Comparison of Feature Extraction Techniques for Handwriting Recognition

Abstract

Feature extraction is an important phase for image processing purposes since the output of the feature extraction is the input for classifiers. The importance of it applies to handwriting recognition problem, too. Distinctive features result in higher accuracy recognition of characters, or words. Therefore, it is crucial to be able to extract relevant and distinctive features from the image. In this study, we compare different feature extraction techniques for Hungarian handwriting recognition purpose. In order to be able to compare the techniques, the output of feature extraction phase is classifier using three classifiers namely, Support Vector Machines (SVM), Rough Sets Theory (RST) and Bayesian Networks (BN) using the WEKA machine learning tool. The results indicated that, the best classification results were retrieved using features calculated by the distribution of points in the image. However, it can be said that the combinations of different feature extraction types provide a greater deal of distinctiveness.