An algorithm for generating color scales for both categorical and ordinal coding

Abstract

Previous research has shown multihue scales to be well-suited to code categorical features and shown lightness scales to be well-suited to code ordinal quantities. We introduce an algorithm, Motley, that produces color scales varying in both hue and lightness, intended to be effective for both categorical and ordinal coding, allowing users to determine both absolute and relative quantities efficiently and accurately. The algorithm first determines the lightnesses of scale colors to maximize perceived lightness differences and establish the lightness ordering, generating separate search spaces for each scale position. It then selects hues by heuristic search to maximize the discriminability of the scale. It produces scales that are ordered with respect to lightness but unordered with respect to hue and thus more discriminable than typical multihue lightness scales. In an experimental evaluation on human subjects, Motley's scales enabled accurate judgments of relative quantity, with response times superior to unordered multihue scales and comparable to ordered lightness scales, and enabled accuracy and speed of judgments of absolute quantity superior to lightness scales and comparable to multihue scales.

Second-order mappings in grapheme–color synesthesia

“…Take map reading as an example. Defining regions by hue typically allows for faster and more accurate judgments of categorical distinctions than does defining them by luminance, while luminance scales afford advantages for judgments about relative quantity or continuous magnitudes (Breslow, Trafton, McCurry, & Ratwani, 2010). This likely reflects the fact that variations in luminance have an underlying continuity, from dark to light, while hues are perceived categorically.…”

Pseudocolours for thermography—Multi-segments colour scale

“…Literature has demonstrated that luminance-based colour scale is best for direct comparison between spots [18,19], but poor for value reading [14,15]. Whereas hue-based colour scale is good for value reading [14,15,20], but poor in any other else. For diverging colour scale, its specialty is to make peaks noticeable [10], a strength neither luminance nor hue based colour scale possesses.…”

“…If the colours of a hue-based colour scale are distinctive enough among themselves, the user identifies value key better than luminance-based colour scale [15]. Furthermore, it is possible to design a hue-based colour scale that is distinctive among colours in the scale and ordered by luminance, as illustrated by Breslow et al in their Motley algorithm [15].…”

“…The second and the third are diverging colour scales, under red/blue and orange/purple colour schemes respectively, with white in the middle, using only three colour handles. The fourth colour scale is a type of rainbow colour scale commonly used in thermography, and the last is a Motley scale, proposed by Breslow et al [15].…”

Visualization enhancement of segmented images using genetic algorithm

“…Carter and Huertas in [15] studied the methods for displaying small visual targets for ultra-large color difference using several color difference metrics and color spaces. Breslow et al in [9] introduced an algorithm called Motley, generating color scales predesignated for nominal and ordinal coding, in which k search spaces are generated based on lightness and saturation and then single color from each search space that maximizes the color discriminability is chosen. Bianco and Citrolo in [10] tested four different heuristics for creation of high contrast color sets under single and multiple illuminants.…”