We show experimentally that parametric optical preamplification greatly improves the signal-to-noise ratio of an image if the detector has a poor quantum efficiency and/or a great level of readout noise. Results are fully consistent with the theory of quantum-noise-limited amplification.
We demonstrate experimentally that a phase-sensitive optical parametric amplifier used to amplify the incoming optical signal on a lossy photodiode improves the noise figure of detection. At high parametric gains this noise figure tends to 0 dB corresponding to a detection quantum efficiency approaching 100%. This result contrasts with the performance of preamplification by a classical laserlike amplifier in which the noise figure tends to 3 dB giving a saturation of the detection quantum efficiency at 50%.
The operational theory of homodyne detection by nonideal detectors, used in quantum tomography, was recently modified in order to incorporate the preamplification (before homodyne detection) of the input signal, thus enabling one to beat the handicap of the lower than 0.5 detector efficiency. In the present work we set expressions for the Mandel Q parameter and the signal-to-noise ratio in terms of the operational (measured) moments of the preamplified homodyne detection formalism. These quantities furnish important information on the statistical properties of the input signal field and the photocounts at the output. We illustrate the theory by considering several kinds of fields (for the input signal) and determine the effects of the preamplification on the output signal-to-noise ratios. Here we essentially verify that (i) the preamplification shifts the statistics towards the super-Poissonian limit, without jeopardizing the capacity of reconstructing a sub-Poissionan input signal, and (ii) the preamplification is more effective, i.e., the rate of increase of the signal-to-noise ratio of the output photocount is larger for low-efficiency detectors than for ideal ones.
We investigate amplification of optical images by means of a traveling-wave optical parametric amplifier. As shown recently by Kolobov and Lugiato [Phys. Rev. A 52, 4930 (1995)] for a cavity-based geometry, such a scheme can amplify images, without deteriorating their sign-to-noise ratio, thus working as a noiseless amplifier. Here we consider a configuration without cavity, which is more realistic for a possible experimental realization. We investigate in detail the quantum fluctuations of the amplifier and formulate criteria for its noiseless performance. The spatial resolution power, which guarantees noiseless amplification is estimated. We demonstrate how the optimum phase matching of a phase-sensitive wavefront of the image can improve the noise performance of the amplifier and bring it to the ultimate value achievable under given physical conditions. We discuss the effect of improvement of the signal-to-noise ratio in the case of observation of both the input and the amplified image, with an ideal photodetector of small area (i.e., much less than the coherence area of the amplifier). This improvement is due to the fact that the signal-to-noise ratio in the input is deteriorated by the observation with the photodetector of small area.
Classical information theory can be used to quantify the resolution performance of optical imaging systems. When an optical parametric amplifier (OPA) operated as a phase-sensitive amplifier (PSA) in the transverse spatial domain is used for point source imaging, the angular resolution improvement can approach the de Broglie resolution (i.e. Heisenberg limit). In this paper, classical information theory is employed to quantify the signal-to-noise ratio (SNR) improvement for both an ideal and a realistic multimode PSA applied to the problem of sub-Rayleigh imaging. When only considering the noise originating from the detector, the SNR improvement is found to scale quadratically as a function of the PSA gain, in the limit of noise power comparable to signal power. Differences in performance of an ideal PSA and a realistic PSA are discussed.