Smartphone-based method for the determination of chlorophyll and carotenoid contents in olive and avocado oils: An approach with calibration transfer

Abstract

Chlorophylls and carotenoids in vegetable oils are usually analyzed by spectrophotometric methods. On the other hand, a simple and economical analytic method is to apply a portable colorimetric method based on digital images. However, in some cases, the results can be influenced by external variables, such as the type of camera and lighting. Thus, the objective of this study was to use a method that could take data from digital images taken by smartphones, calibrate the data by multiple linear regression (MLR) or least squares support vector machine (LS-SVM), and thereby predict the levels of chlorophyll and carotenoids in olive and avocado oils. In addition, a type of calibration transfer was proposed based on images of FeCl3·6H2O solutions and was tested under different conditions (type of camera and lighting). The effect of the color space used to describe the images, including RGB, Y, HSV, CMYK, L*a*b, and XYZ, was also evaluated. The models performed well, but the LS-SVM models outperformed the MLR models on both chlorophylls and carotenoids. The predictions under secondary conditions were more strongly influenced by lighting than by smartphone type. Considerable improvements in the predictions under secondary conditions were observed when applying the calibration transfer, especially under different lighting conditions, as the RMSE decreased and the R² increased for both pigments. The best performance after applying the calibration transfer was obtained using the XYZ/LS-SVM model for chlorophylls and Y/LS-SVM for carotenoids, which had mean RMSEs of 0.96 and 0.29 mg/kg with mean R² of 0.96 and 0.83 for chlorophylls and carotenoids, respectively. Therefore, the method based on digital images taken by smartphones and the proposed calibration transfer shows great potential to determine pigments in vegetable oils, with good accuracy under different lighting conditions and with different smartphone models.