Improving handwritten character recognition by using a ranking-based feature selection approach

In handwriting recognition, because of the large variability of the writers, the selection of a suitable set of features is a challenging task. This has led to the development of a large variety of feature sets, which, in many cases, contain a large number of attributes, causing performance problems in terms of classification results and computational costs. In this paper, we considered a widely used set of features in handwriting recognition, to verify if it is possible to improve the classification results for handwriting recognition by using a reduced set of features. To this aim, we adopted a feature ranking based approach and tested several univariate measures. The experiments, performed on two real-world databases, confirmed the effectiveness of our proposal.