ECG biometric identification for general population using multiresolution analysis of DWT based features

Electrocardiogram (ECG) is not only a vital sign of life but also contains important clinical information and even identical features. Similarly, ECG provides various significant characteristics to advocate its use as a biometric system such as uniqueness, permanence and liveness detection etc. This research provides with the complete systematic approach of ECG based person identification for general population and consists of preprocessing of signal for noise reduction, feature extraction, feature selection and classifier performance. Feature extraction was performed by extraction of cardiac cycle followed by discrete wavelet transform (DWT) to extract wavelet coefficients as feature vector. Feature reduction is performed with best first search and classification is performed by using single nearest neighbor classifier. System is tested on three publicly available databases like MIT-BIH/Arrhythmia (MITDB), MIT-BIH/Normal Sinus Rhythm (NSRDB) and ECG-ID database (ECG-IDDB) including all subjects both separately and in combined manner. For arrhythmic database, identification rate of 93.1% was achieved by using proposed methodology. System is also tested on normal population based databases and accuracy of 99.4% is achieved using NSRDB database and 82.3% for a challenging ECG-ID database. The combined approach for general population results in accuracy of 94.4% with false acceptance rate (FAR) of 5.1% and false rejection rate of 0.1%, proving the effectiveness of suggested approach as non invasive for general population with better results as compared to previous approaches in literature.