What is biometrics?
A biometric is a substantially stable physical or behavioral characteristic or trait of a person which can be automatically measured and characterized for comparison. In the context of signatures, biometrics are physiological and behavioral characteristics that make a signature unique.
Over 2000 years ago, Aristotle noticed the connection between handwriting and personality. How we make our loops in letters, the alignment of upper, middle, and lower sections of text, letter spacing, and many other characteristics of handwriting are inherent in each individual. Accordingly, a signature or the way people write their names is unique and cannot be repeated by others. This phenomenon originated a centuries-old tradition for important documents to be signed as proof of their authenticity.
Comprehensive signature verification systems analyze two different areas of an individual’s signature:
1) the specific features of a static image of one’s signature, which are the reflection of person’s biometrics,
2) the specific features of the dynamic process of signing, which reflect physical and behavioral peculiarity.
The first type includes applications dealing with the graphical, two-dimensional static image of the signature resulting from an action of signing that has already taken place. The image of a signature results from a series of actions inherent to a particular individual and therefore is an impression of the process of signing that bears unique characteristics of this process and reflects personal peculiarities and traits. However, in these applications the signature’s biometric characteristics of motion itself are unobtainable, and only its static, two-dimensional image is available for verification. Therefore, there are ways to make a copy of a signature (for example tracing) that are nearly undistinguishable from an original. Systems that analyze only the static data of a signature image are called off-line.
The second type embraces applications that allow tracking the motion in the process of signing at the point of presentation. Accordingly, systems that treats the signature as a series of movements and can be used for locally or remotely originated transactions. These are called online or dynamic.
Can biometrics help identify forged signatures?
The key to the online verification of the authenticity of questionable signatures lies in the reconstruction of the writing motion and its elements. Signing is a reflex action based on prior repeated experience (training) and not influenced by deliberate muscular control. In particular, when signing, the hand often moves faster than an individual could volitionally control it to move through hand-muscle coordination. The practiced and natural motion of the original signer would be required to repeat the signature pattern. A copy machine or an expert forger may be able to duplicate what a signature looks like, but it is virtually impossible to mimic such unique behavioral patterns and characteristics of the original signer as succession of touches to the writing surface, speed, acceleration, and pressure. Thus, for dynamic signature verification a handwritten signature is recorded/captured using a variety of pen-enabled devices such as digitizing tablets, membrane touchpads, capacitive touchpads, LCD touchscreens, computer displays or other contact-sensitive technologies.
During the act of signing, a signature is captured and elements and behavioral characteristics that make it unique and identifiable are derived. Signature verification checks biometric characteristics of a questionable signature against biometric characteristics of reference signature(s) and can be executed either in real-time or afterwards. The availability of behavioral characteristics of the signing process that unambiguously distinguish an individual makes it feasible to create robust signature verification systems. Based on this approach, a number of dynamic signature verification products demonstrate similar efficiency, which in most cases is a direct function of the quality of the utilized writing tablet.
