A neural network recognition system for machine printed characters on coils

Abstract

A new system applying the inherent pattern recognition capability of a neural network has been developed for the task of recognizing machine printed alphanumerics on steel coils. The system has been used in the newly constructed No.2 Kashima cold strip mill (2KCM) since April 1993 and has contributed to fully automatic operation in the 2KCM. The system is 100% accurate in detecting coils that have been transported by mistake, and can do so within a few seconds.