MOBILE DEVICE HANDWRITTEN SIGNATURE VERIFICATION BASED ON MODIFIED FACET ANALYSIS METHOD
| Autor: | Hao-Wei Lee, 李晧暐 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 This paper proposed a method for signature verification and made experiments on mobile devices. In the application of online signing such as shopping with credit cards, real-time verification is imperative to prevent forgeries. When we open an account in a financial institution, we need to sign our names several times in various forms that can be used as a basis for extracting features for future comparison. Dynamic attribute values such as XY-coordinates, T(time duration) and P(pen pressure) are collected. Based on the modified Facet Analysis Method, we are able to calculate the average similarity of the signatures. By calculating the Euclidean distance, we may determine the maximum allowable distance, the ceiling, from the center of the basis signature set. By comparing with a signature to be verified, if the Euclidean distance is small than the ceiling, then system accepts the signature. The system works automatically without human intervention. Experimental results show that the false rejection rate is 11.7％ and the false acceptance rate is 7.6％.The low false acceptance rate recommends the applicability of the system. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|