基于加速度特征点提取的步态身份认证

Gait authentication based on acceleration characteristic point extraction

为了提高现有的基于加速度计的步态身份认证方法的性能,提出了一种基于加速度特征点提取的步态身份认证方法。使用基于小波变换的方法提取步态加速度特征点,并使用动态时间规整算法对提取出的特征点序列进行匹配。为了验证该方法的性能,建立了一套步态加速度无线采集系统并采集了18个志愿者的步态信息。实验结果表明,该方法的身份认证平均错误率为6.9%。与现有方法相比,该方法有效地减少了步态模板的数据存储量,缩短了匹配时间,提高了认证算法的性能。

An acceleration characteristic point extraction algorithm was developed to improve the performance of the conventional accelerator based gait authentication. The algorithm uses wavelet transformations to extract the characteristic points from acceleration sequences and matches them using the dynamic time warping method. The algorithm performance was evaluated using a wireless gait acceleration measurement system with measurements of gait information on 18 volunteers. The results show that the equal error rate is 6.9%. This algorithm has better gait verification efficiency than conventional methods by not only reducing the amount of gait template data storage but also shortening the duration of the authentication phase.