ABSTRACT

In this paper we propose the Support Vector Machine (SVM) based recognition scheme towards the recognition of Gujarati handwritten numerals. The preprocessing is done considering morphological operations. For computing the features, each isolated numeral is segmented into blocks. These blocks create base for four sets of features. Then we derived affine invariant moments as features. The features obtained from these blocks are fed to SVM classifier. We obtained the recognition rate of 91% approximately.