Exact Compensation of Anomalous Trichromatisms Based on Matching of Color Discrimination Thresholds |
|
Rika Mochizuki ・ Tatsuya Nakamura ・ Jinhui Chao |
|
|
However, due to notorious absence of linearity, reproducibility and objectivity in human perception, it is generally hard to find rigorous and objective method. Current methods use either color contrast between symbols and background or approximation of color-blind model. Since our perception is not observable from outside, it is difficult to characterize the color-weakness of an observer exactly, therefore until now no criterion has been found to compensate his/her color defect. Besides, the properties of human color perception vary widely among different individuals. Thus, it is highly desirable to have a method taking into account of these properties so that compensation of colorweakness can be adapted to suit a particular observer.
In this paper, the authors propose a compensation algorithm for color-weakness based on a new, objective criterion to compare normal observers and color-weak observers, using Riemann geometric properties of color spaces.
To solve the above-mentioned problems, they focused on the relationship between the discrimination thresholds and the color differences in color space. The authors measured small color differences or discrimination thresholds (known from MacAdam ellipsoids) which is one of the few observable data in color perception and used them as individual characteristics. Such a discrimination ellipsoid defines local geometry at the test color, and a color space becomes a Riemann space with the thresholds as the Riemann metric. Then using Riemann geometry, the authors proposed a new criterion for color-weak compensation to match discrimination threshold between color-weak and color-normal observers. Applying this criterion, a color difference preserving map (called color-weak map) is obtained which maps the color space of the color-weak observer to that of color-normal observers, and simulate color-weak vision. Its inverse map is used to compensate so that the color-weak observer can see the same color stimuli as seen by a normal observer.
The proposed method was then applied to natural images using measured discrimination thresholds of colorweak and normal observers. Its performance is evaluated using SD test. This novel and effective approach is also expected to find more applications in other fields.
