Hybrid framework for tamil handwritten character recognition system using deeplearning approach

Abstract

Optical Character Recognition (OCR) is one of the growing newlinetechnologies in the data processing section. The objective of Optical newlineCharacter Recognition is defined as identifying and understanding the data newlinepresented on paper or digital documents by the system as humans do. Various newlineattempts and research have been carried out in this realm. The success of newlineLatin script recognition opened the research publically, worldwide. According newlineto the Indian research, work-related to Hindi language recognition got a lot of newlineattention, since it is the national language of India. But till date no one could newlinebeat the growth of the Latin script by other non-Latin scripts, in terms of newlineaccuracy rate and successful implementations. This thesis will reveal the newlineterms presented in character recognition in Tamil, one of the popular south newlineIndian languages. Tamil is one of the oldest Dravidian languages with a strong traditional history. Tamil character recognition has two categories such as printed newlinecharacter recognition and handwritten character recognition. Handwritten newlinecharacter recognition can be further classified into two types such as online newlineand offline recognition. During the online identification of Tamil handwritten newlinescript, the OCR engine will acknowledge the written characters spontaneously newlinewhile writing using a digital pen, transducer, and touch screen. But the offline newlineidentification of Tamil handwritten script is just opposite to the online newlineidentification. In this offline recognition, the handwritten Tamil characters are newlineidentified later at the end of the writing and scanning process. Offline Tamil newlineHandwritten Character Recognition (THCR) is characterized as the automatic newlineidentification of Tamil isolated symbols or words from the scanned image newlinedocument. newline newline