Discrimination of olives according to fruit quality using Fourier transform Raman spectroscopy and pattern recognition techniques.

Abstract

Fourier transform Raman spectroscopy combined with pattern recognition has been used to discriminate olives of different qualities. They included samples of sound olives, olives with frostbite, olives that have been collected from the ground, fermented olives, and olive samples with diseases. Milled olives were measured in a dedicated sample cup, which was rotated during spectrum acquisition. A preliminary study of the data set structure was performed using hierarchical cluster analysis and principal component analysis. Two supervised pattern recognition techniques, K-nearest neighbors and soft independent modeling of class analogy (SIMCA), were tested using a "leave-a-fourth-out" cross-validation procedure. SIMCA provided the best results, with prediction abilities of 95% for sound, 93% for frostbite, 96% for ground, and 92% for fermented olives. The olive samples with diseases (too few to define a class) were included in the validation and recognized as not belonging to any class. None of the damaged olive samples was wrongly predicted to the class of sound olives. With this approach a selection of sound olives for the production of high-quality virgin olive oil can be achieved.