In order to confront the information security threats of smart phones,a behavioral biometric user authenti-cation mechanism for mobile devices based upon gestures was presented.It utilized embedded accelerometer to cap-ture the acceleration signal of a person’s authenticating gesture.After automatic gesture endpoint detection,a new al-gorithm based on wavelet packet and mutual information entropy was put forward for de-noising.Finally an improved Dynamic Time Warping method was proposed to measure the similarity between two gesture signal sequences,and the result of authentication was got from these similarities.Experimental results demonstrated that the False Rejection Rate was around 7%while the False Acceptance Rate was approaching 0%. Moreover this scheme has low time com-plexity,so effective and real-time user authentication on smart phones can be achieved.%为了应对智能手机所面临的信息安全威胁，提出一种基于行为生物特征的手势用户认证方案。实时采集手机内置三轴加速度传感器的数据，经有效手势端点检测得到认证数据，在信号去噪环节提出了一种结合小波包分解与互信息熵的新方法，最终由改进的动态时间规整算法进行手势信号序列相似性度量，从而得出认证结论。实验结果表明，当他人模仿手势错误接受率趋近0％时，本人认证手势错误拒绝率维持在7％左右，认证精度良好，同时算法时间复杂度低，可以实时有效对智能手机的持有者进行身份识别。
展开▼
