Recently, demand for authentication technology is growing rapidly in an attempt to overcome counterfeiting of high-value agricultural products, such as patchouli oil. Fingerprinting methods based on spectroscopy are one such technology being used for authentication. However, the spectral datasets obtained are multivariate in nature; containing thousands of data points for a single sample, making data acquisition and processing time-consuming. Therefore, reduction and simplification in the number of variables used required is needed to provide a more rapid and applicable method. Color cameras, which can capture image in the visible region light, could be such an alternative spectral data acquisition approach. In this research, a simplified spectroscopy method was developed for origin authentication of patchouli oil. The system consists of front ultraviolet light induced (365 nm) fluorescence and a white LED-based backlighting imaging system that consecutively captures the fluorescence and transmittance characteristics of the oil in the visible region. From the captured images, features were extracted and analyzed using Principle Component Analysis (PCA) to identify important image features for discrimination of origin. From the samples measured, the samples clustered around three islands of origin in the PCA space. A classification model based on fluorescence and transmittance image features (color values) could discriminate origin classes with a total accuracy of 88.46%. A lower accuracy was found for the Java class due to low sample numbers. This result demonstrates that the proposed system has the potential to be a rapid authentication tool for determining the geographical origin of patchouli oils.
Keywords: Absorbance; Classification; Emission; Essential oil; Fluorescence.
Copyright © 2019. Published by Elsevier B.V.