Fingerprinting of Olive Oil from Spectral Data

Abstract

Olive oil is an extensively used product and extra virgin olive oil is much costlier than other edible oils. Hence, purity of olive oil is a very significant issue. Fluorescence spectroscopy is a largely acceptable, simple, reliable and quick technique for adulteration detection and fingerprinting of olive oil. In this project, principal component analysis has been performed on fluorescence spectral data of 100 samples including pure extra virgin olive oil and adulterated ones with sunflower oil. The analysis has been able to successfully map the samples in a clear pattern for adulteration detection. The maximum tolerance limit for detection of adulteration is ±4.71% for the range of 0%-80% adulterated samples and ±5.67% for the range of 80%-100% adulterated samples. Also, by using two third of the samples as training set, this system can detect the rest one third samples (test set) quite accurately with an average tolerance of only ±3.42%. It has also been found that, short time exposure to laser, as a crude indication of possible long time exposure to sunlight, can definitely affect the fluorescence emission spectra. The two most significant wavelengths have been found (using variability) and validated (by principal component loading), that can replace the use of spectrometer with two color fiber optic probe. In this way, the computational complexity can be reduced to a great extent to make the adulteration detection system more affordable at retailer level.