Rapid and Non-destructive Determination of Oil Content of Peanut (Arachis hypogaea L.) Using Hyperspectral Imaging Analysis

Abstract

Peanut (Arachis hypogaea L.) is rich in some important oils such as the high content of polyunsaturated fatty acids. The potential of hyperspectral imaging technique in the spectral range I (400–1000 nm) and II (1000–2500 nm) coupled with chemometrics analysis for predicting oil content in different peanut cultivars was investigated in this study. Hyperspectral images were obtained and the corresponding spectral data was extracted. Quantitative calibration models were established between pre-processing spectral data and the reference measured oil content by partial least squares regression (PLSR) analysis. By comparing the model performances based on different spectral pre-processing methods, the raw-PLSR models using full wavelengths presented better results with the determination coefficient (R2 p) of 0.696 and 0.923, and root mean square errors by prediction (RMSEP) of 0.416 % and 0.208 %, respectively. In addition, six optimal wavelengths in the spectral range II were selected based on the regression coefficients of the established raw-PLSR model. The simplified PLSR model established only using identified optimal wavelengths also showed good performance with R2 p of 0.934, and RMSEP of 0.197 %. The results demonstrated that hyperspectral imaging technique is a promising tool for rapid and non-destructive determination of oil content in peanut.