A Comprehensive Survey on OCR Techniques for Kannada Script

In modern days, there is a pervasive inclination towards digitization of text documents for the ease of their access and maintenance. Digitized documents can be preserved for the future since this form has a longer shelf life. Optical Character Recognition (OCR) system translates a digitized text document from human readable form to machine editable codes. Many commercial OCRs are available today for documents written in English, Japanese, Chinese, Arabic and a few Indian scripts. Kannada is the official language of Karnataka, which is one of the southern states of India. Development of OCR for Kannada script is an active research area currently. Kannada language consists of a large set of characters, many of which are very similar in structure. This makes the job of developing an OCR for this language several magnitudes more complicated than for a language like English. The very fact that research on developing OCRs for Kannada language is very promising and is still emerging necessitated this survey paper. The aim of this paper is to discuss in detail the peculiarities of the Kannada script, challenges they pose for recognition, techniques reported in the literature, recognition accuracies and a comparison with other OCR systems.