Assessment of the authenticity of coconut water (Cocos nucifera L.) samples using digital images and chemometric techniques

Abstract

This study proposes using digital colorimetry combined with unsupervised pattern recognition techniques to obtain a molecular fingerprint profile that allows detection and identification of the non-destructive authenticity of coconut water samples. It also intends to classify the samples sold as in nature, adulterated, or industrialized. The samples were purchased at street markets and local stores in the state of Bahia, northeastern Brazil. The digital images were obtained through direct analysis without pre-treatment of the samples. Then, the combination values of color histograms in RGB channels were extracted using Chemostat software. Principal component analysis and hierarchical clustering contributed to the classification of the samples. It was possible to prove that digital colorimetry is a useful tool that allows confirming the authenticity of foods quickly and at a low cost. It can contribute to the inspection by regulatory agencies, in addition to following the principles of green analytical chemistry.