基于形态特征和KPCA融合特征的ECG身份识别

ECG Identification Based on Fusion Features of Morphological Characteristics and KPCA

ECG信号由于其唯一性,在身份识别中得到了越来越广泛的应用。但以往的研究基本只讨论在平静状态或不同情绪下的身份识别,没有考虑个体在运动中及运动后的识别情况。本文针对这个问题,研究在平静和运动状态混合下ECG身份识别的特征向量选取问题。实验分别提取QRS等特征点组成的形态特征、核主成分以及两者的融合特征,利用支持向量机(SVM)进行识别测试。其中训练使用平静状态下的数据,测试则使用平静与运动状态混合的数据集。实验结果显示当ECG身份识别扩展到平静和运动状态混合的情况下,形态特征和KPCA融合的特征有最优的识别效果,识别率达到98.7342%。

Due to its uniqueness, ECG signal has been more and more widely used in identity recognition. However, the studies before only discuss the identification in the rest state or under different emotion, without considering the identification during or after movement. To address this problem, we research on the choice of the feature vector of ECG identity recognition under the mixed state of rest and motion states. Morphological characteristics, KPCA and fusion features are extracted in the experiment, respectively, to do the identification using SVM. In the experiment, training uses the data collected in the rest state, while the testing uses the data set collected both in rest state and motion state. The experiment results show that when the ECG identification extends to the situation where the rest state is mixed with motion state, the fusion features of morphological characteristics and KPCA has the optimal identification result, achieving a recognition rate of 98.7342%.