Image Processing Reference
In-Depth Information
As a consequence of the existence of three types of photoreceptors, normal color
vision spans a 3-D color space. The color gamut of a dichromat, on the other hand,
is only two-dimensional and can be represented by a surface patch in the same 3-D
color space. Such a reduced gamut is the cause of the ambiguity experienced by
dichromats: many different colors are perceived as the same, when projected onto
such patches. For anomalous trichromats, the color gamut falls in between these two
extremes, moving towards the gamut of a dichromat as the degree of severity of the
anomaly increases. For spectral shifts of approximately 20 nm, the perception of an
anomalous trichromat becomes similar to the perception of a dichromat [
6
,
11
].
Currently, there is no clinical or surgical treatment for color vision deficiency.
Given the relevance of the problem, a few techniques have been recently proposed
to simulate the perception of individuals with CVD [
2
,
4
,
7
], and to enhance image
contrast through recoloring [
3
,
5
,
9
]. Next, I briefly discuss these techniques, showing
how they can assist the design of more inclusive visualization experiences, but also
discussing their inherent limitations, which calls for more research.
2.2 Tools for More Inclusive Visualizations
The first step to produce more effective visualizations for individuals with CVD is to
understand their perceptual limitations. Meyer and Greenberg [
7
], and Brettel et al. [
2
]
presented simulation techniques for the color perception of dichromats. Machado
et al. [
4
] introduced a physiologically-based model that supports the simulation of
dichromatic as well as anomalous trichromatic vision (with arbitrary degrees of
severity) in a unified way. This simulation model works in real time and can be
quickly incorporated into existing systems. Thus, a visualization designer can get
instantaneous feedback on how it would be perceived by individuals with CVD
(Fig.
2.1
). Such knowledge allows the designer to refine the visualization, making
it more effective for wider audiences. While simulation models help to increase the
awareness of the perceptual limitations due to CVD, they do not directly help the
affected individuals to recover the loss of color contrast.
To address the problem of enhancing color contrast, a few automatic image-
recoloring techniques for dichromats have been proposed in recent years [
3
,
5
,
9
].
Essentially, all these approaches define ways of mapping the colors in the original
image to a new set of colors in the dichromat's gamut. This is done while trying to
preserve the perceptual color differences among all pairs of colors in the original
image. Rasche et al. [
9
] proposed an approach that uses a constrained multivariate
optimization procedure applied to a reduced set of quantized colors. The resulting
algorithm does not scale well with size of the input image and the number of quan-
tized colors, and is not applicable to interactive applications. Kuhn et al. [
3
] present
a solution based on a mass-spring optimization that achieves interactive rates, and
tries to preserve the naturalness of the original images (i.e., preserve the colors that
can already be perceived by dichromats). More recently, Machado and Oliveira [
5
]
introduced a projection-based recoloring approach that works in real time, enforces
Search WWH ::
Custom Search