Handwritten Devnagri Character Recognition

Abstract

Recognition of handwritten characters has been a popular research area for many years. Devnagari script is a major script of India and is widely used for various languages. In this paper we propose a system to recognize devnagri handwritten characters. Total 60 devnagri characters (50 letters and 10 digits) are taken in to consideration. 60 samples of each character i.e. total 3600 samples are used for features extraction. Classification is done by four different classifiers which are Multilayer perceptron, K-Nearest Neighbour, Naive Bayes classifier and Classification tree. Performance of different classifiers is compared.98.9 % accuracy is obtained by Multilayer perceptron