Pattern model for recognition of subset of Telugu characters on palm leaf scribings

Abstract

Digitizing documents have enormous benefits such as quick search, saves space, accessibility, etc. Even with the advent of many technological developments like digital pen, compact disc, etc. people still use the natural interface to record text. This indicates that the newlinehandwritten documents could persist even for some more decades. The various applications of character recognition are automatic mail sorting, signature verification, library automation, language newlineprocessing, forensic applications and multimedia design. Telugu script which is an offshoot of Brahmi script has complex newlinestructural characteristics and makes character recognition still difficult. It has 16 vowels and 36 consonants which form basic characters. The number of basic and compound characters is more compared to the foreign language, English. Every Telugu vowel has a modifier to modify the consonants. When a consonant is combined newlinewith a vowel modifier a compound character is formed. These compound characters are more similar and difficult to recognize them. For several centuries, before the invention of paper, a raw material called palm-leaf was used for recording/writing. As per the survey organized by Institute of Asian Studies (IAS) Chennai, more newlinethan one lakh palm-leaf manuscripts are available in the South Indian museums and research institutes. These are recognized as the key to the knowledge in diverse fields that have been passed through newlinegenerations. This includes tracts dealing with medicine, literature, astronomy, mathematics, vedas, astrology, etc. They get deteriorated due to some of the factors such as climatic variations, improper handling, aging, insect bites, etc. Hence preserving and digitizing text newlineon palm-leaf manuscripts is highly essential.