Handwritten Devnagari Special Characters and Words Recognition Using Neural Network

The aim of this paper is to develop software which can recognize off line Devanagari special words which is made up of half consonant and consonant and special characters which is half form of consonant from scanned image of written documents using neural network. This paper will help to support easily digitization of the Devanagri script. Using this methodology for reorganization of the special characters of Devanagri script it is also easy to digitize the books written in Devanagri because ability of this methodology not only recognizes the normal characters but also the special characters. Some conventional methods like feature extraction and edge detection will be used for pre-processing the characters. These characters will be analysed by comparing its features. The process of data training of samples collected by different people will be followed after previous process. The proposed method will provide accuracy up-to 90% for special characters of deavanagri script with less training time.