Algorithm to Identify Kannada Vowels using Minimum Features Extraction method

Abstract

This paper introduces a novel way of feature extraction for Optical Character Recognition (OCR) customized for Kannada characters. The algorithm described here relies on breaking the character into four equal parts and using one of the quarters for extraction. The algorithm is deliberately kept away from all the complexities and the number of features to be extracted is also minimized so as to increase the efficiency and speed of recognition. The algorithm also describes a conflict resolution technique helpful in effectively utilizing the algorithm.