ECG Biometric Analysis Using Walsh–Hadamard Transform

Abstract

The electrocardiogram (ECG) signal expresses unique cardiac features among individuals. This paper proposes a novel method to human identification using ECG. The proposed method utilizes a band-pass filter for quality check and autocorrelation (AC) for feature extraction. Furthermore, the Walsh–Hadamard transform (WHT) is used for feature transformation. To get cost- and time-efficient classification performance, the dimensionality of feature vector is reduced using linear discriminant analysis (LDA). Experimental results show the best identification rate of 95 and 97% over MIT-BIH arrhythmia database and QT database, respectively.