Neural networks and support vector machines classifiers for writer identification using Arabic script

Abstract EN

In this paper, we present an approach for writer identification carried out using off-line Arabic handwriting.

Our proposed method is based on the combination of global and structural features.

We used genetic algorithm for feature subset selection in order to eliminate the redundant and irrelevant ones.

A comparative evaluation between two classifiers is done using Support Vector Machines and Multilayer Perceptron (MLP).

The best results have been achieved using optimal feature subset and MLP with an average rate of 94%.

Experiments have been carried out on a database of 120 text samples.

The choice of the text samples was made to ensure the involvement of the various internal shapes and letter locations within asubword.