The development of a hyperspectral imaging method for the detection of Fusarium-damaged, yellow berry and vitreous Italian durum wheat kernels

The possibility of using hyperspectral imaging (HSI) techniques to classify different types of wheat kernels, vitreous, yellow berry and Fusarium-damaged, was investigated. Conventional optical techniques adopted by industry for wheat grain sorting usually have too high misclassification errors. Reflectance spectra of selected wheat kernels of the three types were acquired by a laboratory device equipped with an HSI system working in near infrared field (1000–1700 nm). The hypercubes were analysed applying different chemometric techniques, such as principal component analysis (PCA) for explorative purposes, partial least squares discriminant analysis (PLS-DA) for classification of the three wheat types and interval PLS-DA (iPLS-DA) for the selection of a reduced set of effective wavelength intervals. The study demonstrated that good classification results were obtained not only considering the entire investigated wavelength range, but also selecting only three narrow intervals of four wavelengths (1209–1230 nm, 1489–1510 nm and 1601–1622 nm) out of 121. The procedures developed could be utilised at industrial level for quality control purposes or for the definition of innovative sorting logics for wheat kernels after an extensive classification campaign, both at laboratory and industrial level, applied to a large wheat sample sets.

Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slideHighlights► Classification of vitreous, yellow berry and Fusarium-damaged wheat by NIR-HSI. ► Different chemometric strategies applied to classify wheat kernels typologies. ► The different types of wheat kernels were correctly identified in the prediction images. ► 3 effective wavelength intervals were identified to speed up the classification procedure. ► The HSI-based procedure is useful for wheat quality control and on-line sorting strategies.