ECG biometrics: A template selection approach

Abstract

Electrocardiography (ECG) biometrics is emerging as a viable biometric trait. Recent developments at the sensor level have shown the feasibility of performing signal acquisition at the fingers and hand palms, using one-lead sensor technology and dry electrodes. These new locations lead to ECG signals with lower signal to noise ratio and more prone to noise artifacts; the heart rate variability is another of the major challenges of this biometric trait. In this paper we propose a novel approach to ECG biometrics, with the purpose of reducing the computational complexity and increasing the robustness of the recognition process enabling the fusion of information across sessions. Our approach is based on clustering, grouping individual heartbeats based on their morphology. We study several methods to perform automatic template selection and account for variations observed in a person's biometric data. This approach allows the identification of different template groupings, taking into account the heart rate variability, and the removal of outliers due to noise artifacts. Experimental evaluation on real world data demonstrates the advantages of our approach.

Heartbeats in the Wild: A Field Study Exploring ECG Biometrics in Everyday Life

“…We attribute this to the variance of heartbeats which might vary between days or even shorter timeframes. This is in line with existing work [40,46,51]. In addition, in the wild we expect extra noise due to varying real life contexts, such as activities, body postures and movements, and so on.…”

“…In terms of study design, a key difference between ECG biometrics studies is the number of sessions of data recording: Some studies collected data from one session only [31,41,42,45,49], where others used two [2,10,32,40], or more sessions [1,4,51]. These were all lab sessions; only two studies explored ECG biometrics in the wild [56,64].…”

“…Alternatively, new data can be acquired from participants in experimental setups. This has been done with medical-grade devices [4], custom prototypes [10,13,33,41,1 PhysioBank Databases: https://physionet.org/physiobank/ database 42,40,45], consumer devices [2,11,31,47], or wearables [64]. In some cases, researchers highlighted the resulting "in-house databases" with regard to the need for standardised datasets and evaluations for ECG biometrics [1,47,51].…”

“…Features can be extracted based on the segmented cardiac cycles: The fiducial points can be used to determine temporal features, amplitude features, as well as angles [59]. Feature representations are then utilised for decision-making via classification, including random forest classifiers (RFC) [59], support vector machines (SVM) [11,42,40,64], k-nearest neighbors (kNN) [1,42,41], and neural networks (NN) [10,49,54,60,66]. Work by Choi et al provides an overview, evaluating nine different classifiers for ECG biometrics [11].…”