Abstract
This paper introduces a new approach for recognizing cursively handwritten isolated Arabic characters. The approach is based on an active analysis of human handwriting movements, which was proposed and discussed extensively by the authors in a previous paper1). The handwriting process is modeled through autoregressive moving average (ARMA) model. Model Reference Adaptive Control System (MRACS) theory is applied, in order to extract control pulses of the handwriting movements. The control pulses, which describe in details the handwriting movement's behavior, are then used for recognizing the shape of the cursively handwritten character. Although the presented approach is proposed for recognizing handwritten Arabic characters, it can, however, be applied equally well to actively recognize any other cursively handwritten scripts.
