A new ECG feature extractor for biometric recognition

In this paper, a new wavelet based framework is developed and evaluated for automatic analysis of single lead electrocardiogram (ECG) for application in human recognition. The proposed system utilizes a robust preprocessing stage that enables it to handle noise and outliers so that it is directly applied on the raw ECG signal. Moreover, it is capable of handling ECGs regardless of the heart rate (HR) which renders making presumptions on the individual's stress level unnecessary. One of the novelties of this paper is the design of personalized heartbeat template so that the gallery set consists of only one heartbeart per subject. This substantial reduction of the gallery size, decreases the storage requirements of the system significantly. Furthermore, the classification process is speeded up by eliminating the need for dimensionality reduction techniques such as PCA or LDA. Experimental results for identification over PTB and MIT healthy ECG databases indicate a robust subject identification rate of 99.61% using only 2 heartbeats in average for each individual.