Optical Character Recognition Using Feedforward Neural Network Trained by Backpropagation Algorithm

Abstract

Recognizing characters is a problem that at first seems simple but it's extremely difficult in practice to program a computer to do it. And yet, automated character recognition is of vital importance in many industries, which handle floods of paper works everyday. In this study, a feedforward neural network based character recognition system was implemented. The system consists of two parts. The first part is a preprocessor, which is intended to produce a binarized, segmented, and normalized representation of the input pattern. The preprocessed output will then be classified by a neural network classifier trained by a backpropagation training algorithm. Results are shov/n concerning training data consists of characters of font styles Arial, Verdana and Times New Roman. All are of font size 12. The system yields a 78% recognition rate on the training data.