Testing uniform colour spaces using colour differences of a wide colour gamut

Area of Science:

• Color science
• Visual perception
• Experimental data analysis

Background:

• Accurate color difference models are crucial for various industries.
• Existing models have limitations in predicting perceived color differences.
• A wide-gamut experimental dataset is needed to evaluate and improve these models.

Purpose of the Study:

• To evaluate the performance of seven leading color difference models using a new experimental dataset.
• To identify which models, with parametric optimization, best predict human visual assessment of color differences.
• To assess the impact of color difference magnitude on model performance.

Main Methods:

• Assembled an experimental dataset (WCG) with 416 sample pairs across 28 color centers and a wide color gamut.
• Tested seven color difference models: CIELAB, CIEDE2000, CAM16-UCS, DIN99d, OSAGP, ICTCP, and Jzazbz.
• Fitted color discrimination ellipses to compare color space uniformity and optimized model parameters (kL, kC, power factor).

Main Results:

• The kL-optimized CAM16-UCS, DIN99d, and OSAGP models demonstrated significantly better performance than other models.
• Parametric adjustments, including kL optimization, notably improved model fit to the experimental data.
• The magnitude of color difference was found to influence the visual assessment and model accuracy.

Conclusions:

• Optimized versions of CAM16-UCS, DIN99d, and OSAGP provide more accurate predictions of perceived color differences.
• Parametric tuning is essential for enhancing the performance of color difference models.
• The developed dataset and findings offer valuable insights for color difference modeling and applications.