Honey botanical origin classification using hyperspectral imaging and machine learning

Publication date: Available online 5 August 2019Source: Journal of Food EngineeringAuthor(s): Ary Noviyanto, Waleed H. AbdullaAbstractHoney has been growing as a trade commodity that has a significant impact on the economy. Development of honey quality assessment methods is vital to protect consumers from fraudulence. Automatic and non-invasive methods are interesting concepts which can deal with problems of conventional chemical-based methods and may complement them. A combination of hyperspectral imaging and machine learning offers a promising quick, automatic and non-invasive approach for honey botanical origin classification. The present research investigates techniques in building a honey botanical origin classification model based on reflectance spectra. The developed techniques include noisy band elimination, spectral normalisation and hierarchical classification. The proposed model showed promising results under several classification scenarios. The classification performances were around overall balanced accuracy values of 91% for close-set and 80% for open set in a general scenario; also almost 90% and around 88% balanced accuracy in average for a producer-specific scenario with close and open set cases respectively.
Source: Journal of Food Engineering - Category: Food Science Source Type: research