Design and development of an efficient model for handwritten modi script recognition

Abstract

Machine simulation of human reading has caught the attention of computer science newlineresearchers since the introduction of digital computers. Character recognition, a branch of pattern recognition and computer vision, is the process of identifying either printed or handwritten text from document images and converting it into machinecoded text. Character recognition has been successfully implemented for various foreign language scripts like English, Chinese and Latin. In the case of Indian language scripts, the character recognition process is comparatively difficult due to various complexities such as the presence of the vowel modifiers and a large number of characters (class). MODI script is a shorthand form of Devanagari script and it was used as an official script for writing Marathi until 1952. Presently the script is not used officially, but has historical importance. MODI script is a cursive script and the character recognition task is difficult due to various reasons such as variations in the shapes of a character with different individuals and the presence of identical looking characters. MODI documents do not have any word demarcation symbols and that adds to the complexity of the task. The advances in various Machine Learning newlinetechniques have greatly contributed to the success of optical character recognition. newlineThe proposed work is aimed at exploring various Machine Learning techniques/ newlinemethods which can be effectively used in(to) recognizing(recognize) MODI script and newlinebuild a reliable and robust character recognition model for handwritten MODI script. This research work also aims at the development of a Machine Transliteration and text recognition system for MODI manuscripts.