Multiple Classifiers to verify the Online Signature

Nowadays biometric increasingly used in many applications that has strong relation to our live; it's a reliable mean as an alternative to the traditional methods of personal identification. As a behavioral biometric, an online signature still has some shortcomings because of that nature. Furthermore, features in online signature verification system can be either global or local; the techniques that can be used also variety. In this paper both global and local features were used. To classify the mentioned features; the back-propagation neural network (BPNN) technique was used to classify the local features, whereas, the global features was classified by the probabilistic model. Once the results obtained from the local classifier and global classifier, the “AND” fusion was used to combine the two classifiers for final decision. SVC2004 dataset was used to evaluate the proposed method in term of False Rejection Rate (FRR) and False Acceptance Rate (FAR). The obtained results for FRR and FAR were 0.3% and 0.5% respectively. These results are encouraging when compared with related existing studies.