Systematic comparison of head mounted display colorimetric performance using various color characterization models

Abstract

The advancement of virtual reality in recent times has seen unprecedented applications in the scientific sphere. This work focuses on the colorimetric characterization of head mounted displays for psychophysical experiments for the study of color perception. Using a head mounted display to present stimuli to observers requires a full characterization of the display to ensure that the correct color is presented. In this paper, a simulation is done to mimic a practical display with color channel interactions and characterization of simulated data is done using the following models: gain offset gamma model, gain offset gamma offset model, gain gamma offset model, piecewise linear assuming chromaticity constancy model, piecewise linear model assuming variation in chromaticity, look-up table model, polynomial regression model, and an artificial neural network model. an analysis showed that the polynomial regression, artificial neural network, and look-up table models were substantially better than other models in predicting a set of rgb values, which can be passed as input to a head mounted display to output desired target xyz values. both the look-up table and polynomial regression models could achieve a just noticeable difference between the actual input and predicted output color of less than 1. the gain offset gamma, gain offset gamma offset, and gain gamma offset models were not effective in colorimetric characterization, performing badly for simulations as they do not incorporate color channel interactions. the gain offset gamma model was the best among these three models and the lowest just noticeable difference it could achieve was over 13, clearly too high for color science experiments.