[Paper Review] Far-Field Microscopy of Sparse Subwavelength Objects

Alexander Szameit, Yoav Shechtman+10 more|arXiv (Cornell University)|Oct 4, 2010

Near-Field Optical Microscopy25 references3 citations

TL;DR

This paper demonstrates real-time, scanning-free super-resolution far-field microscopy by leveraging compressed sensing to reconstruct sub-wavelength features from sparse optical objects. Using only knowledge of sparsity in a known basis, the method achieves 30 nm resolution for 100 nm features with 532 nm light—exceeding the diffraction limit by a factor of 8.

ABSTRACT

We present the experimental reconstruction of sub-wavelength features from the far-field of sparse optical objects. We show that it is sufficient to know that the object is sparse, and only that, and recover 100 nm features with the resolution of 30 nm, for an illuminating wavelength of λ=532 nm. Our technique works in real-time, requires no scanning, and can be implemented in all existing microscopes - optical and non-optical.

Motivation & Objective

• To overcome the diffraction limit in far-field optical microscopy without near-field scanning or fluorescent labeling.
• To enable real-time, non-scanning super-resolution imaging using only the prior knowledge that the object is sparse.
• To demonstrate experimental feasibility of sub-wavelength feature reconstruction in conventional microscopes with standard illumination.
• To validate that sparsity-based compressed sensing enables high-resolution recovery from limited far-field data.

Proposed method

• The method formulates sub-wavelength reconstruction as a convex optimization problem using L1-norm minimization (Basis Pursuit) to approximate the sparsest solution.
• It models the imaging system as a linear transformation F, where the measured far-field data b is related to the object x via FAx ≈ b, with constraints on measurement noise ε.
• The object is represented in a known basis (here, real-space), and the sparsest solution is found by minimizing the L1 norm of the coefficient vector x.
• The approach relies solely on sparsity in a known basis and does not require prior knowledge of the object's structure or scanning.
• The technique is applicable to any imaging system where the transfer function F is known and the object is sparse in a known basis.
• Experimental validation uses fabricated masks with 100 nm holes and 150 nm stripes to generate sub-wavelength optical patterns.

Experimental results

Research questions

• RQ1 Can sub-wavelength features be reconstructed from far-field measurements using only the knowledge of sparsity?
• RQ2 Is it possible to achieve super-resolution beyond the diffraction limit without scanning or near-field probing?
• RQ3 Can compressed sensing-based reconstruction provide robust, real-time imaging in conventional microscopes?
• RQ4 Does the method maintain high resolution even when features are significantly smaller than the wavelength?
• RQ5 Can this approach be generalized to non-optical imaging systems with sparse signals?

Key findings

• The method successfully reconstructed 100 nm features with a resolution of 30 nm using 532 nm light, achieving a resolution enhancement of 8× beyond the diffraction limit.
• The technique achieved sub-wavelength imaging in real-time without scanning, enabling immediate reconstruction of sparse objects.
• The experimental setup used fabricated masks with 100 nm diameter holes and 150 nm wide stripes to generate test sub-wavelength patterns.
• The reconstruction was robust to noise and required no prior knowledge of the object beyond sparsity in a known basis.
• The method is compatible with all existing microscopes, both optical and non-optical, due to its reliance on far-field measurements and standard imaging hardware.
• The results confirm that sparsity-based compressed sensing enables accurate bandwidth extrapolation of evanescent spatial frequencies from propagating far-field data.