The physics of ghost imaging

Ghost images are obtained by correlating the output of a single-pixel (bucket) photodetector-which collects light that has been transmitted through or reflected from an object-with the output from a high spatial-resolution scanning photodetector or photodetector array whose illumination has not interacted with that object. The term ghost image is apt because neither detector's output alone can yield an image: the bucket detector has no spatial resolution, while the high spatial-resolution detector has not viewed the object. The first ghost imaging experiment relied on the entangled signal and idler outputs from a spontaneous parametric downconverter, and hence the image was interpreted as a quantum phenomenon. Subsequent theory and experiments showed, however, that classical correlations can be used to form ghost images. For example, ghost images can be formed with pseudothermal light, for which quantum mechanics is not required to characterize its photodetection statistics. This paper presents an overview of the physics of ghost imaging. It clarifies and unites two disparate interpretations of pseudothermal ghost imaging-two-photon interference and classical intensity-fluctuation correlations-that had previously been thought to be conflicting. It also reviews recent work on ghost imaging in reflection, ghost imaging through atmospheric turbulence, computational ghost imaging, and two-color ghost imaging.