基于动态阈值和DTW的ECG身份识别算法

人体心电信号(ECG)虽然是非稳态时变信号,但现有研究表明其可以作为一种新型的生物识别特征用于对识别分辨率要求不高的应用场合,或与传统生物特征相结合来提高识别性能。本文给出了一种能针对不同实验对象动态设定阈值以提取典型波形,并利用加权动态时间弯曲法(DTW)进行ECG身份识别的算法。在模板登记阶段,首先对经去噪处理的一段ECG波形进行R点检测,然后通过动态阈值设定得到一段典型的包含完整心电周期信息的心电波形,最后将其作为模板波形存入用户模板库。在身份识别阶段,基于前述经动态设定的阈值进行心电波形选择,实时提取测试波形,然后基于经长度加权的DTW算法进行测试波形和模板波形的相似度计算,最终实现用户身份识别。

ECG Identification Based on Dynamic Threshold and DTW Algorithm

Existing studies have shown that non-stationary and time-varying ECG signals can be used as a new biometric trait for human recognition in situations with a less-demanding resolution. This study proposes a dynamic setting of thresholds and dynamic time warping （DTW） algorithm based ECG identification. During the stage of template recording , R peaks are firstly detected from ECG signal with de-noising preprocess, then a waveform of ECG containing a complete period of heartbeats is obtained by setting dynamic thresholds , followed by the storage of the obtained typical waveform in the database. During the stage of identification , testing waves are obtained in the real time by using the above mentioned determined thresholds , then the length weighted DTW algorithm is applied to calculate the difference between the test and template waveforms to finally reach the results of human identification.