This work describes a new procedure for classification of mineral waters based on digital images acquired by smartphones. Commercial waters from eight mineral springs plus distilled water and tap water were combined with eriochrome T black or murexide and transferred to a cuvette, which was positioned into a light controlled chamber. RGB (Red, Blue and Green) measurements of cuvette images were acquired in real time, using a free smartphone app, and employed as variables for the exploratory analysis. 2D data dispersion along component B for murexide (x axis) and component R for eriochrome T black (y axis) provides the clear visualization of clusters using the raw variables. Hierarchical cluster analysis (HCA) applied to this data confirmed the efficient discrimination of samples providing the characterization of nine clusters for the ten classes of water investigated. The classification of samples based on a k-nearest neighbors (k-NN) modelled to the efficiency rate of 100% for 8 classes and of 94.4% and 50% for the remaining classes, respectively, indicating the adequate performance of the proposed strategy. Considering the facilities to acquire the data, such as low cost instrumentation and reagents, and the rapidity of the procedures, this alternative may be applied for verification of commercial water adulteration.
Keywords:
smartphone; mineral waters; colorimetry; exploratory analysis
