Pressure Analysis in Dynamic Handwriting for Forgery Detection

摘要

Handwritten signatures are extensively used over many fields for processing transactions and contracts. Online (dynamic) handwritten signatures have been used as a way of identification and there has also been intensive research done on distinguishing genuine signatures from forgeries for over past 40 years. However, dealing with high-skilled forgeries still remains a challenge. Therefore, handwritten signatures have not been used for authentication in electronic systems. Unlike static signatures, online handwritten signatures contain dynamic features and are able to identify individuals with more precision. The pressure value recorded during the online signature process has been a crucial information for automatic signature verification systems. In addition, the pressure an individual applies to mouse is also shown to enable continuous authentication. This paper focuses on the analysis of the pressure patterns during the process of dynamic signature. Comparing the pressure values for genuine signatures and skilled forgeries allows us to detect forgery while having the minimal information about the genuine signature. This fact significantly decreases the need of storing very sensitive biometric information with the potential to make the process writer independent.