Modeling and Evaluation of Honey Adulteration Based on Image Processing of Water-Soluble Samples

Adulteration, especially industrial type, is caused by the direct addition of natural or fermented syrups to honey. In this study, fennel honey was supplied from beekeepers located in Kangavar city.After ensuring the authenticity of honey, 39 samples of counterfeit honey were prepared by adding and stirring sucrose, fructose and a combination of 0.9 fructose-glucose syrups in natural honey at different levels of 0 to 100 wt%.Different samples were dissolved in water and their images were recorded using a camera.In order to process the images in each 33 monochrome channels, 15 parameters (495 parameters in total) were measured.  Few parameters were selected by sensitivity analysis using ANFIS fuzzy neural inference network, ANN artificial neural network and RSM response level. The explanation coefficient of the presented models for water-soluble samples was 0.9512, 0.9882 and 0.9904, respectively.Considering all the statistical error valuesof the RSM model, it was introduced as the best model to determine the amount of honey fraud in this method by the desirability function.