Measuring and Modeling the Feature Detection Threshold Functions of Colormaps

Area of Science:

• Scientific Visualization
• Color Science

Background:

• Pseudocoloring is essential in scientific visualization, mapping scalar field values to colors via colormaps.
• Colormap uniformity is traditionally assessed using uniform color spaces like CIELAB.
• Existing methods for colormap evaluation lack direct assessment of feature detection.

Purpose of the Study:

• To introduce a novel method for evaluating colormap uniformity based on feature detection thresholds.
• To test the efficacy of uniform color spaces (CIELAB) in predicting human perception of features in pseudocolored data.
• To investigate alternative color space models for improved colormap design.

Main Methods:

• Developed a new method to evaluate the feature detection threshold function across colormaps.
• Utilized crowdsourced studies to directly assess nine colormaps for three feature sizes.
• Compared perceptual models based on uniform CIELAB, reduced-weight CIELAB, and luminance-only predictions.

Main Results:

• Crowdsourced evaluations rejected the hypothesis that CIELAB accurately models feature detection thresholds.
• The hypothesis that luminance alone predicts feature detection was also rejected.
• A modified CIELAB model with reduced weights on chromatic components improved prediction accuracy for small feature detection.

Conclusions:

• Uniform color spaces like CIELAB are insufficient for predicting feature detection in pseudocolored data.
• A modified CIELAB color space, emphasizing luminance and de-emphasizing chromaticity, offers a more accurate perceptual model.
• The developed evaluation method and modified CIELAB provide valuable tools for colormap design and assessment.