A multi-stage processing technique for character recognition

Optical character recognition research has been developed for a long time and there are many techniques which provide very high identification rate. One major drawback of the existing methods is the long training time. Speeding up the training time for most techniques usually suffers the high reduction on the recognition rate. In this work, we present a neural network based approach to largely reduce the training time while maintain the high recognition rate. The main idea is to perform a preprocessing stage on the training data prior to the neural network training and use a template matching technique in the recognition stage. We have implemented the algorithms using different grouping features. The experimental results on real images have demonstrated the effectiveness of the proposed method.