A digital trajectory authentication method based on behavioral feature recognition

Handwritten signature authentication, a biometric authentication technology based on gesture behavior characteristics, is widely used for identity verification in fields like e-commerce, electronic contracts, finance, and legal sectors. However, it faces challenges such as high error rates, security vulnerabilities, and privacy concerns. To address these, this study designs a digital trajectory authentication method that leverages behavioral feature recognition. Utilizing the MediaPipe hand detection model, the method captures the air-writing trajectory through video analysis. Then, the model generates joint temporal feature descriptors from both time and frequency domains. Furthermore, a weighted probability matching strategy is adopted to construct a digital trajectory authentication model. Experiment results show that our method has an average authentication error rate (EER) of 3.04% on edge devices, which fully meets the accuracy of authentication recognition and is of great significance for the development of identity verification technology.