Detection of sprout damage in wheat kernels using NIR hyperspectral imaging

Abstract

The use of near-infrared (NIR) hyperspectral imaging (HSI) for detecting sprout damage in wheat kernels was investigated. Experiments were carried out to determine which spectral bands had the best potential for discriminating between sound and sprouted kernels. Two wavelengths were selected and combined into an index that was used to indicate the presence or absence of sprouting. Experiments with three wheat cultivars revealed that the proposed method is effective in identifying kernels for which the germination process has initiated, achieving 100% accuracy for the samples used in this study. It was also observed an imperfect correlation with the Falling Number (grain quality), making it challenging to accurately determine the degree of germination, especially if sprouts are not yet clearly visible. These results confirm the usefulness of the near-infrared spectral range for detecting chemical alterations in wheat kernels, as well as the fact that most information is usually contained in a few specific bands within such range.

Standardisation of near infrared hyperspectral imaging for quantification and classification of DON contaminated wheat samples

“…Thus, although in the present work the whole spectra were used, if better models were achieved, wavelength selection would be desirable. As described above, some authors chose a discrete number of wavelengths based on local spectra minima or maxima, and their differences related to visually damaged grains (Barbedo, Guarienti, & Tibola, 2018;Delwiche et al, 2010), other used PLS coefficients (Tekle, Mage, Segtnan, & Bjornstad, 2015) or used smarter techniques (Delwiche et al, 2019;Liang et al, 2018). In the NIR region, the lower wavelength region demonstrated closer similarity between sound and Fusarium-damaged kernels reflectance for most wavelengths pairs (Delwiche et al, 2011 (Dowell et al, 1999).…”

Use of hyperspectral imaging as a tool for Fusarium and deoxynivalenol risk management in cereals: A review

“…The comparison between the studies using HSI shows that the predominant method used for HSI-NIR is a line-by-line scanning (push-broom imaging) at different pixel resolution (x, y, ). Barbedo, Tibola, & Fernandes, (2015) and Barbedo, Guarienti, & Tibola, (2018) used this method to capture a 3D hypercube with a 320 x 800 x 256 dimension. Besides, Delwiche, Kim, & Dong (2011) also acquired a push-broom imagine but with a different pixel resolution of 320…”

Rapid screening of DON contamination in whole wheat meals by Vis/NIR spectroscopy and computer vision coupling technology

“…Currently, various methods have been developed for DON determination, including fast screening and chromatography or mass spectrometry‐based quantitative methodologies, such as enzyme‐linked immunosorbent assay (ELISA), lateral flow immunoassay (LFI), high‐performance liquid chromatography‐mass spectrometry (HPLC‐MS) and gas chromatography‐mass spectrometer (GC‐MS) (Pereira et al ., 2019). Although these methods are more reliable, special pretreatment made them impossible for large‐scale screening in an online sorting and production system (Barbedo et al ., 2018).…”