Tri-scripts handwritten numeral recognition: a novel approach

In this paper an automatic recognition system for isolated handwritten numerals recognitionfor three popular south Indian scripts. Kannada, Devanagari, and Telugu numeral sets are used for theirrecognition. The proposed method is thinning free and without size normalization. The structuralfeatures viz. directional density of pixels, water reservoirs, maximum profile distancess, and fill holedensity are used for handwritten numerals recognition. A Eclidian distance criterion and K-nearestneighbor classifier is used to classify the handwritten numerals. A total of 5250 numeral images areconsidered for experimentation, and the overall accuracy of 95.40%, 90.20%, and 98.40% for Kannada,Devanagari and Telugu numerals respectively are achived. The novelty of the proposed method isthinning free, fast, and without size normalization.