This paper describes a system for performing handwritten sigture verification using complementary statistical models. The system alyses both the static features of a sigture (e.g., shape, slant, size), and its dymic features (e.g., velocity, pen-tip pressure, timing) to form a judgment about the signer's identity. This approach's novelty lies in combining output from existing Neural Network and Hidden Markov Model based sigture verification systems to improve the robustness of any specific approach used alone. The system performs reasobly well and achieves an overall error rate of 2:1% in the best case. The results of several other experiments are also presented including using less reference sigtures, allowing multiple signing attempts, zero-effort forgery attempts, providing visual feedback, and signing a password rather than a sigture.
Unless otherwise indicated, works by Griffith University Scholars are © Griffith University. For further details please refer to the University Intellectual Property Policy.