Evaluating the accuracy of color vision deficiency simulation: Methodologies and a comparative analysis of current models

Color vision deficiency (CVD) simulation seeks to approximate the color perception of color-deficient observers (CDOs) for color-normal observers (CNOs). This technique provides a basis for developing color-enhancement technologies aimed at helping CDOs overcome challenges in color-dependent and visualization-related tasks. However, the accuracy of current CVD simulation models remains unclear, and an effective evaluation method is lacking. In this study, we propose an evaluation method based on the quantification of color vision tests for both CDOs and CNOs. Specifically, the accuracy of a simulation model is evaluated by comparing the test results of CDOs viewing original colors and CNOs viewing model-simulated colors. Our human-factor study demonstrates that the proposed method can effectively evaluate the accuracy of CVD simulation models. Specifically, the Machado and Yaguchi models significantly outperform the Yang model in simulating the color perception of CDOs, revealing the dominant influence of the stage theory in human color vision on simulation accuracy. Moreover, our findings indicate that the decline in color discrimination ability caused by CVD is not linearly correlated with the severity of the deficiency. Minor and moderate shifts in spectral sensitivity have very limited impacts on individuals’ actual color perception. These findings can contribute to the advancement of CVD simulation models and support their applications in enhancing the color vision of CDOs.