Handwritten signature recognition in mobile scenarios: Performance evaluation

Abstract

Following the idea of improving our previous work on dynamic handwritten signature recognition on portable devices, a performance evaluation in a mobile scenario was done. A database with 11 users and 8 mobile devices (using stylus and finger) has been collected in order to study different parameters such as screen size, operative system and the interoperability between the devices. The evaluation was divided by 3 sessions of 20 signatures per device each; 20 skilled forgeries signatures per user were used also. The devices used were mobile phones, tablets, laptops and two specific devices for signing. The algorithm used to assess the signatures was a DTW-based signature recognition algorithm.