Quantum imaging uses correlated or squeezed light (entangled photon pairs, squeezed beams) to form images with contrast, wavelength coverage or noise properties that a classical beam of the same brightness cannot provide, and as of 2026 it remains a set of laboratory demonstrations whose most convincing advantage is wavelength conversion rather than raw sensitivity.
Summary
Quantum imaging is a family of optical techniques that exploit non-classical light. Three strands dominate. Ghost and two-photon imaging use spatially entangled photon pairs, typically from spontaneous parametric down-conversion (SPDC), where one photon probes the object and the other is detected on a camera or scanned detector; the image appears only in the coincidence statistics between the two arms. Imaging with undetected light (QIUL), based on nonlinear interferometers and induced coherence, probes the sample with one photon of the pair but detects only its partner, so the sample can be illuminated at, for example, mid-infrared wavelengths while detection happens in the visible, where detectors are cheap and efficient. Sub-shot-noise imaging and sensing uses squeezed or twin-beam light to push intensity or phase measurements below the noise floor set by Poissonian photon statistics, in principle towards Heisenberg scaling.
The physics is settled; the engineering parameters decide whether any of it is useful. The first is photon budget. Correlation-based methods throw away most detections: a scattering-media contrast demonstration improved image contrast by post-selecting spatially correlated coincidences, but at the cost of higher shot noise because far fewer events survive. The second is acquisition time. Conventional camera-based covariance imaging needs tens of thousands of frames to pull correlations out of noise; recent work using a kurtosis-difference weighting reports usable results at 5,000 frames and removes the need to pre-calibrate a single correlation centre. The third is loss. Heisenberg scaling in SU(1,1) interferometers is known to be fragile under realistic detection efficiency and loss, which is why recent work concentrates on differential intensity measurements that degrade gracefully rather than on ideal Yurke-type schemes.
Evidence base
Open questions
Assessment drafted 2026-08-31 from up to 13 KB sources using the technology-scorecard framework; scores are a draft read pending review.