Prediction of significant oil properties using image processing based on RGB pixel intensity.

[Display omitted] • A total of 13 different edible and non-edible oils are tested under image processing. • 1300 images are captured under a fluorescent lamp with CANON-EOS-200D digital camera. • KNN and SVM models are used for validation. • High Accuracy RGB image pixel extraction process with Fiji. • Significant properties of oils are predicted accurately. Renewable oils from different feedstocks are considered a potential replacement for the current petroleum products, and they are widely used in various applications like alternative fuels to heat engines and bio-coolants for metal removal operations, and their analysis of significant physicochemical properties is significant before their application. This study attempts to analyze the significant oil properties of different edible and non-edible oils using an image processing technique. For this purpose, 11 different oils are considered for experimentation, and their surface images of all the oils are captured and utilized to extract significant pixel information of Red, Green and Blue (RGB). The RGB pixel information is correlated with experimental properties (viscosity, density, flash point, fire point, cloud and pour points) and the developed model is validated with KNN, SVM and Fiji techniques and observed a high prediction accuracy (99%, 90.9% and 99%). Finally, two different oils which are not used in the experiment are chosen to test the prediction accuracy of the developed model. The results show that the RGB pixel predicts the significant oil properties, and they are validated experimentally and observed that the developed model predicts the oil properties with 91.7% accuracy. [ABSTRACT FROM AUTHOR]