ECG as biometric using autocorrelation of normalized R-R interval

A technique has been presented to use ECG (Electrocardiogram) for human identification. Earlier researches were tested on PTB or MIT-BIH ECG data which was recorded once a person is at rest and will give erroneous results if the heart beat rate changes depending upon human activity. Hence time and amplitude normalization is necessary for identification. In this technique, R peaks of ECG signal are detected and R-R interval is then resampled to consider a few in order to proceed further. Time and amplitude normalization is achieved in order to compensate for the variation in heart beat rates. The proposed technique was implemented on real world, self collected data taken from Armed Forces Institute of cardiology (AFIC) and 100 % correct results were obtained with the data set of 20 healthy persons.