Non-destructive discrimination of sesame oils via hyperspectral image analysis

Sesame oil is produced from cheap imported raw materials, but its quality differs depending on the imported form. The purpose of this study was to determine the quality characteristics of sesame oil through hyperspectral image (HSI) analysis. The following types of sesame oil samples were used in this study: sesame oil WS (WS1, WS2) prepared using imported whole-sesame seeds, sesame oil SP (SP1, SP2) prepared using only imported sesame powder and sesame oil WSP (WSP1, WSP2) prepared by mixing imported sesame powder with whole-sesame seeds. The principal component analysis revealed that WS and SP or WS and WSP had different principal components. As a result of partial least squares-discriminant analysis, WS and WSP had an average R² of 0.9289 in a training model, and prediction R² was the highest at 0.8333 in a test model. The developed model was affected by C-H stretch second overtones for fatty acids and C-H and O-H bonds corresponding to phenolic absorbance. The average R² of partial least squares regression developed owing to fatty acid and phenolic compound content was the highest. Therefore, it was possible to analyze the quality characteristics of sesame oil according to the form of raw material using hyperspectral image.