Human identification method using time and wavelet domain features based on modified dECG

In this paper, an effective method of human identification is proposed based on time-frequency domain features extracted from modified differential electrocardiogram (dECG) signal. In comparison to the ECG data, the discrimination in terms of QRS complex among different persons is more prominent in the dECG signal. It is shown that the use of a modified dECG signal can further enhance the level of discrimination as it also includes the effect of P and T waves. First, in order to obtain time domain features, a number of reflection coefficients are extracted from the modified dECG signal using Yule-Walker based algorithm. Next, the discrete wavelet transform (DWT) of different levels are employed on the modified dECG signal to obtain time-frequency domain features. It is shown that, instead of using the proposed two features separately, use of combined features can provide high within class compactness and between class separation. In the recognition phase, a linear discriminant based classifier is employed, where the leave-one-out cross validation technique is utilized. The proposed human identification method has been tested on standard ECG database and high recognition accuracy is achieved with a low feature dimension.