Fusarium damage assessment in wheat kernels by Vis/NIR hyperspectral imaging

Classification of Fusarium-infected and healthy wheat kernels based on features from hyperspectral images and flatbed scanner images: a comparative analysis

“…Compared to the literature data in our own studies even higher classification accuracies were obtained. Delwiche et al [29] classified Fusarium-damaged and healthy wheat kernels with around 95% average accuracy. Shahin and Symons [30] discriminated Fusarium-infected and healthy wheat kernels with approximately 90% total accuracy.…”

“…The near-infrared (NIR) range was more suitable for identifying infected kernels than the visible (VIS) range. Delwiche et al [29] and Shahin and Symons [30] relied on Vis/NIR hyperspectral imaging to assess fungal infections in wheat kernels based on reflectance analysis. Barbedo et al [31] used hyperspectral imaging with reflectance measurements in the wavelength range of 528-1785 nm to develop an algorithm for detecting wheat kernels infected with Fusarium fungi, and to estimate the concentration of deoxynivalenol (DON).…”

“…The cited authors considered the use of only kernel fractions which are less susceptible to specular reflection to eliminate washed out segments from the image. However, according to Delwiche et al [29], analyses performed on whole kernels usually deliver more accurate results.…”

Principal Component Analysis in the Development of Optical and Imaging Spectroscopic Inspections for Agricultural / Food Safety and Quality

“…Conventional grain inspection procedures for Fusarium damage are heavily reliant on human visual analysis. As an inspection alternative, Delwiche et al (2011a) investigated the potential of hyperspectral image systems (1000 -1700 nm NIR vs. 400-1000 nm visible) in the detection of Fusarium-damaged wheat kernels. On a limited set of wheat samples that their conditions were subjectively assessed and also using a linear discriminant analysis (LDA) classifier, they found that hyperspectral imaging in either visible or NIR regions was able to discriminate Fusarium-damaged kernels from sound kernels at an average accuracy of approximately 95%.…”

Hyperspectral imaging in quality control of inkjet printed personalised dosage forms

“…Near infra-red hyperspectral imaging is a non-invasive tool that has been used in many fields, mainly in agriculture and food products research for quality and safety assessments (Delwiche et al, 2011;Csefalvayova et al, 2011;Daleab et al, 2013;Kamruzzaman et al, 2011;Lopes et al, 2009). This technique has also been applied in the pharmaceutical studies from early research and development (R&D) to final product analysis (Shi et al, 2012;Gendrin et al, 2007;Sulub et al, 2011;Franch-Lage et al, 2011;Ravn et al, 2008).…”