Quality Analysis for Online Signature Based on Entropy and Consistency over F-Tablet
Hao Chang · Journal of Information and Computational Science · 2013
A quantitative analysis for quality measure based on entropy and consistency was presented and evaluated on the online signature database acquired by our F-Tablet. Firstly, we carried out the work by presenting a survey of the recent literatures on the major factors affecting the signature acquisition process and quality measure methods. Noise of biometric sensor, physiological or behavioral factor of person and interaction between the users and the devices were three major factors impacting the signature acquisition process intrinsically. Then, the F-Tablet was briefly introduced. The F-Tablet was capable of capturing three perpendicular writing forces of the pen-tip to the contacting plane and two dimension torques directly. Writing forces are important dynamics of online signatures and they are invisible and harder to be imitated by forgers than signature shape. Finally, quality was measured on entropy and consistency in terms of trajectory and writing forces respectively. On one hand, performance assessment showed that consistency was principal component of quality measure and EER was positive correlated to quality. On the other hand, writing forces outperformed trajectory in terms of both entropy and consistency at the same level, which implied a huge information loss in mobile environment.