Total Variation PCA-Based Descriptors for Electrocardiography Identity Recognition

Electrocardiographic(ECG) signals have been successfully used for biometric recognition. However, the accuracy of ECG-based biometric systems is lower than systems based on other physiological traits. To improve the performance of ECG-based biometric systems, in this study, we propose a local feature learning method for ECG biometric recognition. Specifically, we first extract the multi-scale differential feature(MDF) for each point in the training ECG heartbeats using the difference between each point and its neighboring points. Second, we learn a feature mapping to project these MDFs into low-dimensional descriptors in an unsupervised manner, where 1) The error between the original MDF and reconstructed MDF is minimized. 2) The total variation in the reconstructed MDFs was minimized. Third, we represented each ECG heartbeat as a histogram feature by clustering and pooling these descriptors. Finally, we adopted global feature learning methods to obtain a representation of the ECG heartbeat. Experiments on the MIT-BIH Arrhythmia database, ECG-ID database, and Physikalisch Technische Bundesanstalt database verified the performance of the proposed method over other existing ECG biometric recognition methods by within-session analysis. Moreover, we evaluated the performance of our proposed method using across-session analysis of the ECG-ID database.