Open AccessNear-infrared spectroscopy is feasible to discriminate hazelnut cultivars
Author(s) -
Elisabetta Stella,
Roberto Moscetti,
Letizia Carletti,
Giuseppina Menghini,
Francesco Fabrizi,
V. Cristofori,
Danilo Monarca,
Massimo Cecchini,
Riccardo Massantini
Publication year - 2013
Publication title -
journal of agricultural engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.3
H-Index - 18
eISSN - 2239-6268
pISSN - 1974-7071
DOI - 10.4081/jae.2013.(s1):e55
Subject(s) - cultivar , linear discriminant analysis , absorbance , spectroscopy , near infrared spectroscopy , mathematics , partial least squares regression , chemometrics , analytical chemistry (journal) , infrared spectroscopy , chemistry , horticulture , pattern recognition (psychology) , biological system , artificial intelligence , biology , statistics , computer science , chromatography , optics , physics , organic chemistry , quantum mechanics
The study demonstrated the feasibility of the near infrared (NIR) spectroscopy use for hazelnut-cultivar sorting. Hazelnut spectra were acquired from 600 fruit for each cultivar sample, two diffuse reflectance spectra were acquired from opposite sides of the same hazelnut. Spectral data were transformed into absorbance before the computations. A different variety of spectral pretreatments were applied to extract characteristics for the classification. An iterative Linear Discriminant Analysis (LDA) algorithm was used to select a relatively small set of variables to correctly classify samples. The optimal group of features selected for each test was analyzed using Partial Least Squares Discriminant Analysis (PLS-DA). The spectral region most frequently chosen was the 1980-2060 nm range, which corresponds to best differentiation performance for a total minimum error rate lower than 1.00%. This wavelength range is generally associated with stretching and bending of the N-H functional group of amino acids and proteins. The feasibility of using NIR Spectroscopy to distinguish different hazelnut cultivars was demonstrated.