Optical Character Recognition for Brahmi Script Using Geometric Method

Abstract

Optical character recognition (OCR) system has been widely used for conversion of images of typed, handwritten or printed text into machine-encoded text (digital character). Previous researches on character recognition of South Asian scripts focus on modern scripts such as Sanskrit, Hindi, Tamil, Malayalam, and Sinhala etc. but little work is traceable to Brahmi script which is referred to as the origin of many scripts in south Asian. This study proposes a method for recognition of both handwritten and printed Brahmi characters which involve preprocessing, segmentation, feature extraction, and classification of Brahmi script characters. The geometric method was used for feature extraction into six different entities, followed by a newly developed classification rules to recognize the Brahmi characters based on the features. The method obtains accuracy of 91.69% and 89.55% for handwritten vowels and consonants character respectively and 93.30% and 94.90% for printed vowel and consonants character respectively.