Aplicação de imagens digitais e técnicas quimiométricas para detecção de adulteração em mel.

Abstract

Honey is a natural food composed of enzymes, amino acids, organic acids, carotenoids, vitamins, minerals, aromatic substances, and carbohydrates, its main component. Given its high amount of phenolic compounds, such as flavonoids and phenolic acids, which act as antioxidant agents, honey stands as a highly appreciated product. However, due to the product scarcity, cases of adulteration through the addition of commercial sugars, commercial glucose, molasses, inverted sucrose solution or syrup are increasingly common in the market. This study aimed to develop a low-cost methodology with the use of digital images and multivariate analysis to identify adulterations in honey. In this study six (6) honey samples and nine (9) mixture of these were used. The adulterations were made with a brown sugar solution in the following concentrations 0.1 g/ml, 0.15 g/ml and 0.2 g/ml. Aiming to analyse the purity of the honey samples, three control tests to check adulterations were performed:Lund, Fiehe, and Lugol with analysis performed in triplicates. Digital images were acquired with the LifeCam Cinema HD webcam. The ImagensGui interface coupled with the Matlab R2010a software was used to capture the images.PCA and SIMCA supervised pattern recognition analyzes were made on TheUnscrambler 9.7. According to results obtained from the physical-chemical analyzes, composed by the three testsLund, Fiehe, and Lugol of adulterant honey control, in the samples with pure honey, we could infer that these were not adulterated. From the exploratory study, based on the application of principal component analysis (PCA), in the histogram colors obtained from the RGB and HLS systems, we observed that there was discrimination among the 15 samples of pure honey (Uruçu and Italian species and mixtures of the Uruçu species) and the 45 adulterated honeys. The SIMCA model presented I and II errors type, however the result obtained by this model can be improved with the increase of numbers of associated samples.