Abstract:
Optical Sinhala Character recognition is a fairly researched area in the local natural
language initiative. Although a fair amount of research and development has been
undertaken in this field, a comprehensive product is yet to be developed. However, a
Sinhala OCR has a formidable academic and commercial value. The value of such a tool
is evident in the e-gonvemance intiative as well, where vast numbers of Sinhala
documents, forms etc will have to be digitised. OCRs have been developed in English and
many other languages, and some of these basic methodologies can be extended for a
Sinhala OCR as well. A forerunning concept on OCR is the use of weighted Artificial
Neural Networks. However, this method is requires heavy processing power. This
abstract proposes a fresh idea to reduce processing power, using image processing and
comupter vision techniques.
