Biometric technologies are increasingly used for authentication purposes instead of passwords or cards. Different biometrics are suitable for different applications; for example, retina or iris can be used in applications requiring very high levels of security, while fingerprints are used in applications requiring more moderate security. The signature is a biometrics that is well-accepted, easily collected and widely used in banking applications. The challenge of the signature verification problem is to find the most appropriate computer vision and machine learning approaches to learn the natural variations of a person’s signature in very few (3-5 signatures) and to distinguish imitations from genuine signatures.

Online and offline signature verification systems developed by Prof. Yanıkoğlu and his two doctoral students have ranked first or second place in almost all international signature recognition competitions since 2004 (SVC2004, SigComp2011, 4NsigComp2010, SigWIComp2013, SigWIComp2015). In these competitions, researchers and companies are evaluated by organizers of the contest, so technologies in a certain area can be compared fairly. The success of the systems developed by Prof. Yanikoglu and his students is over 95% for online and 85% for offline. For more information, please see http://people.sabanciuniv.edu/berrin.