Pattern recognition and machine learning
Pattern recognition and machine learning
On-line signature verification using LPC cepstrum and neural networks
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
A study on enhanced dynamic signature verification for the embedded system
BVAI'05 Proceedings of the First international conference on Brain, Vision, and Artificial Intelligence
| Hi-index | 0.00 |
Automatic verification of handwritten signatures hasnumerous applications in checking the authenticity and validityof cheques and documents. Intra-class differences betweengenuine signatures and difficulty in collecting representativeforgeries for comparison have been the main obstacles for itspractical implementation. In this paper, a new standpoint ofpaying selective attention to the stable parts of genuinesignatures is proposed to overcome the obstacles, and an experimental system based on it is given. To realize theselective attention, two strategies are addressed. One is totrain the classifier with artificial forgeries generated byremoving stable components from genuine signatures, so that theclassifier can detect these stable components when verifyingsignatures. The other is to force the neural network classifierto pay special attention to local stable parts of signatures by weighting their corresponding node responses through a feedbackmechanism. The experimental results demonstrate the potential ofthe proposed approach to compensate for the lack ofrepresentative forgeries for system training, and in improvingthe system‘s ability to identify skilled forgeries.