Synthesis of the Perceptionally Linear Color Space Using Machine Learning Methods

Abstract

Currently, there is no color model that exactly matches the color perception of a person. It would be desirable for automatic processing of sensor data. The experiment was conducted to create such a model. The test samples specifications are defined in the article, taking into account the known parameters and characteristics of the visual analyzer. The experiment results are statistically processed using machine learning methods. Perceptional visibility saturation and hue thresholds in for given brightness, saturation, and hue values are calculated and represented as ellipses in the luminance plane. For the indicated ellipses, isocontrast space in equal brightness slices is calculated by processing the parameters of the ellipses obtained.