Offline Handwritten Kannada Numerals Recognition

Abstract

Handwritten Character Recognition (HCR) is one of the essential aspect in academic and production fields. The recognition system can be either online or offline. There is a large scope for character recognition on hand written papers. India is a multilingual and multi script country, where eighteen official scripts are accepted and have over hundred regional languages. Recognition of unconstrained hand written Indian scripts is difficult because of the presence of numerals, vowels, consonants, vowel modifiers and compound characters. In this paper, recognition of handwritten Kannada numeral characters is implemented and the different Wavelet features are used as feature extraction in this paper. The zonal densities of different region of an image have been extracted in the database. The database consists of 50 samples of each Kannada numeral character. For classification, the K-Nearest Neighbor method is used. Recognition accuracy of 88% has been achieved.