A handwritten digit recognition system oriented for VLSI implementation is proposed

in which a modified Hamming neural network is used to extract local features from patternand a fuzzy logic processor is adopted to do the recognition. Some new techniques are used to improve the feature extraction performance under the constraint of the feature set size. Another twotechniques are also proposed in fuzzy logic processor design. Experiments show that it has strongability to deal with distortion and shift variations in handwritten characters recognition.