Detection of fat content in peanut kernels based on chemometrics and hyperspectral imaging technology

Abstract

Hyperspectral imaging technology combined with chemometrics were applied to detect fat content in peanut kernels. Four varieties of peanuts were scanned to acquire hyperspectral images by using a “push-broom” system. Then, the spectral data was extracted from hyperspectral images. Principal component analysis (PCA) was used to detect outliers. After outliers removed, five different pre-processing methods were used to preprocess spectral data. Successive projections algorithm (SPA) and regression coefficient (RC) were adopted to select effective wavelengths. Finally, based on the full wavelengths and the effective wavelengths, the models of partial least squares regression (PLSR), support vector machine regression (SVR) and multiple linear regression (MLR) were established respectively. Comparing these models, Baseline-SPA-MLR was the most excellent with determination coefficient (R2p) of 0.9736, root mean square errors (RMSEp) of 0.4635% and residual prediction deviation (RPD) of 6.1273 in the prediction set. All results in this study indicated that the combination of chemometrics and hyperspectral imaging technology provided an efficient and non-destructive method for detecting the fat content in peanut kernels.