DEVANAGARI HANDWRITTEN CHARACTER RECOGNITION METHODS AND ANALYSIS

Presently these days recognizing the handwritten character recognition is getting high essentialness in light of various applications like educational field, digitized signature verification, bank processing, postal code acknowledgment, electronic library and so on. exceptionally less work is ac-counted in the research of Devanagari hand written character acknowledgment, so that there is an enormous extent of research right now. Some potential challenges adding to the unlawful execution of different frameworks for seeing deciphered characters are: various shapes, broken characters, various tendencies and measures, and so on. To overcome these types of issues, initially, we introduce a pixel intensity histogram based feature for the special character recognition, it identifies the special symbols and characters and different types of characters. Further, we used selection process done by improved rule based feature set selection algorithm. Dataset is collected and with help of this proposed improved rule based feature set selection algorithm, the accuracy of character identification is improved. Further, we use recurrent-artificial neural network classifier for classification and recognition process to classify different types of handwritten characters. The performance of proposed model is compared with the existing designs in terms of higher accuracy and speed in classification and recognition.