CHARACTER RECOGNITION USING ARTIFICIAL NEURAL NETWORK

Abstract

Among the numerous advances those have been made in developing intelligent systems using computers, some are inspired by biological neural networks. These system can be mould in a way , that can be used for character recognition. This dissertation deals with the first step in alleviating this effort, that is electronically reading and recognizing hand-written characters directly from manuscript, thus removing the need of typing. Backpropagation algorithm is adopted here to deal with the variation and deformities that occur in normal hand-writing. The domain chosen for recognition is the English alphabet (vowels). In this dissertation, MS-Paint Brush is used in making hand-written type manuscript in BMP file format, which is used for testing of the used neural network . But the manuscript can be scanned with image scanner in BMP file format. The BMP file in addition to the image pixel data, contains a header in which various information related to the scanned image resides. In preprocessing stage of BMP file, first the pixel values are segregated by detaching the header. The extracted pixel data is then scanned by the program for detecting the existence of characters. The character found are then scaled and quantized into data matrix of Is and Os .These matrix data is then fed to feedforward multilayered error backpropagated trained neural network for recognition. The output is then interpreted to get recognized character. The recognition efficiency can be enhanced by proper training of neural network. The program code has been divided into various modules and total software package is made using C/C++ language.