Uncertainty Visualization and Color Vision Deficiency - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 30

Fig. 2.3 Limitation of recoloring techniques. a Original image with colors over most of the RGB
color space. b Simulated perception of a deuteranope. c Due to the dichromat's limited color gamut,
by trying to solve some color ambiguities, recoloring techniques may introduce new ones
temporal coherence, and can be easily integrated with existing visualization appli-
cations. This makes it suitable for use in interactive visualizations. Figure 2.2 shows
an example of image recoloring produced by this technique.
All recoloring techniques for dichromats share an inherent limitation: they define
mappings from a 3-D color space to a 2-D color gamut . Thus, such techniques tend
to become ineffective as the original image content spans the entire or most of the
3-D color space. In those situations, trying to solve some ambiguity by rearranging
colors on the dichromat's 2-D color space might introduce new ambiguities (Fig. 2.3 ).
Moreover, current recoloring techniques are restricted to the set of colors found in
each input image or video. Thus, mappings between pairs of colors in one image
or video may not be preserved in different ones (e.g., some cell structures may be
recolored in blue in one image, while appearing yellow in another).
2.2.1 Open Research Questions
In order to address the limitations discussed in the previous paragraph, we need to
consider the following open research questions:
Q1 How can one enhance visualizations by encoding additional information in order
to compensate for the reduced color gamut of dichromats? Or, in other words,
how can one lift the two-dimensional color gamut restriction?
Q2 Can the experiences learned from addressing the previous question also be
exploited to enhance visualizations for normal trichromats?
Q3 How can one obtain content-independent solutions that can be consistently used
over different images and videos?
Q4 Is it possible to satisfactorily extend these solutions to also represent natural
scenes, where colors have some associated meanings to the viewers?
We have started to explore some of these questions. In one initial effort, we have inves-
tigated augmenting colors with simple patterns to encode information for dichro-
mats [ 1 ]. Our results suggest that such combination can improve the performance