Determination of grated hard cheese adulteration by digital image analysis and multivariate analysis

Abstract

Hard-cheeses are foods with high commercial value, thus fraudulent practices are frequent and difficult to detect, which reduce hard-cheese quality and consequently, decreases nutritional value. This work presents a novel and feasible tool to identify the adulteration of grated-cheese with higher levels of additives than those permitted (cellulose and silicon dioxide), and foreign substances used to increase weight and volume (wheat-flour, wheat-semolina and sawdust). The methodology is based on the use of colour histograms (obtained from digital images) and multivariate classification analysis. Satisfactory results were obtained with support vector machines with average values of accuracy and sensitivity above 82%, as adulterated samples were discriminated from original (unadulterated) samples. The proposed screening methodology represents a fast, simple, reliable and affordable way to perform quality control of grated-cheeses to avoid food fraud and nutritional detriments of worldwide consumption foods such as hard-cheeses.