Script identification and language detection of 12 Indian languages using DWT and template matching of Frequently Occurring Character(s)

摘要

India is a diverse country with various cultural and traditional differences. There are more than 12 distinguished different languages in the country, viz., Hindi, Bangla, Marathi, Oriya, Tamil, Telugu, Assamese, Manipuri, Gujarati, Kannada, Malayalam, Panjabi, Nepali, Tibetan, Urdu, etc. Optical Character Recognition (OCR) of Indian Languages needs to be designed in such a way that it automatically identifies the language of the input document for further processing. There are many techniques which are already implemented, but the problem lies in identifying and detecting the correct language as some of the languages uses the same or similar script. For example, the Bangla script is used to write - Bengali, Assamese and Manipuri languages. Though the scripts can be distinguished using global technique, the problem with languages having similar script still exist. To deal with this problem, a robust wavelet transform cumtemplate-matching invariant to rotation, scale and translation technique is deployed to identify the script and detect the language of the document automatically.