Handwriting Process Modelling by Artificial Neural Networks

Abstract

The handwriting is considered among the fastest and the most complex motor activities of our biological directory. This process also has a side which differentiates it from other human behavior as it is a physical manifestation of a complex cognitive process. Therefore, the modelling of a handwriting system is difficult to implement. Considering the complexity of the biological system involved in this process, several studies have been proposed in the literature based on different approaches. However, the validation results of these models remain unsatisfactory and the basic models have been improved to approach the reality as much as possible. This paper deals with new unconventional handwriting process characterization approaches based on the use of soft computing techniques namely the exploitation of artificial neural networks and more precisely the Radial Basis Function (RBF) neural networks. The obtained simulation results show a satisfactory agreement between responses of the developed RBF neural model and the experimental electromyographic signals (EMG) data for various letters and forms then the efficiency of the proposed approaches.