Angličtina
This volume highlights some of the multidisciplinary aspects of automatic signature verification. The first two chapters serve as an introduction. The first constitutes a review of the literature of the past five years, and the second addresses the problem of parallel strategies to construct and optimize feature vectors to describe a signature. The remaining six chapters are divided into two sections: research on static systems and research on dynamic systems. The section on off-line systems describes a system, based on co-operative neural networks for the automatic processing of signatures on cheques, a system for background removal as well as a model-based system that segments the signature into elements which are then processed locally and globally to evaluate the similarity between two specimens. The three papers on on-line systems, compare three verification methods, based on statistical models of signature features, using the same benchmark, describe a step-wise verification method based on the analysis of the signature components and describe the different design phases of an operational system, focusing on the various decisions that have to be made throughout the development of such a prototype.
