mmSign: mmWave-based Few-Shot Online Handwritten Signature Verification

• DOI: 10.1145/3605945
• Publication Date: 2023-06-24T11:14:38Z
• Authors: Mingda Han; Huanqi Yang; Tao Ni; Di Duan; Mengzhe Ruan; Yongliang Chen; Jia Zhang; Weitao Xu
• Keywords: Signature (topology)

Handwritten signature verification has become one of the most important document authentication methods that are widely used in financial, legal, and administrative sectors. Compared with offline based on static images, online handwritten more reliable because temporary dynamic information (e.g., signing velocity, writing force, stroke order) alleviates risk being forged. However, existing solutions reliant specific devices customized pens or pads) require extensive data collection during registration phase, resulting poor adaptability applicability for new users. In this article, we propose mmSign, a millimeter wave (mmWave)–based system, which enables accurate sensing user’s hand movements when through superior capability mmWave. mmSign extracts time-velocity feature maps from captured mmWave signals by carefully designed signal processing algorithms then exploits transformer-based model verification. addition, novel meta-learning strategy proposed task generation augmentation is introduced to teach learn effectively limited samples, allowing our quickly adapt Extensive experiments show robust, efficient, secure achieving 84.07%, 87.31%, 91.12%, 96.54% accuracy 1, 3, 5, 10 labeled signatures available, respectively, while resistant common forgery attacks.