An Engineering Model for Color Discriminability as a Function of Size

This work describes a first step towards the creation of an engineering model for color discriminability as a function of size. Our approach is to non-uniformly scale CIELAB using data from crowd-sourced experiments, such as those run on Amazon Turk. In such experiments, the inevitable variations in viewing condi-tions reflect the environment many applications must run in. Our intent is to define discriminability in a way that is robust, on av-erage, to these conditions. We make no claim that our model de-scribes color perception with a high degree of precision. Our goal is to create a useful model for design applications where it is im-portant to make colors distinct, but for which a small set of highly distinct colors is inadequate