[Paper Review] Blind Ptychography: Uniqueness and Ambiguities

Albert Fannjiang, P. Chen|arXiv (Cornell University)|Jun 6, 2018

Advanced X-ray Imaging Techniques48 references3 citations

TL;DR

This paper establishes global uniqueness for blind ptychography under mild phase constraints on a random mask, proving that the reconstructed masked object is uniquely determined up to a constant scaling and affine phase factor. Using mixing schemes to connect ambiguity across the domain, it resolves local phase drifts into a global affine profile, ensuring simultaneous recovery of object and mask despite unknown illumination.

ABSTRACT

Ptychography with an unknown mask and object is analyzed for general ptychographic measurement schemes that are strongly connected and possess an anchor. Under a mild constraint on the mask phase, it is proved that the masked object estimate must be the product of a block phase factor and the true masked object. This local uniqueness manifests itself in the phase drift equation that determines the ambiguity at different locations connected by ptychographic shifts. The proposed mixing schemes effectively connects the ambiguity throughout the whole domain such that a distinct ambiguity profile arises and consequently possess the global uniqueness that the block phases have an affine profile and that the object and mask can be simultaneously recovered up to a constant scaling factor and an affine phase factor.

Motivation & Objective

• To resolve the fundamental ambiguity in blind ptychography where both object and probe are unknown.
• To establish conditions under which the masked object and probe can be uniquely recovered from ptychographic measurements.
• To characterize measurement schemes that eliminate local phase ambiguities through global connectivity.
• To show that random masking, under mild phase constraints, enables global uniqueness via mixing schemes.
• To prove that the ambiguity structure reduces to an affine phase profile across the entire domain.

Proposed method

• Analyzes ptychographic measurement schemes that are strongly connected and possess an anchor to ensure global coherence.
• Introduces a phase drift equation to model local ambiguities arising from overlapping scans.
• Uses random phase masks with bounded phase fluctuations to limit ambiguity support and enforce statistical independence.
• Applies probabilistic arguments to bound the likelihood of spurious solutions, showing failure probability decays exponentially with domain size.
• Employs mixing schemes to connect disjoint regions, transforming local ambiguities into a global affine phase profile.
• Derives conditions under which the only consistent solution is the true object and mask up to scaling and affine phase.

Experimental results

Research questions

• RQ1 Under what conditions can the object and unknown mask be uniquely recovered in blind ptychography?
• RQ2 How do local phase ambiguities propagate across overlapping scan regions?
• RQ3 What role does random masking play in eliminating twin-image and translation ambiguities?
• RQ4 Can a global uniqueness result be achieved even when the mask phase is unknown?
• RQ5 How does the structure of the ambiguity profile (e.g., affine vs. block) affect recovery?

Key findings

• The masked object estimate is uniquely determined up to a constant scaling factor and an affine phase factor under mild constraints on the mask phase.
• Local phase ambiguities are governed by a phase drift equation that links shifts across the scan domain.
• Mixing schemes effectively connect ambiguity across the entire domain, transforming local drifts into a global affine phase profile.
• With random masks and sufficient overlap, the probability of incorrect recovery decays exponentially with the number of independent measurements.
• The analysis proves that a vanishing residual in the noiseless case implies correct recovery only when the ambiguity structure is constrained to an affine profile.
• The results establish global uniqueness for strongly connected ptychographic schemes with an anchor, even when the mask is unknown and randomly phased.