A hybrid handwritten digits recognition system based on neural networks and fuzzy logic

A hybrid handwriting recognition system based on neural networks and fuzzy logic is proposed. The system consists of two stages. In the first stage, a Hamming neural network is used to extract local features from a pattern, and based on the feature maps, a fuzzy logic recognizer is adopted to do the recognition. In the second stage, the horizontal/vertical connected component features are extracted, the recognition task is also performed by a fuzzy logic recognizer. Experiments show that the performance of the hybrid system is better than either of both stages. It has very high recognition speed and large ability to deal with distortion and shift variations in handwriting characters