Exact computations were performed on an electronic computer to check the previous approximate formulas for the information content of photon beams. When no noise beam is present, the approximate formulas are found to be satisfactory for large incident-photon and pulse numbers. For small pulse numbers they give erroneous results. Thus 0.7 bits of information are obtained with an incident beam of 4–5 pulses, instead of the 17 pulses found from the approximate formulas. In the case of sample noise, the approximate formulas may become incorrect when the noise is of the same order of magnitude as the signal beam. This changes the discrimination step to 4.6 times the rms fluctuation of the pulse number at the lower level, when i = 0.7 bits of information are chosen as the minimum acceptable amount of information per macrocell. The exact computations confirm the previous result that the information per detector depends, to a good approximation, only on the ratio between the pulse-number step and the rms deviation of the pulse number of the background.
© 1969 Optical Society of AmericaFull Article | PDF Article
More Like This
J. Opt. Soc. Am. 55(11) 1446-1454 (1965)
J. Opt. Soc. Am. 58(7) 977-984 (1968)
J. Opt. Soc. Am. 57(1) 35-43 (1967)