Mitigation of Color Appearance Shifts in Augmented Reality Under Aligned Conditions

Abstract

This study investigates how geometric alignment affects color appearance in optical see-through (OST) augmented reality (AR). Building on our previous work, we conducted a color-matching experiment using a simulation setup that optically merges a display and printed background via a half mirror. Fifteen observers matched the test colors to either printed targets or printed targets with AR overlays across four lightness levels and six hues, with precise geometric registration maintained throughout the experiment. We analyzed the results using perceptual weighting functions and distance-based metrics. The perceptual weighting coefficients were consistently close to unity, indicating minimal discounting between the virtual and physical components. The color differences between targets and the mean of the matched colors were generally small, and the matched colors clustered closely around their respective targets. The observed mean color difference from the mean (MCDM) was comparable to the previously reported observer variability, while the mean color difference from the target (MCDT) was approximately 1.2 times the MCDM. Although lightness and hue had statistically significant effects on matches, their absolute impacts remained minor. These findings demonstrate that accurate geometric alignment substantially mitigates color appearance shifts in OST AR, even under varying stimulus conditions.