Performance analysis of multiple classifiers for Marathi digit recognition

Abstract

For many researchers over the globe; the Marathi Language Handwritten digit recognition is an area of importance.. This study has its primary application to recognize the correct Marathi handwritten numerals. Many systems have been developed for numeral recognition using soft computing paradigms such as Artificial Neural Networks, Fuzzy logic-based methods. In this paper, three classifiers were used for handwritten Marathi numeral recognition that is K- Nearest Neighbour (KNN) Classifier, Bayes classifier and SVM Classifier. In this study primary database with 250 samples were used to perform the experiment. The SVM classifier gives 98% highest recognition accuracy.