Performance evaluation of handwritten signature verification techniques

Abstract

Handwritten Signature Verification (HSV) is a biometric method of determining an individual's identity by examining the manner in which s/he performs his/her written signature. An Off-line HSV system examines static signature characteristics such as shape, slant and size. An On-line system verifies dynamic signature features such as velocity, pressure and timing. This paper describes important techniques and achievements in on-line HSV research and assesses their performance based on published literature. This work forms the basis for a dissertation that examines techniques specific to applying Neural Networks and Hidden Markov Models to on-line HSV systems (which is documented in complementary papers).