Abstract
Using the tools of Fourier analysis, a sampling requirement is derived that assures that sufficient information is contained within the samples of a distribution to calculate accurately geometric moments of that distribution. The derivation follows the standard textbook derivation of the Whittaker-Shannon sampling theorem, which is used for reconstruction, but further insight leads to a coarser minimum sampling interval for moment determination. The need for fewer samples to determine moments agrees with intuition since less information should be required to determine a characteristic of a distribution compared with that required to construct the distribution. A formula for calculation of the moments from these samples is also derived. A numerical analysis is performed to quantify the accuracy of the calculated first moment for practical nonideal sampling conditions. The theory is applied to a high speed laser beam position detector, which uses the normalized first moment to measure raster line positional accuracy in a laser printer. The effects of the laser irradiance profile, sampling aperture, number of samples acquired, quantization, and noise are taken into account.
© 1990 Optical Society of America
Full Article | PDF ArticleMore Like This
Stephen K. Park and Robert A. Schowengerdt
Appl. Opt. 21(17) 3142-3151 (1982)
Noah Bareket
Appl. Opt. 18(17) 3064-3069 (1979)
A. VanderLugt
Appl. Opt. 29(23) 3352-3361 (1990)