ON-LINE SIGNATURE RECOGNITION USING A GLOBAL FEATURES FUSION APPROACH

The paper presents a feature-based approach regarding on-line signature recognition, by employing the TESPAR - DZ matrices extracted from every signature. These matrices describe the different signature features shapes, such as x, y - coordinates, x, y - velocity, pressure etc. For better results, the fusion of different features was studied. Three signature databases have been employed in our experiments: DB1, DB2, our own signature databases and the Task2 database available for research purposes from the SVC2004 contest. In our work, two types of experiments were made: verification and identification (surveillance). These results are very promising taking into account the simplicity of template generation and the fact that we employed only five features in our experiments.