Comparison of digital photography and spectrometry for evaluating colour perception in humans and other trichromatic species

Abstract

Digital photography and spectrometry are widely used for colour measurement, but both methods have a number of advantages and disadvantages. Comparative studies can help determine the most appropriate method for quantifying animal colour perception, but few have attempted to compare them based on colour model conversion. Here we compare colour measurements from digital photography and spectrometry in a controlled standard experimental environment using the three-dimensional colour space model CIE L*a*b* which is designed to approximate colour perception in humans and assess the repeatability and agreement of the two methods. For digital photography, we first extracted RGB values from each colour patch and transferred these to L*a*b* values using colour model conversion. For spectrometry, we measured the spectral reflectance (SR) value and subsequently transferred SR values to L*a*b* values. Using a consensus of correlation analysis, intraclass correlation coefficients, concordance correlation coefficients, and Bland-Altman analysis, we found that although spectrometry showed a slightly higher repeatability than photography, both methods were highly repeatable and showed a strong agreement. Furthermore, we used Bland-Altman analysis to derive the limits of agreement, which can be used as criteria for identifying when photography and spectrometry could be as a suitable alternative for measuring colour perception in humans and other trichromatic species. We suggest that our workflow offers a practical and logical approach that could improve how we currently study colour perception in trichromats. Measuring colour efficiently and accurately is necessary for investigating the evolutionary biology of colour perception in animals. Digital photography and spectrometry are two methods widely used for colour measurement, but there are benefits and limitations to using either method. Comparative studies based on colour model conversion are therefore critical for helping researchers determine which method is most appropriate. Here we test the repeatability and agreement of the two measuring methods using standard colour patches, as a comparative case study of broader interest in measuring colour perception in humans and similar primates. Our results demonstrate that both methods are highly repeatable, and the two methods may be used interchangeably to measure colour perception in humans under experimental conditions.