Fast, accurate, and predictive method for atom detection in site-resolved images of microtrap arrays

Optical detection of atoms or molecules in microtrap arrays is one of the enabling techniques for quantum simulation and computation. In many settings the individual traps are not optically resolved, and detection accuracy depends crucially on an image-processing algorithm. This study introduces an optimal method, rooted in estimation theory, that can drastically improve detection accuracy compared to familiar algorithms. It also provides a rigorous definition for the signal-to-noise ratio of the problem, which can be used to define the practical conditions under which accurate detection is possible, and to rationalize the design of future experiments in affected fields.