Abstract
The effects of random errors in measurements of diffuse reflectance made on an abridged spectrophotometer are evaluated by means of Monte Carlo computer-simulation techniques. These techniques allow the analysis of nonadditive or non-Gaussian errors such as signal-dependent noise and quantization of the photometric scale. The effects of measurement error are evaluated in terms of computed color difference, with results based on the largest color difference errors that occur in 500 measurements. Analysis of representative colors demonstrates that neutrals are at least as sensitive to measurement errors as any of the colors considered. It is demonstrated that relative errors with a standard deviation near 0.1% result in negligible color difference errors. Colors of high luminous reflectance are shown to be more susceptible to relative errors than are dark colors. Analysis of linear and logarithmic quantization schemes shows that logarithmic quantization uses considerably fewer quantization levels than a linear quantization scheme having the same color difference error. The results presented in this paper are applicable as guidelines in the design of abridged spectrophotometers.
© 1977 Optical Society of America
Full Article | PDF ArticleMore Like This
A. R. Robertson
J. Opt. Soc. Am. 57(5) 691-698 (1967)
A. R. Macdonald and G. P. Bentley
J. Opt. Soc. Am. 45(5) 366-372 (1955)
Thorne Shipley and Gordon L. Walker
J. Opt. Soc. Am. 46(12) 1052-1060 (1956)