Perceptual color similarity measurement

Abstract

Color perception remains a difficult and little understood problem, where the perceived color is generally subjective to the viewer such as human eye or a digital camera. In the area of vision there are a number of applications where measuring the similarity between several colors is required. However, existing models for measuring the color similarity use complex mathematical expressions and thus are slow and difficult to compute, especially on embedded systems where available number of instructions and mathematical capabilities can be quite limited. In many cases, the mathematical complexity of existing color similarity models makes them inadequate for applications running in real-time. In this work a new model for measuring the perceived similarity is proposed. It requires a minimal set of mathematical calculations and its formula can be computed easily on an embedded system, even when running in real-time. The proposed model is the result of theoretical and experimental studies that were made by the authors of this document with the purpose of finding a better alternative for measuring the color similarity on embedded systems. This work also presents a series of experiments comparing the existing perceptual color similarity models with the proposed model of this work, where the human eye can judge the result visually, that is, in perceptual terms.