Human Identification Using Heartbeat Interval Features and ECG Morphology

摘要

This paper presents a novel method to characterize the ECG signal for human identification. The characterization process utilizes the analytical and appearance based techniques to analyze the ECG signal with an aim to make the measurements insensitive to noise and non-signal artifacts. We extract heartbeat interval features and interbeat interval features using analytical based technique and use them as a complementary information with the morphological features that are extracted using appearance based technique for improved identification accuracy. We perform identification using one-to-many comparisons based on match scores that are generated using statistical pattern matching technique. Results demonstrate that the proposed method for automated characterization of the ECG signal is efficiently used in identifying the normal as well as the arrhythmia subjects. In particular, the recognition accuracy for the subjects of MIT-BIH Arrhythmia database is reported to 87.37% whereas the subjects of our IIT(BHU) database are recognized with an accuracy of 92.88%.