It is shown that, for an arbitrary discrete process embedded in independent additive discrete noise, the classical binary detection problem using a likelihood-ratio test reduces to a simple comparison of the number of events with a single threshold. Only a weak condition on the noise distribution is required. Our results are appropriate for the analysis of photocounting optical communications and photocounting radar systems as well as neural counting in auditory psychophysics. We specifically apply our method to a signal-detection theory model of the human visual system and draw a comparison to the analysis of Hecht, Shlaer, and Pirenne [J. Gen. Physiol. 25, 819 (1942)].
Malvin Carl Teich, Paul R. Prucnal, and Giovanni Vannucci, "Optimum photon detection with a simple counting processor," Opt. Lett. 1, 208-210 (1977)