in advanced imaging applications.