A neural network approach to quality control of padlock manufacturing

Abstract

We describe a neural network based visual quality control system especially suited for the needs of padlock manufacturing. In this system the image produced by the CCD camera is converted into a binary image by using a custom filter. The image is divided into subimages which are aligned via a set of reference images. The horizontal and vertical projections of the subimages are used as input features to a MLP neural network classifier. We also present some preliminary empirical results.