This invention allows the verification and identification of a person based on features extracted from one or more electrocardiographic leads or channels. For identification purposes, the invention can be used to identify a subject from a group of known subjects. For verification purposes, the invention can be used to allow access to secured facilities or remote services over the Internet to allow access to or control of computers, cars, airplanes, ships, and submarines; to enable/disable electronic alarms; to open safes or security deposits; and to enable guns and special weapons that require restricted access.
The purpose of this work was to build an automatic means of verifying the identity of a human subject by recording a sample of the electrocardiographic signals (PQRST) generated by the person’s heart, and comparing this sample with the previously recorded electrocardiographic features from the claimed person (subject verification) or from a group of known persons (subject identification).
The method of subject verification or identification uses as biometric modality the electrocardiographic signals generated by the heart during its contraction (depolarization) and relaxation (repolarization). Normally, these two phases are repeated continuously with a periodicity that reflects the heart rate. The method automatically extracts from one or more electrocardiographic leads (channels) a set of biometric features that are characteristic to an individual and can be employed to verify the identity of an individual or to identify an individual from a group of known individuals. The electrocardiographic signal is recorded in each lead from a pair of electrodes placed on the arms or legs; for example, between right leg and left arm or between right arm and left leg.
Each electrocardiographic lead characterizes the heart’s electrical vector as projected on a separate dimension. This allows a multidimensional characterization of the heart’s electrical activity that can result in improved subject verification performance.
The advantage of the invention is that it creates a probabilistic model of the electrocardiographic features of a person instead of a single signal template of the average heartbeat. The probabilistic model, described as Gaussian mixture model, allows the various mixtures to model various modes of the feature distribution in contrast to a template model that only characterizes a mean waveform. Another advantage of the invention is that it uses both physiological and anatomical characterization of the heart, unlike other methods that mainly use only physiological characterization.