Abstract

All interferometric imaging is based on a direct application of the Van Cittert-Zernike theorem. Because of the existence of cooled solid-state photodetectors with hardly any readout noise, the sensitivity of optical interferometric imaging is limited almost exclusively by the Poissonian shot noise of photoelectron counting. We shall derive the signal-to-noise ratios (SNRs) in the reconstructed image, both for ideal imaging, which is possible for space-based arrays, and for bispectral (or closure-phase) imaging, which is necessary for ground-based arrays. We shall show¹ that the beam combination geometry plays no essential role in determining SNR. Also, at high photon rates, the image SNR is the same for ideal arrays as it is for ground-based arrays. However, the sensitivity of ground-based imaging is severely degraded for faint objects that produce less than about 0.3 photon per spatial coherence area per coherence time. I shall compare bispectral imaging with self-calibration and discuss the tradeoff between sensitivity and resolution.

© 1990 Optical Society of America