Identification and Transliteration of Braille Letters into Sinhala Text

Braille, a tactile reading and writing system, provides visually impaired people with access to information and educational resources. Despite successful transliteration of Braille into many languages, Sinhala Braille faces ongoing difficulties due to insufficient datasets and error-prone conversion processes. This research aims to bridge this gap by concentrating on precise transliteration from Braille to Sinhala text, utilizing a dataset of roughly 230 characters. Our methodology incorporates cutting-edge image processing, cell segmentation, and letter recognition techniques to improve transliteration accuracy. In contrast to previous studies that used handwritten or computer-generated Braille documents, we employed authentic machine-produced datasets to ensure relevance. The proposed system attained an overall accuracy of 90% for underrepresented datasets by filtering noise well and identifying letters fast. This showcases its capacity to enhance accessibility and aid for the vision-impaired community in Sri Lanka.