Purchase this article with an account.
Shankaran Ramaswamy; Does computerized simulation of dichromatic color appearance predict VDT color identification error rate? . Journal of Vision 2009;9(14):74. doi: 10.1167/9.14.74.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Introduction: Several algorithms are available to transform colored digital images into simulated dichromatic color perception. These algorithms can be very illustrative of the problems dichromats experience in discriminating colors. We were interested in whether this type of transformation could provide a more quantitative account of error rates in identifying colors displayed on a computer monitor.
Methods: The task required observers to identify the color of small rectangles displayed in a black background. There were two different sets of eight colors in each set. The number of errors for each color was recorded. Four deuteranopes and five protanopes participated. Color differences were calculated using normal trichromatic and dichromatic values. The dichromatic color differences were calculated using the procedure developed by Brettel et al. (1997).
Results: There was a significant inverse linear correlation between the dichromatic error rates and the color differences calculated in the normal trichromatic color space for each color set. [r = −0.47 (p=0.034) for Set 1 and r= −0.49 (p=0.008) for Set 2]. A nonlinear fit [polynomial inverse first order: y=m+(c/x)] provided a better fit of the data for the second set of colors (r=0.45 p=0.001) but not for the first set (r=0.70 p=0.001). The correlations between the error rates and the dichromatic color differences were no better than the correlations using colour differences calculated in trichromatic color space.
Conclusions: Correlations between the error rate in identifying colors for dichromats and the color differences were moderate and nearly identical for color differences calculated based on color-normal vision or dichromatic vision. This suggests that it may be sufficient to calculate the color difference only in color-normal space in order to determine whether the colors will be confused by a person with a congenital color vision defect. Although algorithms are useful in illustrating color discrimination problems experienced by dichromats, they may not offer any advantage over typical trichromatic color spaces in predicting performance in color identification. The lack of any advantage may be due to how dichromats use brightness information to identify colors.
This PDF is available to Subscribers Only