Identification of Peanut Kernels Infected with Multiple Aspergillus flavus Fungi Using Line-Scan Raman Hyperspectral Imaging

The mold contamination caused by Aspergillus flavus poses a serious threat to food safety. In this study, three artificially inoculating strains of Aspergillus flavus (A. flavus 142,801, A. flavus 142,803, A. flavus 336,156) were used to infect two healthy peanut varieties (variety A: GS1210, variety B: fengyingluohan) kernels. These healthy and Aspergillus flavus-infected peanut kernels were identified and differentiated by using a line-scan Raman hyperspectral imaging system. Firstly, the average spectra of healthy and infected peanuts were extracted, followed by preprocessing using Savitzky-Golay smoothing and airPLS for fluorescence background removal. Finally, four feature variable selection methods were used to optimize the models. In the binary classification model (healthy vs. A. flavus), the SVM method yielded the best modeling results, with accuracy above 99