Comparative Study of Devanagari Handwritten and printed Character Numerals Recognition using Nearest-Neighbor Classifiers

摘要

In this paper we are using Devanagari script OCR for recognition. The handwritten data set is created by us and for printed characters we have used ISM font. Here we are using gradient and curvature based feature extraction method. We have compared Nearest Neighbor, K-Nearest Neighbor, Euclidian Distancebased K-NN, Cosine Similarity -based K-NN, Condensed Nearest Neighbor, Reduced Nearest neighbor, Farthest like neighbor and Nearest unlike Neighbor.