<i>Investigating the Development of Aspergillus flavus inoculated on Maize Kernels Using Visible and Near Infrared Hyperspectral Imaging</i>

Visible and near infrared (VNIR) hyperspectral imaging (HSI) combined with multivariate image analysis and chemometric techniques were used to investigate the development of Aspergillus flavus inoculated on maize kernels. Multivariate image analysis was used for cleaning the images and for making score plots of principal component analysis (PCA). Using the brush technique all irrelevant pixels (background, reflection from Petri dish, and bad pixels) were removed. Mean spectra of each kernel were calculated on the cleaned images. Partial least square-discriminant analysis (PLS-DA) models based on full wavelengths were developed to classify different infection periods of Aspergillus flavus on maize kernels. Informative wavelengths were selected using successive projection algorithm (SPA) to develop the simplified model. The correct classification rate (CCR) of calibration set, cross-validation set, and prediction set were 95.7 %, 93.0 %, and 93.3 % for full wavelengths model, 96.5 %, 94.8 %, and 93.3 % for simplified model, respectively. Results indicating that VNIR HSI can be used to investigate infection periods of Aspergillus flavus on maize kernels.