Determining the location of industrial bar-codes using neural networks

Although bar-codes are normally thought of in the context of the labelling of products in retail situations, they increasingly find use in a manufacturing context for identification purposes. This paper describes a technique which locates bar-codes in a computer image of a manufacturing cell. The bar-code detecting sub-system is based on a neural network. Two neural network paradigms (the multilayer perceptron and the radial basis function networks) are compared for use in this application. The work is part of a programme of research in which the aim is to develop a system which locates and decode bar-codes in real-time.