Abstract: In this paper, a Character Recognition system is designed using Multilayer Feed forward Artificial Neural Networks. For the training of neural network Backpropagation Learning algorithm is preferred. Input is typed English alphabets with a space between the individual characters, which are input to the segmentation algorithm and then fed as input to the neural network. A neural network is trained onto the dataset containing 6 different samples for each 26 alphabets for recognition. We set the size of our training images as 40x40 gray scale image. From such training images we create image profile. Each element of the profile vector is connected with one input to the neural network. An implicit approach for character recognition is implemented in this paper for its recognition, which improves the accuracy significantly.

Keywords: Character recognition, ANN, Segmentation, Recognition.