FNN model for multi-font character recognition

A novel fuzzy neural network (FNN) model for multi-font character recognition was presented, which can efficiently process the fuzzy pattern classification problem. This FNN model is built by fuzzifying the input layer, output layer and the training algorithm of a conventional multilayer perceptron (MLP). The simulation with a lot of multi-font character samples shows that the FNN presented here can get a high recognition rate, and has low sensitivity for different character fonts in comparison with classical MLP. Also, this FNN is proved to have a good robustness.