Abstract
Biometric identification techniques are widely used as individual identification methods in security systems. We have studied the high frequency component of electrocardiogram (HFECG) as a new biometric modality. In this technique, a HFECG segment containing individual characteristics is extracted and used for identification. Identification performance depends greatly on extraction accuracy, but the current extraction method using the peak point on R waves as fiducial points can result in unsatisfactory performance. In this study, we propose a new fiducial point determination technique utilizing waveforms transformed from ECGs. The algorithm is based on differential calculus and produces sufficient performance despite having less complexity than other signal detection techniques. Comparative evaluation established that identification performance was improved using the new method.
