An evolutionary algorithm based Feature extraction and selection to Persian and Arabic Handwritten Recognition

Abstract

There are many feature extraction methods for handwritten letters. And selecting an effective subset of features is an important point in analyzing correlation rate in handwritten recognition. Feature selection is needed to select a subset of features that gives good recognition accuracy and has low computational overhead. In this article a methodology for feature selection in unsupervised learning is proposed. The main purpose of this article is enhancing characters recognition and classification, creating quick and low-cost classes, and eventually recognizing Persian and Arabic handwritten characters more accurately and faster. In this paper, to reduce feature dimensionality of datasets a hybrid approach using artificial neural network evolutionary algorithms algorithm is proposed that can be used to distinguish handwritten letters. A key property of our approach is that it does not require any a priori knowledge about the number of features to be used in the feature subset Implementation results show that evolutionary algorithm are applied here to improve the recognition speed as well as the recognition accuracy.