Bee honey classification using a combined texture analysis in sugar crystal images

Different chemical and physical methods to identify honey are a growing topic due to their impact on health and the economy. For decades, specific tools have been devoted to this task, but nowadays they face advances in adulteration or mixing methods. Monofloral honey crystallizes in a natural form, and it is possible to observe those crystals’ complex structures without the signal reduction of liquid honey using second harmonic generation microscopy. These structures are registered as images, allowing the observation of crystal clusters that are specific to each type of honey’s floral origin. In this work, a Kurtosis matrix, serving as a texture map derived from microscopic images, is introduced. A grey-level matrix correlation algorithm then processes the resulting patterns to determine the degree of correlation among several honey samples, allowing them to be identified solely from images of the sugar crystals. Results suggest a possible tendency to identify and group honey by the same floral origin, regardless of the location and harvest season in which the honey was produced. • Optical combined analysis identifies monofloral honey using its sugar crystals. • The use of Kurtosis can create a texture pattern of honey crystals. • Grey level correlation uses the Kurtosis matrix to classify honey. • Monofloral honey families are grouped according to their floral origin.