April 2, 2007 14:42 World Scientific Review Volume - 9in x 6in Main˙WorldSc˙IPR˙SAB
56 Synthesis and Analysis in Biometrics
and mining large volumes of signature biometrics. Most current state-of-
the-art solutions require proprietary hardware and are incompatible with
legacy systems (i.e. unable to verify the bitmap images of the signatures).
Moreover, none of the existing products solve the problem of identification
for large sets of data.
To address those problems, we created a system for signature
verification, identification and synthesis with flexible mechanisms of
database organization. The system offers the following enhancements over
existing packages in pervasive environments:
Advantage 1. An extended applicability to numerous security application
areas, e.g. personal security, financial transactions and document
automation.
Advantage 2. Rapid retrieval of information based on the attributes
stored with the image and/or on relevant visual attributes.
Advantage 3. A unified approach to the design of image processing tools
that are independent of image resolution, size or orientation.
Advantage 4. A software package which is able to process human
biometrics of different types (e.g., handwriting, signatures, voice, etc.).
Advantage 5. An easy-to-use environment for signature verification and
most importantly identification available “on demand” and accessible
online.
The functionality of the system is enabled through the integration
of three main components: (1) a unified data format with supporting
mathematical models, (2) an image processing toolbox with image
generation capabilities, and (3) a database of images with self-learning
capabilities. The diagrammatic structure of the system is depicted in
Fig. 2.14.
Data acquisition. Signatures can be obtained by various image
acquisition devices. Dynamic signature acquisition records kinematic
characteristics of human handwriting such as pressure, speed, acceleration,
duration of writing, etc. Dynamic acquisition devices include numerous
tablets, writing pads, etc. Pen-based interfaces are becoming more
integrated in human-computer interfaces because of overall convenience
of pen enabled devices vs. traditional keyboards and mice. There are
various pen-enabled devices ranging from pressure sensitive tablets to
gyropens. Recently developed devices include handheld computers (PDAs
and smart phones), and Tablet PCs. In the study presented here, we use
a Wacom compatible pressure sensitive tablet. It is set to provide
100 samples per second containing values for pen coordinates, pressure,