A neural network based recognition scheme for the classification of industrial components

This paper outlines a method for representing the silhouettes of industrial components by generating a vector sequence of Euclidean distances between the shape centroid and each boundary pixel, which is translation invariant and can exhibit scale and rotation invariance if required. The sequence can be re-sampled to form a suitable input vector for an artificial neural network (ANN). Three different ANN topologies have been implemented: the multilayer perceptron, a learning vector quantisation network and hybrid self organising map. This method of representing industrial components has been used to compare the ANN architectures when implemented as classifiers based on shape and dimensional tolerance. A number of shortcomings with this methodology have been highlighted; most importantly the identification of a unique sequence start point, vital for rotation invariance. Another problem may arise due to the conflict between the inherent robustness of ANNs when dealing with noise, and classifying components which are similar but display subtle dimensional differences