基于心电图的身份识别方法

Human Identification Approach Based on ECG

针对当前心电图(ECG)身份识别中存在的小样本、多特征点检测问题,提出基于小波变换和动态时间规整(DTW)相结合的方法。利用小波变换对ECG信号进行预处理并提取R波峰值点,提取并保存肢导联QRS波及心拍模板,根据QRS波测试数据与各QRS波模板间的相关性分析以及阈值条件缩小身份识别范围,采用DTW算法确定心拍测试数据与各心拍模板之间的最优匹配距离,实现身份识别。实验结果表明,该方法在112个个体ECG数据中的身份识别准确率为97.3%,个体识别时间为4.4 s,解决了单检测点和大样本条件下的ECG身份识别问题。

In view of the problems in current human identification based on Electrocardiography（ECG）, such as small sample and multi-feature points detection, a novel approach using wavelet and Dynamic Time Warping（DTW） for human identification is investigated. The noise is reduced in preprocessing and the R peaks are detected by using wavelet transform. Then the reference templates of QRS wave and heart beat in limb leads are extracted and saved. Candidates are selected by the result of correction analysis between the test and templates of QRS waves. The only candidate is selected by the minimum distance between the test and templates of heart beat in limb leads using DTW. Experimental results show that the identification accuracy is 97.3% in 112 individuals and individual identification time is 4.4 s, and the proposed method solves the problems of human identification based on ECG under single feature point detection and large sample.