Multi-scale differential feature for ECG biometrics with collective matrix factorization

doi:10.1016/j.patcog.2020.107211

Pattern Recognition

Volume 102, June 2020, 107211

https://doi.org/10.1016/j.patcog.2020.107211 Get rights and content

Highlights

•
A novel Multi-Scale Differential Feature for ECG Biometrics with Collective Matrix Factorization is proposed.
•
The micro texture and multi-scale differential signal characteristics of ECG is efficiently captured.
•
The intra-subject and inter-subject similarities are maximally preserved.
•
The extracted discriminative ECG representation is more descriptive and robust towards noise.

Abstract

Electrocardiogram (ECG) biometrics has recently received considerable attention and is considered to be a promising biometric trait. Although some promising results on ECG biometrics have been reported, it is still challenging to perform this technique robustly and precisely. To address these issues, this paper presents a novel ECG biometrics framework: Multi-Scale Differential Feature for ECG biometrics with Collective Matrix Factorization (CMF). First, we extract the Multi-Scale Differential Feature (MSDF) from the one-dimensional ECG signal and then fuse MSDF with 1DMRLBP to generate the MSDF-1DMRLBP, which acts as the base feature of the ECG signal. Second, to extract discriminative information from the intermediate base features, we leverage the CMF technique to generate the final robust ECG representations by simultaneously embedding MSDF-1DMRLBP and label information. Consequently, the final robust features could preserve the intra-subject and inter-subject similarities. Extensive experiments are conducted on four ECG databases, and the results demonstrate that the proposed method can outperform the state-of-the-art in terms of both accuracy and efficiency.

Introduction

Biometrics is the field of study that models the identity of people using their physiological or behavioral traits [1], such as face [2], fingerprint [3], iris [4], finger vein [5], gait [6], and so forth. Over the past decade, some biosignals that are generally used for medical purposes, including but not limited to electrocardiogram (ECG) [7], electromyogram (EMG) [8] and electroencephalogram (EEG) [9], have been examined as biometrics for multiple reasons.

Compared with other biometric traits, ECG biometrics have several unique advantages, which can be summarized as follows. 1) ECG is recorded by attaching sensors to the body, which means that it can only be captured from a living person. 2) ECG is difficult to replicate or spoof, leading to the high security of ECG biometrics. 3) ECG signals can be acquired from all living individuals regardless of whether they are healthy; thus, ECG biometrics are universal for all people. In addition, ECG signal-capturing devices are becoming increasingly portable and inexpensive, such as wearable devices. The acquisition method makes ECG biometrics superior to several other biometric systems because acquiring other biometric signals may distract the user. 4) The acquired ECG is a one-dimensional signal that is easy to store and process.

G.E. Forsen et al. [10] first proposed using ECG as a human identification trait in 1977, and the most influential work on ECG biometrics was reported by Biel [11] in 2001. Since then, ECG biometrics have attracted increasing research interest and are regarded as one of the most promising biometric techniques. In general, the ECG signal is a quasiperiodic signal with a frequency of 1-1.5 heartbeats per second [12]. Exploiting the repetitive property can increase biometric system accuracy. As shown in Fig. 1, a healthy ECG heartbeat has six fiducial points, namely, P, Q, R, S, T, and U.

Many challenges are encountered when processing ECG biometrics. Since sensors need to be attached to the body to acquire the ECG regardless of whether the configuration is on-the-person settings (medical acquisitions) or off-the-person settings, the interface between the body and electrodes may create contact noise, which affects the quality of the ECG signal to some degree. In addition, ECG signals are easily affected by physiological and psychological changes. For example, it could change along with activities, diets, diseases, positions of the electrodes, and other factors [7]. With these considerations, it is urgent to propose proper feature extraction methods or other techniques to overcome these challenges. Thus, in this paper, an effective feature extraction framework for ECG biometrics is introduced.

We design a novel feature extraction framework to generate a discriminative ECG biometrics feature representation, which can help improve the ECG biometrics accuracy. First, the ECG signal is represented by a new type of base feature. Inspired by one-dimensional Multi-Resolution Local Binary Patterns (1DMRLBP for short) [12], which was proposed by Louis et al., we design an improved multi-scale differential feature fused of one dimensional multiresolution local binary patterns (MSDF-1DMRLBP for short) by fusing the Multi-Scale Differential Feature (MSDF) with 1DMRLBP to act as the intermediate base feature of the ECG signal. The proposed base feature takes the differential changes with temporal variations into consideration and improves the capability of capturing micro texture and multi-scale differential signal characteristics. Then, we extract discriminative features from the output of the former step and generate the final enhanced features. Specifically, we adopt the collective matrix factorization (CMF) technique, which simultaneously embeds MSDF-1DMRLBP and label information, to seek a latent implicit feature space. The merits of this step are twofold: 1) The final features could preserve the intra-subject and inter-subject similarities. 2) The final features are robust and applicable to the noisy 1D ECG signals.

The contributions of this paper are summarized as follows:

•
First, we propose a novel multi-scale differential feature extraction method to efficiently capture micro texture and multi-scale differential signal characteristics.
•
Second, we generate the final feature by leveraging the CMF technique, which simultaneously embeds MSDF-1DMRLBP and label information. Due to CMF and semantic embedding, we can make full use of the supervised information and maximally preserve the intra-subject and inter-subject similarities.
•
Extensive experiments are conducted on four ECG datasets. The results demonstrate that the proposed method outperforms several state-of-the-art methods.

The remainder of this paper is organized as follows. Section 2 reviews the literature on ECG biometrics. Section 3 describes the proposed method in detail. Section 4 reports the experimental results and provides a comprehensive analysis. Finally, Section 5 presents the conclusions and future work.

Section snippets

Studies on ECG biometrics

Based on the ECG fiducial points in Fig. 1, research on ECG biometrics can be divided into two main categories, i.e., fiducial-based and non-fiducial-based approaches.

In fiducial-point-based approaches, features are extracted from fiducial points and distances between points, angles, areas, amplitudes, and so forth. There are some related examples, i.e., Juan et al. [13] extracted the time intervals and difference values in amplitude among several fiducial points, such as PR, QR, SR and TR. L.

Proposed method

To perform ECG biometrics, we designed a novel ECG biometrics framework. First (mentioned in Section 3.2), an improved base feature extraction algorithm is introduced to generate the intermediate base feature, i.e., MSDF-1DMRLBP. Second (mentioned in Section 3.3), CMF is leveraged to generate the final discriminative feature for ECG signals. Finally, the matching procedure is performed for the testing data. Before presenting the framework, we first preprocess and segment the ECG signals into

Database

To comprehensively validate the effectiveness of the proposed ECG-based feature extraction method, we conducted extensive experiments on four widely used benchmark datasets [42], i.e., MIT-BIH Arrhythmia (MITDB)[43], ECG-ID [44], PTB Diagnostic ECG Database [45] and the University of Toronto ECG Database (UofTDB) [46]. The summary of the four adopted datasets is presented in Table 1.

ECG-ID is a database entirely focused on biometrics. Twenty-second ECG recordings are collected from 90 subjects

Conclusion andfuture work

In this paper, we propose a novel ECG biometrics framework, including Multi-Scale Differential Feature fused with one-dimensional Multi-Resolution Local Binary Patterns (MSDF-1DMRLBP) and final feature learning with collective matrix factorization (CMF). First, the MSDF-1DMRLBP is extracted, which can express the intrinsic characteristics of the ECG, such as the local wave shape and the growth direction of the ECG signal, preserving the ECG heartbeat morphology about micro texture and

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported in part by the National Natural Science Foundation of China under Grants U1903127, 61703235 and 61876098 and in part by the Key Technology Research and Development Program of Shandong Province under Grant 2018GGX101032, 2019GGX101056.

Kuikui Wang received her master degree in computer science from the Shandong University, in 2014. Currently, she is pursuing her Ph.D. degree at Shandong University. Her main research interests are biometrics and machine learning.

References (50)

Y.N. Singh
Human recognition using fisher discriminant analysis of heartbeat interval features and ecg morphology
Neurocomputing
(2015)
G. Wübbeler et al.
Verification of humans using the electrocardiogram
Pattern Recognit. Lett.
(2007)
S.-C. Fang et al.
Human identification by quantifying similarity and dissimilarity in electrocardiogram phase space
Pattern Recognit.
(2009)
M. Hejazi et al.
Ecg biometric authentication based on non-fiducial approach using kernel methods
Digit. Signal Process.
(2016)
Y. Zhang et al.
Ecg-based personal recognition using a convolutional neural network
Pattern Recognit. Lett.
(2019)
Y.N. Singh et al.
Fusion of electrocardiogram with unobtrusive biometrics: an efficient individual authentication system
Pattern Recognit. Lett.
(2012)
M.D. Bugdol et al.
Multimodal biometric system combining ECG and sound signals
Pattern Recognit. Lett.
(2014)
C.-X. Li et al.
Scratch: a scalable discrete matrix factorization hashing for cross-modal retrieval
Proceedings of the International Conference on Multimedia
(2018)
B. Toth
Biometric liveness detection
Inf. Secur. Bull.
(2005)
F. Schroff et al.
FaceNet: a unified embedding for face recognition and clustering
Proceedings of the Conference on Computer Vision and Pattern Recognition
(2015)

K. Cao et al.

Automated latent fingerprint recognition

IEEE Trans. Pattern Anal. Mach. Intell.

(2018)

A. Gangwar et al.

Deepirisnet: deep iris representation with applications in iris recognition and cross-sensor iris recognition

Proceedings of the International Conference on Image Processing

(2016)

L. Yang et al.

Finger vein recognition with anatomy structure analysis

IEEE Trans. Circuits Syst. Video Technol.

(2017)

J. Han et al.

Individual recognition using gait energy image

IEEE Trans. Pattern Anal. Mach. Intell.

(2005)

I. Odinaka et al.

Ecg biometric recognition: a comparative analysis

IEEE Trans. Inf. For. Secur.

(2012)

K. PG et al.

Electromyography analysis for person identification

Int. J. Biomet. Bioinf.

(2011)

S. Barra et al.

EEG/ECG signal fusion aimed at biometric recognition

Proceedings of the International Conference on Image Analysis and Processing

(2015)

G.E. Forsen et al.

Personal Attributes Authentication Techniques.

Technical Report

(1977)

L. Biel et al.

Ecg analysis: a new approach in human identification

IEEE Trans. Instrum. Meas.

(2001)

W. Louis et al.

Continuous authentication using one-dimensional multi-resolution local binary patterns (1DMRLBP) in ECG biometrics

IEEE Trans. Inf. For. Secur.

(2016)

J.S. Arteaga-Falconi et al.

ECG authentication for mobile devices

IEEE Trans. Instrum. Meas.

(2015)

M. Li et al.

Robust ECG biometrics by fusing temporal and cepstral information

Proceedings of the International Conference on Pattern Recognition

(2010)

I. Odinaka et al.

Ecg biometrics: a robust short-time frequency analysis

Proceedings of the International Workshop on Information Forensics and Security

(2010)

M.N. Dar et al.

ECG biometric identification for general population using multiresolution analysis of DWT based features

Proceedings of the International Conference on Information Security and Cyber Forensics

(2015)

S.I. Safie et al.

Electrocardiogram (ECG) biometric authentication using pulse active ratio (par)

IEEE Trans. Inf. For. Secur.

(2011)

Cited by (32)

Multi-stage stacked temporal convolution neural networks (MS-S-TCNs) for biosignal segmentation and anomaly localization
2023, Pattern Recognition
In the computer vision domain, temporal convolution networks (TCN) have gained traction due to their lightweight, robust architectures for sequence-to-sequence prediction tasks. With that insight, in this study, we propose a novel deep learning architecture for biosignal segmentation and anomaly localization based on TCNs, named the multi-stage stacked TCN, which employs multiple TCN modules with varying dilation factors. More precisely, for each stage, our architecture uses TCN modules with multiple dilation factors, and we use convolution-based fusion to combine predictions returned from each stage. Furthermore, aiming smoothed predictions, we introduce a novel loss function based on the first-order derivative. To demonstrate the robustness of our architecture, we evaluate our model on five different tasks related to three 1D biosignal modalities (heart sounds, lung sounds and electrocardiogram). Our proposed framework achieves state-of-the-art performance for all tasks, significantly outperforming the respective state-of-the-art models having F1 score gains up to $\approx 9$ %. Furthermore, the framework demonstrates competitive performance gains compared to traditional multi-stage TCN models with similar configurations yielding F1 score gains up to $\approx 5$ %. Our model is also interpretable. Using neural conductance, we demonstrate the effectiveness of having TCNs with varying dilation factors. Our visualizations show that the model benefits from feature maps captured at multiple dilation factors, and the information is effectively propagated through the network such that the final stage produces the most accurate result.
BAED: A secured biometric authentication system using ECG signal based on deep learning techniques
2022, Biocybernetics and Biomedical Engineering
Citation Excerpt :
The method extracts fiducial features of each heartbeat to compute statistical features and decisions made with sequential procedures. K.Wang et al. [31] extracted QRS Complex, onset and offset of P, Q, S and T waves for Lead-I ECG and performed biometric recognition. Methods based on non-fiducial features do not recommend characteristic points for extracting feature sets.
Biometric authentication technology has become increasingly common in our daily lives as information protection and control regulation requirements have grown worldwide. A biometric system must be simple, flexible, efficient, and secure from unauthorized access. The most suitable and flexible biometric traits are the face, fingerprint, palm print, voice, electrocardiogram (ECG), and iris. ECGs are difficult to falsify among these biometric traits and are less attack-prone. However, designing biometric systems based on ECG is very challenging. The major limitations of the existing techniques are that they require a large amount of training data and that they are trained and tested on an on-person database. To cope with these issues, this work proposes a novel biometric authentication scheme based on ECG detection called BAED. The system was developed based on deep learning algorithms, including a convolutional neural network (CNN) and a long-term memory (LSTM) network with a customized activation function. The authors evaluated the proposed model with on-and off-person databases including ECG-ID, Physikalisch-Technische Bundesanstalt (PTB), Check Your Bio-signals Here Initiative (CYBHi), and the University of Toronto Database (UofTDB). In addition to the standard performance parameters, certain key supportive identification parameters such as FMR, FNMR, FAR, and FRR were computed and compared to increase the model’s credibility.The proposed BAED system outperforms prior state-of-the-art approaches.
Robust multi-feature collective non-negative matrix factorization for ECG biometrics
2022, Pattern Recognition
Citation Excerpt :
Hence a challenging problem for LBP-based recognition methods is to design a robust method to extract the latent semantics. The widely used collective matrix factorization (CMF) can extract the latent semantics and remove redundant information [15–17]. Most CMF-based methods assume that all input data matrices share the same factor and should have the same semantics.
The field of electrocardiogram (ECG) biometrics has received considerable attention in recent years. Although some promising methods have been proposed, it is challenging to design a robust and precise method to improve the recognition performance of ECG signals with noise and sample variation. While the advantage of improved local binary pattern (LBP) for establishing identities has been widely recognized, extracting the latent semantics from multiple LBP features has attracted little attention. We propose a robust multi-feature collective non-negative matrix factorization (RMCNMF) model to handle noise and sample variation in ECG Biometrics. We extract multiple LBP histograms as feature descriptors from segmented ECG signals, and propose a multi-feature learning framework that learns unified representations in the shared latent semantic space via collective non-negative matrix factorization. To further enhance the discrimination of learned representations, we integrate label information and multiple norms in the proposed model, which not only preserves intra- and inter-subject similarities but also mitigates the influence of noise and sample variation. RMCNMF can be solved by an efficient iteration method, for which we provide a convergence analysis in detail. Extensive experiments on four ECG databases show that it performs competitively with state-of-the-art methods.
Multi-view discriminant analysis with sample diversity for ECG biometric recognition
2021, Pattern Recognition Letters
Currently, electrocardiogram (ECG) biometric recognition is a novel research trend, and many methods have been developed. Due to the influence of physical and psychological activities, there are heartbeats diversities of the same person. However, the existing ECG biometric recognition methods do not make use of sample diversity information. In this paper, we present a multi-view discriminant analysis approach in the consideration of sample diversity for ECG biometric recognition. Firstly, we propose a method of generating multiple views by using single lead ECG signal. Secondly, we present a multi-views learning framework, which takes sample diversity into account to generate a more discriminative subspace. Thirdly, to obtain a more robust solution, we introduce a denoising constraint to learn the relationships between different views, which can create a stable representation against ECG noise. At last, experimental results demonstrate that compared with the state-of-the-art methods on four databases, the proposed method can achieve competitive performance compared to state-of-the-art ECG biometric recognition methods.
ECG-based biometric under different psychological stress states
2021, Computer Methods and Programs in Biomedicine
Citation Excerpt :
Different people show different ECG features, but it is difficult to design representative manual features. Kuikui Wang et al. proposed a Multi-Scale Differential Feature for ECG biometric identification with Collective Matrix Factorization (CMF) in 2020 [4]. By extracting multi-scale differential features from the one-dimensional ECG signal, and at the same time extracting discriminative information from the intermediate basic features, a robust ECG feature is finally generated.
In recent years, people have been exploring methods for biometric identification through electrocardiogram (ECG) signals. Under the same psychological pressure state, biometric identification through ECG signals is a traditional verification method. However, ECG signals are affected by changes in psychological stress, and ECG-Based biometric under different psychological stress states are still challenging. In this paper, we propose a method combining manual and automatic features for ECG-based biometric under different psychological stress states. And propose a new indicator Stress Classification Coefficient (SCC) that assesses the effect of different psychological stress on heart rate variability (HRV) features.
In our method, we obtain manual features to be a three-step process: first, HRV features obtained from the ECG signals. Second, based on HRV features, the mental state of the experimental subjects is assessed by using the Gaussian mixture model (GMM). Finally, use cluster centers to process the original HRV features to reduce the Stress Classification Coefficient (SCC). Also, the one-dimensional convolutional neural network is constructed to automatically extract the implied features of ECG signals. Finally, the manual feature and the automatic feature are combined, and the final recognition result is obtained through the support vector machine (SVM) model. The major attribute of the proposed method is that it can perform ECG biometric under different psychological stress states. The combination of manual and automatic features expands the application scenarios of ECG-based biometric.
Based on this method, we used the Montreal stress model with calculation experiment in the laboratory to induce stress on 23 healthy students (10 women and 13 men, aged 20–37), and obtain their ECG signals under different stress conditions. Through this method to recognize the above data, an average recognition rate of more than 95% can be achieved, the average F1 score is 0.97.
The proposed method in this article is a promising approach to deal with the effects of different psychological stresses on ECG-Based biometric. It provides the possibility of ECG-Based biometric under different psychological stress.
Total Variation PCA-Based Descriptors for Electrocardiography Identity Recognition
2024, IEEE Access

View all citing articles on Scopus

Gongping Yang received the Ph.D. degree in computer software and theory from Shandong University, China, in 2007. He is a Professor with the School of Software, Shandong University and an adjunct professor in the school of Computer, Heze University. His research interests are pattern recognition, image processing, biometrics, and so forth.

Yuwen Huang received the master degree in computer science from Guangxi Normal University, in 2009. He is an associate professor in the school of computer, Heze university and pursuing his Ph.D. degree at Shandong University. His research interests include ECG recognition, biometrics and machine learning.

Yilong Yin received the Ph.D. degree from Jilin University, Changchun, China, in 2000. From 2000 to 2002, he was a Post-Doctoral Fellow with the Department of Electronic Science and Engineering, Nanjing University, Nanjing, China. He is currently the Director of the Machine Learning and Applications Group and a Professor with Shandong University, Jinan, China. His research interests include machine learning, data mining, and biometric.

View full text

Multi-scale differential feature for ECG biometrics with collective matrix factorization

Highlights

Abstract

Introduction

Section snippets

Studies on ECG biometrics

Proposed method

Database

Conclusion andfuture work

Declaration of Competing Interest

Acknowledgements

Neurocomputing

Pattern Recognit. Lett.

Pattern Recognit.

Digit. Signal Process.

Pattern Recognit. Lett.

Pattern Recognit. Lett.

Pattern Recognit. Lett.

Biometric liveness detection

Inf. Secur. Bull.

FaceNet: a unified embedding for face recognition and clustering

Proceedings of the Conference on Computer Vision and Pattern Recognition

Automated latent fingerprint recognition

IEEE Trans. Pattern Anal. Mach. Intell.

Deepirisnet: deep iris representation with applications in iris recognition and cross-sensor iris recognition

Proceedings of the International Conference on Image Processing

Finger vein recognition with anatomy structure analysis

IEEE Trans. Circuits Syst. Video Technol.

Individual recognition using gait energy image

IEEE Trans. Pattern Anal. Mach. Intell.

Ecg biometric recognition: a comparative analysis

IEEE Trans. Inf. For. Secur.

Electromyography analysis for person identification

Int. J. Biomet. Bioinf.

EEG/ECG signal fusion aimed at biometric recognition

Proceedings of the International Conference on Image Analysis and Processing

Personal Attributes Authentication Techniques.

Technical Report

Ecg analysis: a new approach in human identification

IEEE Trans. Instrum. Meas.

Continuous authentication using one-dimensional multi-resolution local binary patterns (1DMRLBP) in ECG biometrics

IEEE Trans. Inf. For. Secur.

ECG authentication for mobile devices

IEEE Trans. Instrum. Meas.

Robust ECG biometrics by fusing temporal and cepstral information

Proceedings of the International Conference on Pattern Recognition

Ecg biometrics: a robust short-time frequency analysis

Proceedings of the International Workshop on Information Forensics and Security

ECG biometric identification for general population using multiresolution analysis of DWT based features

Proceedings of the International Conference on Information Security and Cyber Forensics

Electrocardiogram (ECG) biometric authentication using pulse active ratio (par)

IEEE Trans. Inf. For. Secur.