The so-called Rayleigh limit is not an intrinsic property of light, but a consequence of the imaging instrument measuring collected photons in a suboptimal way. A new framework, dubbed Adaptive Information Mining (AIM), proposes to overcome this barrier through sequentially refined measurements of incoming photons, aiming to approach the fundamental quantum limit of resolution even in regimes well below the Rayleigh scale.

The approach employs a reconfigurable imaging system that dynamically reshapes the instrument's measurement responses to maximise Fisher information and guide scene estimation. Crucially, it requires no prior knowledge of the positions or brightnesses of the observed point sources, a requirement common to other super-resolution strategies that limits their applicability in real-world scenarios.

In the demonstrations, AIM achieves an estimation error 2 to 5 times closer to the fundamental quantum limit than conventional direct imaging when observing point sources. The authors further show that the framework can be extended to fully blind multi-parameter estimation, including scenarios with an unknown number of sources or with turbulence-induced wavefront distortions.

The proposal thus addresses the limitations of conventional fixed-response imaging and offers, according to its authors, a practical route towards quantum-limited super-resolution imaging. The work is published as a preprint on arXiv and opens the door to future experimental verification in real optical systems.