AN IMPROVED ACCURACY OPTICAL CHARACTER RECOGNITION ALGORITHMS USING ARTIFICIAL NEURAL NETWORK

ABSTRACT

This research work represents an Artificial Neural Network based approach for the optical Character recognition (OCR) using back propagation neural network. Noise has been seen and considered as one of the major problems that reduces the accuracy and performance of the Optical character recognition system. The back propagation neural network has one input, one hidden and one output layer. The entire recognition system is divided into two sections such as training and recognition section. The two sections include image acquisition, preprocessing and feature extraction. Training and recognition section also include training of the classifier and simulation of the classifier respectively. Preprocessing involves digitization, noise removal, binarization, line segmentation and character extraction. After character extraction, the extracted character matrix is normalized into 12x8 matrix. Then features are extracted from the normalized image matrix which is fed to the network. The network consists of 97 input neurons and 63 output neurons. We train our network by proposed training algorithm in a supervised manner and establish the network. Eventually, the author has tested the trained network with more than 12 samples per character and gives 97% accuracy for numeric digits (0~9), 95% accuracy for capital letters (A~Z), 94% accuracy for small letters (a~z) and 91% accuracy for alphanumeric characters by considering inter-class similarity measurement. The application was created using matlab.