New Heart Features for More Effective Human Identification

Abstract

Biometric verification is a process to authenticate whether the subject is what it claims to be, based on the characteristics of the human body. These characteristics must meet seven (7) conditions to enable them to be used in a practical biometric system. These conditions are namely distinctiveness, performance, collectability, acceptability, universality, circumvention, and permanence. Electrocardiogram (ECG) is a human body characteristic measured from the heart that meets these seven conditions. Choosing the correct features from the ECG is important to get a high authentication rate. This paper proposed a new algorithm known as Bipolar Slope Feature (BSF) for ECG features selection. It is developed based on the relationship of slopes between several locations in a complete ECG cycle. The Receiver Operating Characteristic (ROC) curve is used to measure the effectiveness of this technique for the application of biometric verification.