Digital images-based chemometrics-assisted methodology as a sustainable strategy for blond beers quality control

Abstract

A simple, economical and reliable methodology was developed based on digital images captured with a smartphone for the multiparametric quantification in blonde beers. For this objective, PLS was used to construct multivariate calibration models using grayscale color histograms, RGB (red-green-blue) channels, HSI (hue-saturation-intensity) and their combinations as analytical information. Through these calibrations, the alcohol content, pH, total acidity, bitterness and polyphenols have been successfully quantified with a relative prediction error between 3.90% and 12.60%, demonstrating satisfactory results. In addition, the data were processed using PLS-DA to develop a multivariate classification model that allowed discriminating beer samples according to their production method (industrial or craft) and between alcoholic and non-alcoholic beers. The error rates were of 9% and 2%, respectively. These methods hold promise for improving quality control of blond beers.