In Part I of this series of two papers, it was explicitly postulated that the internal noise of the detector was large compared with the noise produced by steady radiation falling on the detector (radiation noise). In this Part II both kinds of noise are taken into account. In a sense, therefore, the present paper includes Part I. But in another sense the scope of the present paper is more restricted because the greater complexity with respect to the sources of noise has made it desirable to use throughout the simplifying assumption that was used only in Sec. 3 of Part I.
The emphasis is given not to the information capacity itself, but to the information efficiency, which is the information capacity divided by the mean power of the beam.
A parameter β is introduced that is a measure of the relative amount of the radiation noise and the internal detector noise when the mean signal power is chosen optimally. When β is zero, only radiation noise is present, and when β is infinite, only the internal noise is present. In Sec. 6 the information efficiency I is derived as a function of the parameter β. The two cases represented by the extreme values of β are discussed in Secs. 4 and 5. In Sec. 4 it is found that with symmetrical signal modulation, the maximum possible information efficiency is equal to the detective quantum efficiency of the detector. When, however, nonsymmetrical modulation is permitted, then the maximum possible information efficiency may be greater and is equal to QD multiplied by log2(1/p), where p is the probability that the gate is open.
The information efficiency in bits per photon is calculated in Sec. 8, for five different kinds of detectors: a multiplier phototube, human vision, photographic films, an image orthicon, and heat detectors. An error was made in Part I in the calculation of the information efficiency of the 1P21 phototube; the error is corrected in Sec. 8.
© 1962 Optical Society of America
Equations on this page are rendered with MathJax. Learn more.