This book deals with new methods for fingerprint verification. With the advent of public key
infrastructure and e-commerce the demand for reliable user authentication has increased by
leaps and bounds. Authentication via PIN and passwords or tokens may become no longer suitable
to meet the security constraints. Biometrics is the science of measuring physical or behavioral
characteristics that are suitable for the identification of individuals. Fingerprint
verification is one of the most reliable and well-understood biometrics measures. Physical
characteristics cannot be stolen forgotten spied out or transferred. They are tied to the
user and permit the implementation of high security systems that are convenient to use. The
design of professional security systems enhanced with biometric user authentication implies
that the verifying device is a tamper-proof embedded system. Therefore it is valuable to
perform the verification process in a secure manner with microcontrollers as used in common
low-cost security tokens like smart cards. Several concepts for implementing fingerprint
verification within small microcontrollers were developed enhanced and realized throughout
this work. The memory constraints and limited processing power of smart cards and compatible
tokens were overcome as reported by tests with databases of fingerprints. The results and the
strengths as well as limitations of every method yield conclusions on how to implement secure
biometric user authentication with security tokens. The transaction schemes for secure user
authentication with biometrics show a major improvement in system security and privacy. This
book shows that fingerprint verification is possible with standard microcontrollers in today's
security tokens. Several original algorithms were developed implemented and tested against
each other. Furthermore the standard algorithms for fingerprint verification were enhanced
with some new aspects to improve the overall matching accuracy. Particularly a flexible
multi-resolution level fingerprint matching algorithm was derived and implemented. It combines
the verification of minutiae ridge counts and pores while keeping the template storage
requirements in a range that is appropriate for embedded tokens.
