Evaluation of an adulteration detection system of olive oil based on machine vision and dielectric spectroscopy

The main fraud way in olive oil adulteration is mixing with other vegetable oils such as maize, canola, sunflower and soy oil. Therefore, the aim of this study was to evaluate a portable system to detect adulteration in olive oil using combination of machine vision and capacitive methods. To identify adulteration in olive oil by using frequency and color features, Principal Component Analysis (PCA) and Linear discriminant analysis (LDA) were thus applied. The best neural network model with 36-6-1 structure had R2 of 0.944 and mean square error of 0.006, for prediction of adulterated olive oil with sunflower oil. Also, the best model of neural network for olive-canola mixture had the structure of 36-10-1, R2 and mean square error of 0.946 and 0.003, respectively. Finally, a combination of frequency and color properties was used to develop the models. The R2 and mean square error of mixed samples of olive-sunflower were obtained as 0.962 and 0.008 for 38-2-1 network. For mixed samples olive-canola, the R2 and mean square error were obtained 0.961 and 0.0013 for the structure of 38-16-1 neural network.