Research on ECG Biometric in Cardiac Irregularity Conditions

This paper studies the principle of ECG signals applied to identification, particularly considers the case of users´ ECG abnormal conditions. This paper presents an improved multi-template matching algorithm for identification, which can achieve good discrimination effects under ECG abnormality. Normal and abnormal ECG templates are constructed by QRS complex, the discrimination is based on the correlation coefficient of the testing data and template. We used 44 ECG data files from the MIT-BIH Arrhythmia Database (MITDB) to measure the performance of the algorithm, extracted normal templates in 18 data files as well as normal and abnormal templates in the remaining 26 data files. The experiment obtained an 88.06% accuracy of template matching, when considering the discrimination results of all the testing data belong to one user, the individual recognition accuracy reaches 100%. Experiments showed that the improved multi-template matching algorithm characterized by QRS complex can be used to identify individuals in the state of arrhythmia.