An application to analyzing and correcting for the effects of irregular topographies on NIR hyperspectral images to improve identification of moldy peanuts

Near-infrared hyperspectral imaging (NIR-HSI) can be used for nondestructive, rapid, real-time detection in food safety; however, irregular sample topographies introduce variations in the spectral intensity that impair subsequent classification and inversion processes. In this study, the spectral variations in HSI images of peanut samples with irregular topographies were assessed via the classification gradient and singular spectrum analysis (SSA). An SSA based correction model (CMSSA) is proposed that assumes the spectral intensity of all pixels of peanuts should be equal. The method was validated via classification and the coefficient of variation (CV) and was found to eliminate the spectral variation caused by the irregular kernel topography while retaining chemical differences of interest. We anticipate this method will prove useful in food safety detection applications involving the quantitative inversion of parameters.