Character recognition using neural based feature extractor and classifier

The authors present a neural network architecture for the recognition of handwritten digits and machine printed multi-font characters. To reduce the dimension of the input data vector as well as robustness of the system, an appropriate neural net is utilized for the feature extraction part which is cascaded with another neural net for the classification purpose. The proposed architecture has been tested on a large sample of real field data and the results indicate the effectiveness of the proposed technique.<>