A study into the use of reflecting tiles to determine spectrophotometric errors and the corresponding correction values was accomplished. A method was developed to calculate the wavelength scale accuracy (out to a wavelength error of 5nm) using the inflection points of reflecting tiles' spectrophotometric curves. Using simulated error data, wavelength accuracy was determinable at seven distinct wavelengths over the range of 440 nm to 705 nm within an accuracy of 0.2 nm. Three methods, the Matrix, Selection, and Non-Linear, were developed to increase the accuracy of spectrophotometric data. Each method determined and corrected spectrophotometric data differently. The Matrix method was based upon the use of multiple linear regression equations; the Selection method determined and used averages of the different errors; and the Non-Linear method identified and corrected errors based on using non-linear equations. The Selection method was found to be the most effective in correcting simulated spectrophotometric data. The Matrix method was also effective, IV however the determination of a specific error was not as accurate as the other two methods. The Selection method was the simplest to use, but it did not identify the existence of non-linear errors. The Non-Linear method was labor intensive and required very accurate data input, but with proper usage it showed exactly how a spectrophotometer was performing at each measured wavelength. With average (delta)E*ab errors from simulated data of 8.40 (using IIluminant D-65, 2 Observer, wavelength interval of 10 nm, and wave length range from 400 to 700 nm), the Matrix, Selection, and Non-Linear methods corrected to average (delta)E'ab's of 0.06, 0.01, and 0.48, respectively. A limited determination of the effectiveness of each method on actual data was performed. This was accomplished by measuring a set of tiles (from 400nm to 700nm at 10 nm increments) once on two separate instruments. The instruments were the Pacific Scientific Spectrogard and the Applied Color Systems Spectro-Sensor n. The wavelength accuracy of each instrument was measured using the method described in this thesis and then each method was used to identify and correct for the spectrophotometric errors. The Pacific Scientific had an average (delta)E*ab error of 0.81 and the Matrix, Selection, and Non-Linear methods corrected to average (delta)E*ab errors of 0.31, 0.36, and 0.37, respectively. The ACS had an average (delta)E*ab error of 0.43 and the Matrix and Selection each corrected to average (delta)E*ab errors of 0.23 and 0.39. The Non-Linear method failed for the ACS in that a worse average (delta)E*ab error of 0.79 was determined.
Library of Congress Subject Headings
Spectrophotometer--Technological innovations--Evaluation; Colorimetry--Instruments--Evaluation
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Petersen, Kelvin H., "An Investigation of new methods to improve the accuracy of modern color spectrophotometers" (1989). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus