DefinePK

Abstract

Computational efficiency is a matter of great concern in state-of-the-art English alphabet character recognition systems. In this paper, nine state Hidden Markov Model (HMM) for character recognition has been presented. Alphabetical character images are being divided into nine blocks that corresponds to nine respective states of HMM. Corresponding local features of the character are being extracted by using geometric based feature extraction algorithm. Training of the HMM is done by means of the Baum-Welch algorithm. Computational cost of proposed model is minimized by employing Discrete Wavelet Transform (DWT) prior to other dimensionality reduction techniques. The recognition is performed using a Viterbi algorithm to perform best path search in combinations of various character models. Experimental results on handwritten English alphabet character databases demonstrate that recognition accuracy of proposed model is comparable to the existing techniques with reduced computational cost.