AirSign: An innovative VR-based hand-written signature recognition

Abstract

Traditional authentication methods in virtual reality (VR), such as password entry or smartphone-based verification, disrupt immersive experiences and introduce usability challenges. This paper presents a preliminary VR authentication system prototype that leverages dynamic 3D hand-drawn signatures captured via headset-integrated hand-tracking or motion controllers. By converting hand trajectories into unique spatiotemporal point clouds, the system employs Dynamic Time Warping (DTW) to verify signatures directly on-device, eliminating reliance on external hardware or confined 2D input surfaces. Key contributions include: (1) unrestricted 3D signature capture within the VR environment, contrasting with prior approaches requiring predefined 2D planes or specialized sensors (e.g., LeapMotion); (2) a lightweight DTW-based algorithm optimized for real-time processing on resource-constrained VR hardware, demanding low computational overhead; and (3) seamless integration with native handtracking, preserving immersion and minimizing user friction. Experimental validation demonstrates the system’s robustness across 500 unique signatures from 28 users, achieving authentication accuracy suitable for the target application context. Latency benchmarks confirm execution within 350 ms per verification cycle (including data normalization, feature extraction, and the optimized DTW matching process), with an additional power draw of< 2% on standalone headsets. The results show an Equal Error Rate (EER) between 0.030 and 0.165, corresponding to an authentication accuracy of approximately 83.5% to 97% in the most challenging scenarios, depending on the sample set and the input device used: combinations of controllers vs. hand-tracking, and fixed vs. custom signature. The proposed experimental protocol provides preliminary evidence regarding the method’s potential. Also, a fixed-word signature contributed to creating a hostile scenario in which a potential forgery attack was carried out. By decoupling authentication from fixed spatial constraints and prioritizing computational efficiency, this work demonstrates the feasibility of on-device biometric authentication, offering a scalable solution for secure, immersive user identification.