Recognition of Devanagari Handwritten Numerals using Two Different Approaches

This paper proposes two methods for automatic recognition of Handwritten Devanagari Numerals. In first method, Grid features i.e. structural features are extracted and minimum distance is calculated using these features for classification. In second method, ICZ (Image Centroid Zone) & ZCZ (Zone Centroid Zone) features based on distance information are extracted and given to an already trained Neural Network for classification. The accuracies obtained using two approaches are evaluated. Keywords— Artificial Neural Network, classification, feature extraction, Handwritten Numeral Recognition, preprocessing.