Cullets (glass fragments) quality control by artificial vision: a color based approach

Glass fragments (cullets) to be recycled present different market values according to their color. Glass recycling plants perform cullets sorting mainly discriminating colored glasses from white and half white glasses; furthermore sorting presents some other technological limits concerning the minimum cullet size, about 45 mm, that is possible to analyze. In this paper are analyzed the possibility that could be offered by the adoption of a color imaging based approach to realize cullets sorting. This study was mainly focused on the effects that cullets surface status and characteristics produce on the detected color digital spectra and how they can influence the further classification. All the tests have been performed on glass samples as they result after the cleaning stage, impurities removal, of an industrial glass recycling plant.