COMPONENT-ORIENTED ALGORITHMS FOR SIGNATURE VERIFICATION

Signing is a complex and highly variable process. It results from simple bodily motions of a ballistic nature whose combined effects produce the fundamental components of the signature. Therefore, signature verification can be performed by adopting local verification strategies for the detection of personal characteristics in fundamental components.

In this paper, an on-line component-oriented signature verification system is presented. During the training phase, the set of fundamental components of each signer is derived and a suitable component-oriented knowledge-base is created by an automated technique. The verification is accomplished through a step-wise process. In the first step the structural organization of the signature is checked. In the second step each component is examined using spectral analysis. The experimental result shows the effectiveness of this approach compared with other techniques in the literature.