The handwritten signature as a biometric identifier: psychophysical model and system design

The paper presents a signature generation model based on a few fundamental properties of the human neuromuscular system. It suggests some causes of the intrinsic stability and variability of signatures and provides conditions for using the signature as a biometric identifier. A brief critical survey of the different online approaches that have been used to tackle these problems over the past decades is then presented. Finally, a complete system based on the previous neuromuscular model is described, with special emphasis on the various decisions that have been made throughout the design process. The system (US Patents 5,077,802 and 5,101,437) works in a PC compatible environment. In registration mode, it quantifies the stability and variability of three reference signatures of a subject. In verification mode, the system segments and analyzes any test signature using a personalized, multilevel comparison algorithm. A decision about the authenticity of the test signature is generally made within a couple of seconds, based on global parameters, velocity functions and image features. We conclude with an overview of current work in the field.