Open AccessQualitative analysis for sweetness classification of longan by near infrared hyperspectral imaging
Author(s) -
W. Sahachairungrueng,
Sontisuk Teerachaichayut
Publication year - 2019
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/639/1/012038
Subject(s) - sweetness , hyperspectral imaging , partial least squares regression , mathematics , linear discriminant analysis , near infrared spectroscopy , calibration , absorbance , pattern recognition (psychology) , artificial intelligence , chemistry , remote sensing , analytical chemistry (journal) , food science , computer science , statistics , chromatography , sugar , optics , physics , geology
Near infrared analysis is a nondestructive technique used for determining the quality of various materials including fruit and other food. The objective of this study was to test whether near infrared hyperspectral imaging could be used for classifying sweetness of longan. One hundred and twenty samples were divided into a calibration set (n = 80) and a prediction set (n = 40). The average absorbance spectra from samples in the wavelength range of 935-1720 nm were used in this study. The sweetness of longan was represented by total soluble solids (TSS) which was used to separate fruit into a low sweet (TSS≤ 21.30°Bx) and high sweet fruit (TSS> 21.30°Bx). A classification model was developed in order to classify groups of longan based on sweetness, where 0 = low sweet and 1 = high sweet, by partial least squares discriminant analysis (PLS-DA). Spectra were preprocessed using a Savitzky-Golay smoothing method in order to obtain the optimal performance of the classification model. The results showed an accuracy of the classification model in the calibration set of 85% and the accuracy was 77.5% in the prediction set. Therefore, it was concluded that near infrared hyperspectral imaging has a potential for classifying longans nondestructive based on sweetness.