Preliminary Study on Biometric Recognition Based on Drawing Tasks

Abstract

In this paper, we analyze the possibility to identify people using online handwritten tasks different from the classic ones (signature and text). Our preliminary results reveal that some drawings offer a reasonably good identification rate, which is 78.5% in the case of multi-sectional vector quantization applied to classify circles and house drawing tasks. To the best of our knowledge, this is the first paper devoted to biometric recognition based on drawing tasks.