Abstract
Entangled states of light are essential for quantum technologies and fundamental tests of physics. Current systems rely on entanglement in 2D degrees of freedom, e.g., polarization states. Increasing the dimensionality provides exponential speedup of quantum computation, enhances the channel capacity and security of quantum communication protocols, and enables quantum imaging; unfortunately, characterizing highdimensional entanglement of even bipartite quantum states remains prohibitively timeconsuming. Here, we develop and experimentally demonstrate a new theory of camera detection that leverages the massive parallelization inherent in an array of pixels. We show that a megapixel array, for example, can measure a joint Hilbert space of 10^{12} dimensions, with a speedup of nearly four ordersofmagnitude over traditional methods. The technique uses standard geometry with existing technology, thus removing barriers of entry to quantum imaging experiments, generalizes readily to arbitrary numbers of entangled photons, and opens previously inaccessible regimes of highdimensional quantum optics.
Introduction
Broad beams of quantum light are a natural pathway to large Hilbert spaces^{1,2,3,4,5,6,7,8,9,10,11}, as they have highdimensional entanglement in transverse spatial modes^{12}. Spatial correlation of biphotons has led to subshotnoise quantum imaging^{13,14}, enhanced resolution^{15}, quantum ghost imaging^{16}, and proposals for quantum lithography^{17}. Despite this work, highdimensional quantum optics remains underdeveloped, largely due to difficulty in measuring the full joint probability distribution. Traditionally, experiments measure coincidences between two singlephoton counting modules (SPCMs) that are each scanned over their own subspace to build up a measurement pointbypoint. Such a procedure is photoninefficient, making highdimensional measurements tedious and prohibitively time consuming. Full quantumstate measurements are impractical even for a relatively small number of dimensions^{18,19}.
In this work, we present a rapid and efficient method of measuring a highdimensional biphoton joint probability distribution via massively parallel coincidence counting. We use a singlephotonsensitive electronmultiplying (EM) CCD camera as a dense array of photon detectors to measure all dimensions of the joint Hilbert space simultaneously. For example, a typical megapixel camera can record a one trilliondimensional joint Hilbert space nearly 10,000× faster than traditional rasterscanning methods. This speedup enables observation of highdimensional features that cannot be seen when only lowdimensional measurements (projections) are made.
Recent efforts with singlephotonsensitive cameras have characterized spatial entanglement^{20,21,22,23,24,25}, but results relied on projection onto only two dimensions and considered only homogeneous distributions, limiting measurements to EPRtype entanglement. In these works, measurements were spatially averaged over the entire plane (losing local detail). Further, to mitigate complications of accidental counts, coincidence measurements were performed in the lowcountrate regime. Here, we show that this assumption is unnecessary and give a general expression for the biphoton joint probability distribution. The exact expression follows from measurements of single and coincidencecount probabilities observed between every pair of pixels over the entire frame simultaneously. The resulting distribution is valid for arbitrary count rates up to detector saturation, enabling more accurate measurements, faster acquisition speeds, and optimization of the signaltonoise ratio.
To demonstrate our method, we characterize the properties of photon pairs entangled in transverse spatial degrees of freedom. A pure entangled photon state is described by the biphoton wave function \(\psi ({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})\), where \({{\boldsymbol{\rho }}}_{i}={x}_{i}{\hat{{\bf{x}}}}_{i}+{y}_{i}{\hat{{\bf{y}}}}_{i}\), and likewise for \({{\boldsymbol{\rho }}}_{j}\). The joint probability of observing one photon at \({{\boldsymbol{\rho }}}_{i}\) and its partner at \({{\boldsymbol{\rho }}}_{j}\) is \({\rm{\Gamma }}({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})={\psi ({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})}^{2}\), which in a discretized basis is \({{\rm{\Gamma }}}_{ij}\). Since each photon may be found in a 2D space (x_{ i }, y_{ i }), the joint probability distribution is a 4D distribution. Like classical lightfield methods^{26}, observation of the full 4D distribution shows details and features that would be lost with conventional projection methods. While we focus on spatial components, we emphasize that our technique may be readily extended to other degrees of freedom, such as spectral modes or orbital angular momentum, by suitable mapping onto the pixels of the camera.
A schematic of the measurement and processing procedure is shown in Fig. 1. Spatially entangled photon pairs are generated via spontaneous parametric downconversion (SPDC) in a βbarium borate (BBO) crystal, cut for typeI phase matching. The spatial entanglement structure has been extensively studied^{12,13,15,21,24,25,27,28,29,30,31,32,33}, and we use it here for a clear experimental demonstration of highdimensional characteristics of entangled photons. The crystal is pumped by a 120 mW, 400 nm cw laser diode that is spatially filtered and collimated (not shown). Spectral filters block the pump beam and select neardegenerate photon pairs at 800 nm (a large bandwidth of 40 nm (FWHM), gives rise to the relatively thick rings in the far field^{32,33}). These are placed immediately after the BBO crystal to prevent induced fluorescence in the subsequent optics. A lens images the far field of the crystal onto an EMCCD camera (Andor iXon Ultra).
Measurement of the biphoton joint probability distribution \({{\rm{\Gamma }}}_{ij}\) is possible with an EMCCD camera due to its high quantum efficiency and low noise floor. The camera is operated in the photoncounting regime, where each pixel is set to one if its graylevel output is above a threshold and zero otherwise^{34} (see Methods). The data consist of a set of \(N\) frames \({C}_{i,n}\) = {0, 1}, where subscript \(i\) is the pixel index (spatial mode) and \(n\) is the frame number. Each frame consists of many counts from both photon events and electronic noise (mainly due to clockinduced charge^{34}). The singlescount probability is
where \({P}_{m}\) is the distribution of the number \(m\) of photon pairs and \({p}_{el}\) is the electronic count probability (e.g., dark counts). The factors \({\mu }_{im}\) and \({\mu }_{\bar{i}m}\) represent the conditional probabilities of detecting at least one photon and zero photons, respectively, given \(m\) pairs arriving within the detector time window (see Table 1)^{35}.
Since the duration of both the exposure and readout of each frame of the EMCCD is much longer than the biphoton correlation time, photons from each pair arrive at the camera within a single frame. The coincidence count probability between all pixels i and j is measured by the average of the tensor product of each frame with itself:
In addition to genuine coincidence counts from entangled photon pairs, there are also accidental counts from uncorrelated photons and noise. These can be accounted for in general by the expression
Where each of the terms \({\mu }_{pqm}\) are related to \({{\rm{\Gamma }}}_{pq}\) and its marginal (see Table 1). The terms in Eq. (3) are coincidences between (1) at least two photons, (2) at least one photon and one electronic noise event, and (3) two noise events. For a Poissonian distribution of pairs, Eq. (3) simplifies, giving an analytic expression for \(\langle {C}_{ij}\rangle \) in terms of \(\langle {C}_{i}\rangle \), \(\langle {C}_{j}\rangle \), and \({{\rm{\Gamma }}}_{ij}\). With Eq. (1) this yields
where α is a constant that depends on the quantum efficiency of the system (see Supplementary Information).
Equation (4) includes the case when several photons arrive at the same pixel. This case has been excluded explicitly by other treatments^{21,22,35}, even though collinear geometry and high spatial entanglement make this case the most likely one. The paradox is often circumvented by considering the lowphotoncount limit, in which the joint probability distribution \({{\rm{\Gamma }}}_{ij}\) becomes proportional to the measured coincidence count rate \(\langle {C}_{ij}\rangle \). However, this assumption is not necessary here; indeed, Eq. (4) remains valid up to detector saturation. The formalism thus covers the entire range of photon intensities and types of detection events, and generalizes straightforwardly to joint distributions of higher numbers of entangled photons.
Figure 1d shows the coincidence count distribution for a particular pixel \(j\) = [\({x}_{j}\) = 70, \({y}_{j}\) = 33], i.e., a 2D slice for all \(i\) = {\({x}_{i}\), \({y}_{i}\)} through the 4D joint distribution \(\langle {C}_{ij}\rangle \). It includes genuine coincidences as well as a large background from accidental counts. Due to the large number of pairs in each frame (~10^{4}), most coincidences are accidentals between photons from different pairs; indeed, Fig. 1d appears very similar to the singles count distribution \(\langle {C}_{i}\rangle \) in Fig. 1c. Genuine coincidences between photons from the same pair, shown in the inset, rise above the background from accidentals. The corresponding 2D slice through the 4D \({{\rm{\Gamma }}}_{ij}\), calculated via Eq. (4), is displayed in Fig. 1e. When one photon is found at \(j\) = [70, 33], its entangled partner is localized near \(i\) = [−70, −32], indicating a high degree of anticorrelation. Such conditional distributions \({{\rm{\Gamma }}}_{ij}\) are measured simultaneously for all \(j\), thus constituting a full measurement of the 4D biphoton joint probability distribution.
Complete measurements of highdimensional joint Hilbert spaces contain detailed, localized information not available in lowerdimensional projections. To demonstrate this, we show \({{\rm{\Gamma }}}_{ij}\) for entangled photons detected at different radial distances \(j\) = [x_{ j }, y_{ j }] from the center of the beam (Fig. 2a–c). There are two main observations: 1) as x_{ j } increases, x_{ i } decreases, and 2) the width along the radial directions increases. The former is necessary to maintain a fixed sum, i.e., \({x}_{i}\) + \({x}_{j}\) ≈ 0, while the latter arises from the radial dependence of the uncertainty in the wave vector k, \({\rm{\Delta }}{k}_{\rho }\approx {k}_{\rho }{\rm{\Delta }}{\bf{k}}/{\bf{k}}\). This effect comes from the rather large spectral bandwidth of the filter (40 nm), as different frequencies are phasematched at different radial momenta \({k}_{\rho }\)^{27,33}. Observation of such features with traditional rasterscanning techniques requires multiple separate measurements. With an EMCCD camera, they are all captured simultaneously in a single image.
In previous studies, the intercorrelation function was measured via image correlation techniques^{21,22}, without measuring the full 4D \({{\rm{\Gamma }}}_{ij}\). However, such measurements provide only the globally averaged correlation and thus neglect any potential internal variation in the joint probability distribution. Here, we calculate the intercorrelation function by projecting \({{\rm{\Gamma }}}_{ij}\) onto the sum coordinate \([({x}_{i}+{x}_{j})/\sqrt{2},({y}_{i}+{y}_{j})/\sqrt{2}]\) (Fig. 2d). The peak near the center indicates that entangled photon pairs are always found near equal and opposite sides of the center, within anticorrelation widths \({\sigma }_{x,+}\) = 20.9 ± 0.3 μm and \({\sigma }_{y,+}\) = 18.6 ± 0.3 μm. Our moreresolved methods show that, even in this simple case, the corresponding widths of the \({{\rm{\Gamma }}}_{ij}\) in Fig. 2a–c vary significantly, with \({\sigma }_{x}\) = 16.1 ± 1.4 μm, 23.0 ± 1.5 μm, and 34.9 ± 2.5 μm, respectively.
Other slices of \({{\rm{\Gamma }}}_{ij}\), along different coordinates, contain different information about the entangled photon pairs. For example, we examine correlations in vertical position within specific columns of the image by fixing [x_{ i }, x_{j}]. While some variation can survive averaging by projection onto 2D planes, such as phasematching and spatial walkoff effects (as observed in type II SPDC^{28}), in contrast our method is capable of measuring arbitrary 4D joint probability distributions. Examples in Fig. 2e–g show strong vertical anticorrelation that changes depending on the horizontal separation of the selected columns (indicated in the insets), with radial variation that diminishes for larger \(x\). As before, projecting \({{\rm{\Gamma }}}_{ij}\) averages this variation (Fig. 2h), resulting in lost information^{22,23,28}.
The massively parallel capability of EMCCD cameras allows for much faster measurement of \({{\rm{\Gamma }}}_{ij}\) than traditional scanning techniques. Rasterscanning pairs of SPCMs, each in a \(d\)dimensional plane, requires \({d}^{2}\) measurements to build a complete measurement. In contrast, an EMCCD measures the entire plane at once, with pixels at each point in the array. While SPCMs have a high effective frame rate (10s of MHz), the frame rate of an EMCCD camera is limited by the readout process (which scales as \(\sqrt{d}\,\)^{36}). Data shown in Figs. 1 and 2 were taken from a subset of 251 × 251 pixels, corresponding to a fourbilliondimensional joint Hilbert space, and were acquired in a matter of hours. A megapixel EMCCD can record a (1024 × 1024)^{2} ≈ one trillion dimensional joint Hilbert space with a signaltonoise ratio of 10 in approximately 11 hours. The same measurement performed with rasterscanning SPCMs is estimated to take 9 years, giving a camera improvement of ~7000×. The EMCCD camera also outperforms compressive sensing methods^{29} for large joint Hilbert spaces and does not require sparsity or numerical retrieval.
Camerabased methods hold clear advantages for quantum imaging applications. Imaging with perfectly correlated photon pairs—with biphoton wave function \(\psi ({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})=\delta ({{\boldsymbol{\rho }}}_{i}{{\boldsymbol{\rho }}}_{j})\)—gives a probability distribution of both photons at the same position in the image plane
where \(t({\boldsymbol{\rho }})\) is the object transmittance and \(h({\boldsymbol{\rho }})\) is the point spread function. The fact that the square of \(h({\boldsymbol{\rho }})\) appears in Eq. (5) means that biphoton imaging has higher resolution than classical coherent imaging [though it has the same resolution as classical incoherent light (of the same coherence area)^{17,30,31,37,38}]. To demonstrate this, we image a resolution chart using spatially entangled biphoton illumination—where one photon is localized near its partner (\(i\) ≈ \(j\))—by projecting the output facet of the nonlinear crystal onto the object, which is then imaged onto the camera (see Fig. 3a, Methods). To ensure the validity of Eq. (5), we measure the incident \({{\rm{\Gamma }}}_{ij}\) without the object; the results confirm strong spatial correlation, visible in both the conditional distributions (Fig. 3b,c) and the projection onto the difference coordinates (Fig. 3d). By fitting to a Gaussian distribution, we find the correlation width \({\sigma }_{}\) = 8.5 ± 0.5 μm. Measurements are then repeated with the object; a 3D projection of \({{\rm{\Gamma }}}_{ij}\), shown in Fig. 3e, displays the image of the resolution chart, its appropriate basis (diagonal plane), and the final spatial correlation distribution of the biphotons (thickness of the diagonal plane). Furthermore, coincidence images taken with entangled photon pairs (Fig. 3f) show nearly identical resolution as incoherent light^{17,30,31}—as measured by direct imaging (singles counts) of photon pair illumination—(Fig. 3g), and clear improvement in resolution over those with an 808 nm laser diode (Fig. 3h), with less noise and higher visibility. For example, the bars within the red boxed region (group 4, element 6) are clearly resolved with entangled photon pairs (\({{\rm{\Gamma }}}_{ii}\), visibility of 0.33 ± 0.03) and incoherent light (\({{\rm{\Gamma }}}_{i}\), visibility of 0.37 ± 0.03), but not with classical coherent light (visibility < 0.04). Ideally, the visibility for entangled photon pairs and incoherent light should be the same; the discrepancy here may be due to the way we approximate \({{\rm{\Gamma }}}_{ii}\) with \({{\rm{\Gamma }}}_{i,i+1}\,\)using adjacent pixels (see Methods).
By using readily available technology and standard imaging geometries, our method removes barriers of entry to experiments in quantum optics. Timeresolved measurements of coincidence counts are replaced by timeaveraged camera measurements of photon correlations, while lowerorder counts and conditional probabilities are bootstrapped to provide complete characterization of joint distribution functions. Further, the massive parallelization inherent in megapixel cameras enables measurement of states with ordersofmagnitude greater dimensionality than previously possible, with similar increases in acquisition speed. With suitable mapping for other degrees of freedom, e.g., dispersive elements for spectral modes or diffractive elements for orbital angular momentum, other types of quantum states can be characterized as well (including multiphoton quantum states via nfold coincidences). Our results thus extend conventional imaging to the quantum domain, providing a pathway for quantum phase retrieval and coherence/entanglement control, and enable new means of quantum information processing with highdimensional entangled states.
Methods
The EMCCD (iXon Ultra 897, Andor) is a highly sensitive camera in which an avalanche gain of up to 1000 amplifies the signal in each pixel before readout. The camera has a pixel size of 16 × 16 μm^{2} with a quantum efficiency of ~70% at 800 nm. To minimize the darkcount rate compared to other noise sources in the camera, it is operated at a temperature of −85 °C. The camera is first characterized by measuring the histogram of the gray scale output of each pixel from many (~10^{6}) frames taken with the shutter closed. The histogram is primarily Gaussian, due to read noise, with an additional exponential tail towards high gray levels due primarily to clockinduced charge (CIC) noise^{34}. We fit the histogram with a Gaussian distribution to find the center (~170) and standard deviation \(\sigma \) (4 to 20, depending on the readout rate). We have found that a threshold set to 2σ above the mean maximizes the signaltonoise ratio. A pixeldependent threshold is used to account for a minor inhomogeneity across the frame. There is a small crosstalk effect between pixels in a single column due to suboptimal charge transfer efficiency upon readout (see Supplementary Information). For this reason, within each 2D frame of \({{\rm{\Gamma }}}_{ij}\), we set to zero the 10 pixels above and below \(i\) = \(j\).
Operating at higher readout rate increases noise from readout and CIC, but we have found that the increased acquisition speed more than compensates, yielding a higher signaltonoise ratio (SNR) for the same total acquisition time. The camera is therefore operated at the fastest available settings: a horizontal readout rate of 17 MHz and a vertical shift time of 0.3 μs, with a vertical clock voltage of +4 V over factory default. The pump laser power and camera exposure time are set to give an optimum peak count probability \(\langle C\rangle \) of ~0.2^{34}. We acquire a number of frames sufficient to achieve the desired SNR. Typically, a series of ~10^{5}–10^{7} images are acquired at a ~1–5 ms exposure time. Many sets of thresholded frames are saved to disk, where each set contains 10^{4} frames as a logical array \({C}_{i,n}\). Each column of the array represents a single frame, and each row represents a pixel. Equation (2) is used to calculate \(\langle {C}_{ij}\rangle \) by matrix multiplication of each set of frames, which are then averaged. To minimize nonergodic effects, the term \(\langle {C}_{i}\rangle \langle {C}_{j}\rangle \) in Eq. (4) is calculated via matrix multiplication of successive frames (see Supplementary Information). Elsewhere, \(\langle {C}_{i}\rangle \) is the average of all frames.
In general, the biphoton wave function in an image plane is given by
where \({\psi }_{s}({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})\) is the wave function incident on the object. With ideally correlated photon pairs, i.e., \({\psi }_{s}({{\boldsymbol{\rho }}}_{i},{{\boldsymbol{\rho }}}_{j})=\delta ({{\boldsymbol{\rho }}}_{i}{{\boldsymbol{\rho }}}_{j})\), the square amplitude of Eq. (6) simplifies to Eq. (5). The highresolution biphoton image therefore lies within \({{\rm{\Gamma }}}_{ii}\), where both entangled photons hit the same pixel. However, as EMCCDs are not photonnumberresolving, they cannot distinguish between one or both photons hitting the same pixel. Instead, we approximate \({{\rm{\Gamma }}}_{ii}\) by the case where the two entangled photons arrive in adjacent pixels, i.e., \({{\rm{\Gamma }}}_{i,i+1}\), as we do in Fig. 3f. This assumption is valid when the biphoton correlation width and image features are both larger than the pixel size.
For ideal imaging (\(h({\boldsymbol{\rho }})\) ≈ \(\delta ({\boldsymbol{\rho }})\)), intensity images are directly proportional to \({t({\boldsymbol{\rho }})}^{2}\), where \(t({\boldsymbol{\rho }})\) is the complex (field) function for transmission. For entangledphoton images,\(\,{\rm{\Gamma }}({\boldsymbol{\rho }},{\boldsymbol{\rho }})\) ∝ \({t({\boldsymbol{\rho }})}^{4}\) (see Eq. (5)). Therefore, we show in Fig. 3f the the square root of the biphoton images, which is proportional to \({t({\boldsymbol{\rho }})}^{2}\), to allow fair comparison to intensity measurements in Fig. 3g,h. This also explains the relative “flatness” of Fig. 3f compared to 3g (which are both computed from the same set of image frames).
References
 1.
O’Brien, J. L. Optical Quantum Computing. Science 318, 1567–1570 (2007).
 2.
Gisin, N. & Thew, R. Quantum communication. Nature Photon. 1, 165–171 (2007).
 3.
AspuruGuzik, A. & Walther, P. Photonic quantum simulators. Nat. Phys. 8, 285–291 (2012).
 4.
Bouwmeester, D. et al. Experimental quantum teleportation. Nature 390, 575–579 (1997).
 5.
Jozsa, R. & Linden, N. On the role of entanglement in quantumcomputational speedup. Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences 459, 2011–2032 (2003).
 6.
Vidal, G. Efficient Classical Simulation of Slightly Entangled Quantum Computations. Physical Review Letters 91, 147902 (2003).
 7.
Wang, C., Deng, F.G., Li, Y.S., Liu, X.S. & Long, G. L. Quantum secure direct communication with highdimension quantum superdense coding. Physical Review A 71, 044305 (2005).
 8.
Huber, M. & Pawłowski, M. Weak randomness in deviceindependent quantum key distribution and the advantage of using highdimensional entanglement. Physical Review A 88, 032309 (2013).
 9.
Lloyd, S. Enhanced Sensitivity of Photodetection via Quantum Illumination. Science 321, 1463–1465 (2008).
 10.
Mohammad, M. et al. Highdimensional quantum cryptography with twisted light. New Journal of Physics 17, 033033 (2015).
 11.
BechmannPasquinucci, H. & Tittel, W. Quantum cryptography using larger alphabets. Physical Review A 61, 062308 (2000).
 12.
Howell, J. C., Bennink, R. S., Bentley, S. J. & Boyd, R. W. Realization of the EinsteinPodolskyRosen Paradox Using Momentum and PositionEntangled Photons from Spontaneous Parametric Down Conversion. Physical Review Letters 92, 210403 (2004).
 13.
Brida, G., Genovese, M. & Ruo Berchera, I. Experimental realization of subshotnoise quantum imaging. Nature Photon. 4, 227–230 (2010).
 14.
Ono, T., Okamoto, R. & Takeuchi, S. An entanglementenhanced microscope. Nat. Commun. 4, 2426 (2013).
 15.
Walborn, S. P., Monken, C. H., Pádua, S. & Souto Ribeiro, P. H. Spatial correlations in parametric downconversion. Physics Reports 495, 87–139 (2010).
 16.
Erkmen, B. I. & Shapiro, J. H. Ghost imaging: from quantum to classical to computational. Advances in Optics and Photonics 2, 405–450 (2010).
 17.
Boto, A. N. et al. Quantum Interferometric Optical Lithography: Exploiting Entanglement to Beat the Diffraction Limit. Physical Review Letters 85, 2733–2736 (2000).
 18.
Krenn, M. et al. Generation and confirmation of a (100 × 100)dimensional entangled quantum system. Proceedings of the National Academy of Sciences 111, 6243–6247 (2014).
 19.
Martin, A. et al. Quantifying Photonic HighDimensional Entanglement. Physical Review Letters 118, 110501 (2017).
 20.
Reichert, M., Defienne, H., Sun, X. & Fleischer, J. W. Biphoton transmission through nonunitary objects. J. Opt. 19, 044004 (2017).
 21.
Moreau, P.A., MouginSisini, J., Devaux, F. & Lantz, E. Realization of the purely spatial EinsteinPodolskyRosen paradox in fullfield images of spontaneous parametric downconversion. Physical Review A 86, 010101 (2012).
 22.
Edgar, M. P. et al. Imaging highdimensional spatial entanglement with a camera. Nat. Commun. 3, 984 (2012).
 23.
Dąbrowski, M., Parniak, M. & Wasilewski, W. Einstein–Podolsky–Rosen paradox in a hybrid bipartite system. Optica 4, 272–275 (2017).
 24.
Lantz, E., Denis, S., Moreau, P.A. & Devaux, F. EinsteinPodolskyRosen paradox in single pairs of images. Optics Express 23, 26472–26478 (2015).
 25.
Reichert, M., Sun, X. & Fleischer, J. W. Quality of Spatial Entanglement Propagation. Physical Review A 95, 063836 (2017).
 26.
Waller, L., Situ, G. & Fleischer, J. W. Phasespace measurement and coherence synthesis of optical beams. Nature Photon. 6, 474–479 (2012).
 27.
Devaux, F., MouginSisini, J., Moreau, P. A. & Lantz, E. Towards the evidence of a purely spatial EinsteinPodolskyRosen paradox in images: measurement scheme and first experimental results. The European Physical Journal D 66, 192 (2012).
 28.
Moreau, P.A., Devaux, F. & Lantz, E. EinsteinPodolskyRosen Paradox in Twin Images. Physical Review Letters 113, 160401 (2014).
 29.
Howland, G. A. & Howell, J. C. Efficient HighDimensional Entanglement Imaging with a CompressiveSensing DoublePixel Camera. Physical Review X 3, 011013 (2013).
 30.
Saleh, B. E. A., Teich, M. C. & Sergienko, A. V. Wolf Equations for TwoPhoton Light. Physical Review Letters 94, 223601 (2005).
 31.
Abouraddy, A. F., Saleh, B. E. A., Sergienko, A. V. & Teich, M. C. Entangledphoton Fourier optics. Journal of the Optical Society of America B 19, 1174–1184 (2002).
 32.
Kurtsiefer, C., Oberparleiter, M. & Weinfurter, H. Highefficiency entangled photon pair collection in typeII parametric fluorescence. Physical Review A 64, 023802 (2001).
 33.
Yanhua, S. Entangled biphoton source  property and preparation. Reports on Progress in Physics 66, 1009 (2003).
 34.
Lantz, E., Blanchet, J.L., Furfaro, L. & Devaux, F. Multiimaging and Bayesian estimation for photon counting with EMCCDs. Monthly Notices of the Royal Astronomical Society 386, 2262–2270 (2008).
 35.
Tasca, D. S., Edgar, M. P., Izdebski, F., Buller, G. S. & Padgett, M. J. Optimizing the use of detector arrays for measuring intensity correlations of photon pairs. Physical Review A 88, 013816 (2013).
 36.
Andor iXon Ultra EMCCD Specifications, http://www.an\dor.com/pdfs/specifications/Andor_iXon_ULTRA_EMCCD_Specifications.pdf
 37.
Santos, I. F., AguirreGómez, J. G. & Pádua, S. Comparing quantum imaging with classical secondorder incoherent imaging. Physical Review A 77, 043832 (2008).
 38.
Saleh, B. E. A., Abouraddy, A. F., Sergienko, A. V. & Teich, M. C. Duality between partial coherence and partial entanglement. Physical Review A 62, 043816 (2000).
Acknowledgements
The authors would like to thank Nova Photonics, Inc. for providing equipment used in the experiment. This work was supported by the Air Force Office of Scientific Research grant FA95501210054.
Author information
Affiliations
Contributions
M.R. and J.W.F. conceived of the experiment; M.R. and H.D. developed the theory and experimental design, and M.R. performed the experiment; all authors analyzed the data and cowrote the paper.
Corresponding authors
Correspondence to Matthew Reichert or Jason W. Fleischer.
Ethics declarations
Competing Interests
The authors declare no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Received
Accepted
Published
DOI
Further reading

Spatially entangled photonpair generation using a partial spatially coherent pump beam
Physical Review A (2019)

Imaging with quantum states of light
Nature Reviews Physics (2019)

Quantum metasurface for multiphoton interference and state reconstruction
Science (2018)

Optimizing the signaltonoise ratio of biphoton distribution measurements
Physical Review A (2018)

Adaptive Quantum Optics with Spatially Entangled Photon Pairs
Physical Review Letters (2018)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.