A New Method of Verification of Dynamic Signatures Changing over Time with Decomposition and Selection of Characteristic Descriptors

Dynamic signature verification (DSV) has practical applications in banking, commerce, accounting, medicine, law, etc. DSV is important in comparison to the verification of other biometric attributes because the dynamic signature is one of the most widespread and commonly used biometric features. DSV methods are being systematically developed and improved. At the same time, expectations regarding these methods are changing. For example, they are expected to take into account changes of the individual's signature appearing over time. In this paper, we propose a new DSV method that uses partitioning supported by partitions' selection and characterized by the use of signatures created in different acquisition sessions in the training phase. This algorithm has been tested using a publically available, known xLongSignDB, dynamic signature database. In the simulations very good results were obtained.