In-situ-Verfahren zur Klassifikation von Glasfehlern

For automatic quality inspection of glasses (wine glasses, beer glasses , ...) a new method was developed. By the help of this method glass defects will be locate, and classified. The glass surface is projected on a groundglass with the help of a appropriate optical arrangement. Based on images, which are picked up by a CCD camera, various classificators are calculated with methods of industrial image processing. These classificators are the input elements of a neuronal network. The output of the network consists of only one neuron, which gives a good/bad decision.