Alfalfa quality detection by means of VIS-NIR optical fiber reflection spectroscopy

A first approach study for the classification of alfalfa (medicago sativa) quality has been performed by means of VISNIR optical fiber reflection spectroscopy. Reflection spectral data has been obtained from alfalfa samples comprising six different qualities. Obtained data has been classified and organized to feed supervised self-learning algorithms. Neural networks have been used in order to differentiate the quality level of the samples. Obtained results permit to validate the proposed approach with 72% of the samples properly classified. In addition, proposed solution was implemented in a low cost automated detection prototype suitable to be used by non-qualified operators. Obtained equipment consist of a first step towards its utilization in quality monitoring and classification of many other products in the agrifood field.