Standardization of near infrared hyperspectral imaging for wheat single kernel sorting according to deoxynivalenol level

•Single kernel wheat kernels were analysed through hyperspectral image acquisition.•Reflectance spectral data were 1st derivative and SNV transformed as optimum pretreatments.•Root mean square error of prediction (RMSEP) was 4.8 mg/kg for DON quantification.•A 100% of kernels were correctly classified according to symptomatology.•Up to a 98.9% of kernels were correctly classified according to DON.