Neural quality inspection in industrial compact disc print stations

Artificial neural networks (ANN) are becoming powerful tools used for many pattern recognition problems in image processing applications. In this work we present a quality inspection system for compact disc print stations used to classify the quality of the print view. The system design and embedding strategy are based on the general principles described previously by the authors (1993) for optimizing the overall system efficiency. The conflicting demands of limited computing capacity and full resolution control are solved by the interactive mask approach. Parametrized disturbance algorithms, which can easily be adopted in the NEUROSIM environment, are used to automatically create sets of training data