Offline Cursive Word Recognition using Continuous Density Hidden Markov Models trained with PCA or ICA Features

This work presents an Offline Cursive Word Recognition System dealing with single writer samples. The system is a continuous density hiddden Markov model trained using either the raw data, or data transformed using Principal Component Analysis or Independent Component Analysis. Both techniques significantly improved the recognition rate of the system. Preprocessing, normalization and feature extraction are described in detail as well as the training technique adopted. Several experiments were performed using a publicly available database. The accuracy obtained is the highest presented in the literature over the same data.