Hyperspectral imaging for determination of some quality parameters for olive oil

The analysis of the quality of a virgin olive oil involves the determination of a series of chemical indexes and organoleptic characteristics. In this work we propose an online prediction model for three chemical indexes: acidity, peroxide index and humidity content, based on an hyperspectral artificial vision system. Two methods have been developed for the construction of the model: (1) partial least squares regression (PLS) using all the captured spectral components, and (2) partial least squares regression over a subset of the components obtained applying a genetic algorithm (GA-PLS). The design and validation was carried out using olive oil samples from different seasons analysed by a renowned laboratory.