Character Recognition Using Neural Network

Abstract

A method for recognizing handwritten characters using neural network is proposed. It is developed for the isolated handwritten English character (a to z , A to Z and 0 to 9). This method improves the character recognition over old methods which are based on cross correlation. Preprocessing of the Character is done by binarization, thresolding and segmentation method. The ANN is trained using the Back Propagation algorithm. In the proposed method, English Characters are represented by binary numbers that are used as input and then these are fed to an ANN. Training of system is done by using different data set and then system is tested for new samples and accuracy is measured.