Localization and fuzzy classification of manufacturing defects in sheets of glass

Abstract

Artificial Vision Systems are commonly used in industrial applications. The low cost of the equipment facilitates the development of new products. In this paper we describe the use of an artificial vision system in one of the phases of a quality control process related to automotive industries: the windshield manufacturing. We intend to localize and classify the defects that were originated while manufacturing the glass that forms the windshield. We will show that a fuzzy classifier, after being tuned with a genetic parameter adjustement procedure, outperforms a neural networks based classfier.