[Paper Review] Accessible Color Sequences for Data Visualization

Matthew A. Petroff|arXiv (Cornell University)|Jul 5, 2021

Color perception and designPsychology20 citations

TL;DR

This paper proposes data-driven, accessible color sequences for categorical data visualization that balance aesthetic preference with accessibility for color-vision-deficient users, grayscale printing, and high-contrast backgrounds. Using crowdsourced survey data and machine learning, the author develops optimized six-, eight-, and ten-color sequences that outperform common defaults in accessibility while maintaining visual appeal.

ABSTRACT

Color sequences, ordered sets of colors for data visualization, that balance aesthetics with accessibility considerations are presented. In order to model aesthetic preference, data were collected with an online survey, and the results were used to train a machine-learning model. To ensure accessibility, this model was combined with minimum-perceptual-distance constraints, including for simulated color-vision deficiencies, as well as with minimum-lightness-distance constraints for grayscale printing, maximum-lightness constraints for maintaining contrast with a white background, and scores from a color-saliency model for ease of use of the colors in verbal and written descriptions. Optimal color sequences containing six, eight, and ten colors were generated using the data-driven aesthetic-preference model and accessibility constraints. Due to the balance of aesthetics and accessibility considerations, the resulting color sequences can serve as reasonable defaults in data-plotting codes, e.g., for use in scatter plots and line plots.

Motivation & Objective

• To address the lack of accessible, aesthetically pleasing color sequences for categorical data visualization in scientific plotting.
• To develop color sequences that are perceptually distinct for individuals with color-vision deficiencies, including deuteranomaly, protanomaly, and other common deficiencies.
• To ensure legibility in grayscale printing by enforcing minimum lightness differences between colors.
• To maintain high contrast with white backgrounds by limiting extreme lightness values.
• To improve verbal and written describability of colors by incorporating color-saliency scores.

Proposed method

• Crowdsourced aesthetic preference data were collected via an online survey to train a machine-learning model for color-sequence preference.
• A multi-objective optimization framework combined the aesthetic preference model with perceptual distance constraints for color-vision-deficient viewers.
• Minimum lightness-distance constraints were applied to ensure readability in grayscale and on black-and-white prints.
• Maximum lightness constraints were enforced to preserve contrast with white backgrounds and avoid visual dominance of dark or light colors.
• Color-saliency scores from a language-dependent model were integrated to improve the ease of naming and describing colors in text.
• Optimal color sequences were generated by balancing these competing constraints using a data-driven, multi-criteria optimization approach.

Experimental results

Research questions

• RQ1 How can aesthetic preferences for color sequences be quantitatively modeled using real user data?
• RQ2 To what extent do perceptual distance constraints for color-vision deficiencies improve the accessibility of color sequences?
• RQ3 How do lightness constraints affect the usability of color sequences in grayscale and high-contrast environments?
• RQ4 Can color-saliency scores enhance the describability of colors in written and verbal communication?
• RQ5 How do the resulting color sequences compare to existing default sequences in terms of accessibility and visual appeal?

Key findings

• The proposed color sequences—available in six-, eight-, and ten-color versions—achieve a balanced trade-off between aesthetic preference and accessibility constraints.
• The sequences significantly outperform commonly used default color palettes in accessibility for individuals with red–green color-vision deficiencies.
• Aesthetic preference was found to correlate with higher mean chroma and lower maximum chroma, suggesting a preference for moderately saturated colors.
• No significant correlation was found between aesthetic preference and color-saliency scores, indicating that saliency enhances describability without compromising visual appeal.
• The sequences are designed as general-purpose defaults for scatter plots and line plots, with only the first few colors used when fewer categories are present.
• The full dataset, trained models, and analysis code are publicly available to support adoption and extension in data-visualization tools.