Soluble Solid Content Prediction System of Honey based on Spectral Transmittance Profile of Hyperspectral Imaging

Abstract

Honey content is constructed by a combination of glucose and fructose which a high sugar content. One of the honey qualities is contributed by the added adulterant in honey producing such as artificial glucose or fructose. Therefore, the soluble solid content of honey is essential to predict according to differentiate the added content of honey. The honey image was acquired using transmittance mode in the Vis-NIR range of 400–1000 nm. The complete system consists of a Hyperspectral camera at 448 bands, slider, a 200 W halogen lamp and light diffuser. The processing method performs image correction, segmentation, feature extraction, feature reduction, and a regression model. The region interest area of the honey sample was selected at the center of honey that prepared in petry-dish. Partial Least Square Regression (PLSR) was used as the feature reduction and regression model to construct the transfer model of soluble solid content based on the transmittance profile of honey. The Digital Refractometer was used to generate the reference standard of the soluble solid content. Three different producers and five types of botanical origin were used as a sample of honey. The artificial glucose or fructose was added to the original honey to produce five variants of soluble solid content. The result of RMSE for training and testing data is 0.07 and 0.45, respectively. Based on the result, the proposed system could be used as an alternative method to predict the soluble solid content in honey with excellent accuracy.