<title>Fault detection by a Gaussian neural network with reject options in glass bottle production</title>

Abstract

ABSTRACT During the production of translucent glass bottles, many inspection procedures are realized in order to elirni-nate defects which produce dangerous consequences for customs. checks on the neck of a bottle. which look like cracks in the glass. are one of the most important defects. Although an automated visual inspection systems hasbeen developed to solve this specific problems, his ability to cope with variations of the environment is limited and it requires to be tuned very carefully whenever the characteristics of the production change. In this paper. wepropose a new approch based on computer vision and artificial neural network for check detection. The inspection procedure involves extracting features images of necks, the selection of the most discriminant features. and the decision is realised by a gaussian neural network with reject options. Keywords: Industrial process. visual inspection, fault detection. features extraction. features selection. gaus-sian neural network, EM algorithm. minimum description length. reject options