基于CAE和改进式VGGNet的心电身份识别算法

doi:10.19678/j.issn.1000-3428.0068320

摘要/Abstract

摘要：

随着物联网技术和人工智能技术的不断发展, 生物识别技术面临着信息泄露的风险。心电图(ECG)信号因其活体识别的高防伪性在生物识别领域具有一定的优势。针对传统ECG识别算法不能适应多变的采集环境、识别稳定性不高以及基于深度神经网络的ECG识别算法模型参数量较大与难以实现快速响应等问题, 提出一种基于卷积自动编码器(CAE) 和改进式VGGNet的ECG身份识别算法。首先设计了结合小波阈值去噪和单心拍分割的预处理方法, 得到干净的单周期ECG信号作为模型输入。其次构建了基于CAE的信号模态特征提取与降维处理模块, 学习得到输入数据更小维度的潜在表示。最后基于VGGNet优化模型设计, 进一步深入学习特征表示, 得到个体识别的结果。实验结果表明, 该算法在MIT-BIH Arrhythmia Database、European ST-T Database和ECG-ID等数据库的189位测试者中实现了96%以上的识别精度, 其中European ST-T Database的识别精度高达99.82%, 可实现准确率较高、泛化能力较强的个体身份识别。

关键词: 心电图, ECG识别, 卷积自动编码器, 残差网络, 信号预处理

Abstract:

With the continuous development of Internet of Things (IoT) and Artificial Intelligence (AI) technologies, biometric identification techniques face the risk of information leakage. Electrocardiogram (ECG) signals have certain advantages in the field of biometric identification, owing to their highly anti-counterfeiting nature and in vivo recognition capabilities. In response to challenges such as traditional ECG recognition algorithms struggling to adapt to changing acquisition environments and low recognition stability and deep neural network-based ECG recognition algorithms with large model parameter sizes and difficulty in achieving fast response, an ECG identity recognition algorithm based on a Convolutional AutoEncoder (CAE) and an enhanced VGGNet is proposed in this study. First, a preprocessing method combining wavelet threshold denoising and single-heartbeat segmentation is designed to obtain clean single-cycle ECG signals as inputs. Then, a signal-mode feature extraction and dimensionality reduction module based on the CAE is constructed. Learning a lower-dimensional latent representation of the inputs. Finally, based on the optimized design of the VGGNet model, we studied the feature representation and obtained the results of individual recognition. The experimental results showed that the algorithm achieved a recognition accuracy of more than 96% for 189 testers in databases, including the MIT-BIH Arrhythmia Database, European ST-T Database, and ECG-ID. The recognition accuracy of the European ST-T Database is as high as 99.82%, which can realize individual identification with high accuracy and strong generalization ability.

Key words: Electrocardiogram(ECG), ECG recognition, Convolutional AutoEncoder(CAE), Residual Network(ResNet), signal preprocessing

严洁, 张烨菲, 张显飞. 基于CAE和改进式VGGNet的心电身份识别算法[J]. 计算机工程, 2025, 51(1): 295-303.

YAN Jie, ZHANG Yefei, ZHANG Xianfei. ECG Identification Algorithm Based on CAE and Improved VGGNet[J]. Computer Engineering, 2025, 51(1): 295-303.

https://www.ecice06.com/CN/Y2025/V51/I1/295

图/表 14

图1 ECG身份识别模型流程

Fig.1 The procedure of ECG identity recognition model

图2 心电信号分割

Fig.2 ECG signal segmentation

图3 CAE模型结构

Fig.3 CAE model structure

图4 本文提出的VGG-ResNet模型

Fig.4 The VGG-ResNet model of proposed in this paper

图5 心电信号去噪前后对比(采自100号受试者)

Fig.5 Comparison of ECG signal before and after denoising (from subject 100)

图6 CAE模型训练过程

Fig.6 CAE model training process

图7 ECG信号的重构

Fig.7 Reconstruction of ECG signal

图8 ECG身份识别模型的分类性能

Fig.8 Classification performance of ECG identification model

图9 本文提出的算法在不同数据集上的识别性能

Fig.9 The recognition performance of the proposed algorithm on different datasets

图10 混淆矩阵

Fig.10 Confusion matrix

参考文献 25

1	DICK E. Balancing user privacy and innovation in augmented and virtualreality[EB/OL]. [2023-08-10]. https://www.docin.com/p-2637818093.html.
2	BIEL L , PETTERSSON O , PHILIPSON L , et al. ECG analysis: a new approach in human identification. IEEE Transactions on Instrumentation and Measurement, 2001, 50 (3): 808- 812. doi: 10.1109/19.930458
3	汪莉. 基于ECG信号的身份识别技术研究[D]. 济南: 山东大学, 2005.
	WANG L. Research on identity recognition technology based on ECG signal[D]. Jinan: Shandong University, 2005. (in Chinese)
4	SILVA H P D, GAMBOA H, FRED A. Applicability of lead V2 ECG measurements in biometrics[C]//Proceedings of Mede-Tel'07. Washington D. C., USA: IEEE Press, 2007: 177-180.
5	GAHI Y, LAMRANI M, ZOGLAT A, et al. Biometric identification system based on electrocardiogram data[C]//Proceedings of Conference on New Technologies, Mobility and Security. Washington D. C., USA: IEEE Press, 2008: 1-5.
6	贺煜航, 刘棪, 陈刚. 基于自适应图卷积网络的心电图多标签分类模型. 计算机工程, 2022, 48 (12): 261- 269. URL
	HE Y H , LIU Y , CHEN G . Multi-label classification model of electrocardiogram based on adaptive graph convolutional network. Computer Engineering, 2022, 48 (12): 261- 269. URL
7	CHEN C Y , LIN Y T , LEE S J , et al. Automated ECG classification based on 1D deep learning network. Methods, 2022, 202, 127- 135. doi: 10.1016/j.ymeth.2021.04.021
8	MELZI P , TOLOSANA R , VERA-RODRIGUEZ R . ECG biometric recognition: review, system proposal, and benchmark evaluation. IEEE Access, 2023, 11, 15555- 15566. doi: 10.1109/ACCESS.2023.3244651
9	ALEIDAN A A , ABBAS Q , DAADAA Y , et al. Biometric-based human identification using ensemble-based technique and ECG signals. Applied Sciences, 2023, 13 (16): 9454. doi: 10.3390/app13169454
10	颜菲, 胡玉平. 叠加去噪自动编码器结合深度神经网络的心电图信号分类方法. 计算机应用与软件, 2019, 36 (4): 178- 185.
	YAN F , HU Y P . Electrocardiogram signals classification method based on stacked denoising AutoEncodercombined with deep neural network algorithm. Computer Applications and Software, 2019, 36 (4): 178- 185.
11	WANG D , SI Y J , YANG W Y , et al. A novel electrocardiogram biometric identification method based on temporal-frequency autoencoding. Electronics, 2019, 8 (6): 667. doi: 10.3390/electronics8060667
12	JUN S , SZMAJDA M , KHOMA V , et al. Comparison of methods for correcting outliers in ECG-based biometric identification. Metrology and Measurement Systems, 2020, 18 (5): 387- 398.
13	SUN L , ZHONG Z Y , QU Z G , et al. PerAE: an effective personalized AutoEncoder for ECG-based biometric in augmented reality system. IEEE Journal of Biomedical and Health Informatics, 2022, 26 (6): 2435- 2446. doi: 10.1109/JBHI.2022.3145999
14	SILVA R , FRED A , PLÁCIDO DA SILVA H . Morphological AutoEncoders for beat-by-beat atrial fibrillation detection using single-lead ECG. Sensors, 2023, 23 (5): 2854. doi: 10.3390/s23052854
15	卢莉蓉, 王鉴, 牛晓东, 等. 基于心动周期和经验模式分解的心电信号去噪处理. 数据采集与处理, 2020, 35 (4): 702- 710.
	LU L R , WANG J , NIU X D , et al. Electrocardiogram signal denoising based on cardiac cycle and empirical mode decomposition. Journal of Data Acquisition and Processing, 2020, 35 (4): 702- 710.
16	邵伟. 基于深度学习的ECG身份识别研究[D]. 长春: 吉林大学, 2021.
	SHAO W. Research on ECG identity recognition based on deep learning[D]. Changchun: Jilin University, 2021. (in Chinese)
17	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE Press, 2016: 770-778.
18	CHU Y F , SHEN H B , HUANG K J . ECG authentication method based on parallel multi-scale one-dimensional residual network with center and margin loss. IEEE Access, 2019, 7, 51598- 51607. doi: 10.1109/ACCESS.2019.2912519
19	BELO D , BENTO N , SILVA H , et al. ECG biometrics using deep learning and relative score threshold classification. Sensors, 2020, 20 (15): 4078. doi: 10.3390/s20154078
20	EL BOUJNOUNI I , ZILI H , TALI A , et al. A wavelet-based capsule neural network for ECG biometric identification. Biomedical Signal Processing and Control, 2022, 76, 103692. doi: 10.1016/j.bspc.2022.103692
21	MELTZER D , LUENGO D . Efficient Clustering-based electrocardiographic biometric identification. Expert Systems with Applications, 2023, 219, 119609. doi: 10.1016/j.eswa.2023.119609
22	KIM Y , CHOI G , CHOI C . One-dimensional shallow neural network using non-fiducial based segmented electrocardiogram for user identification system. IEEE Access, 2023, 11, 102483- 102491.
23	姚嘉伟, 蔡延光. 基于多核卷积和多头自注意力的心电图身份识别方法. 自动化与信息工程, 2023, 44 (5): 32- 37.
	YAO J W , CAI Y G . ECG identity recognition method based on multi-kernel convolution and multi-head self-attention. Automation[WT《Times New Roman》]& Information Engineering, 2023, 44 (5): 32- 37.
24	ZHANG X , LIU Q F , HE D , et al. Electrocardiogram-based biometric identification using mixed feature extraction and sparse representation. Sensors, 2023, 23 (22): 9179.
25	CAMARA C , PERIS-LOPEZ P , SAFKHANI M , et al. ECG identification based on the gramian angular field and tested with individuals in resting and activity states. Sensors, 2023, 23 (2): 937.

[1]	王晓路, 汶建荣. 基于运动-时间感知的人体动作识别方法[J]. 计算机工程, 2025, 51(1): 216-224.
[2]	王蕾, 党时鹏, 潘丰. 基于卷积神经网络的隐匿性旁路预测模型[J]. 计算机工程, 2024, 50(8): 40-49.
[3]	陈虹, 王瀚文, 金海波. 融合改进自编码器和残差网络的入侵检测模型[J]. 计算机工程, 2024, 50(2): 188-195.
[4]	张高伟, 杨湘. 基于特征融合的多通道心肌梗死定位模型[J]. 计算机工程, 2024, 50(11): 360-368.
[5]	王娅茹, 唐璐, 陈爱斌, 彭伟雄, 沈平. 基于LPDMR-NET的鸟鸣声识别[J]. 计算机工程, 2024, 50(10): 174-184.
[6]	丰芳宇, 罗晓曙, 蒙志明, 王广宇. 基于抗混叠残差注意力网络的人脸表情识别[J]. 计算机工程, 2023, 49(8): 190-198.
[7]	王名茂, 陈向阳, 叶子, 肖利芳, 郑戎. 全局区分性增强与边界监督的篡改检测网络[J]. 计算机工程, 2023, 49(6): 154-161.
[8]	王同官, 赖惠成, 蔡玉玺, 高古学, 汪烈军. 基于注意力残差网络的人脸超分辨率重建[J]. 计算机工程, 2023, 49(6): 234-241.
[9]	谌雨章, 黄逸姿, 张钧涵. 基于多速率空洞卷积的多尺度水下小目标检测[J]. 计算机工程, 2023, 49(6): 257-264.
[10]	丁子轩, 俞雷, 张娟, 李想, 王新宇. 基于深度残差自适应注意力网络的图像超分辨率重建[J]. 计算机工程, 2023, 49(5): 231-238.
[11]	冉瑞生, 翁稳稳, 王宁, 彭顺顺. 基于人脸关键特征提取的表情识别[J]. 计算机工程, 2023, 49(2): 254-262.
[12]	刘群坡, 盛月琴, 高如新, 卜旭辉. 基于关键帧和注意力残差网络的手语识别[J]. 计算机工程, 2023, 49(12): 224-230, 242.
[13]	赫晓慧, 宋定君, 李盼乐, 田智慧, 周广胜. 融合多尺度特征的遥感影像道路提取方法[J]. 计算机工程, 2022, 48(8): 196-205.
[14]	郝阿香, 贾郭军. 结合注意力与批特征擦除的行人重识别模型[J]. 计算机工程, 2022, 48(7): 270-276,306.
[15]	曹瑞阳, 郭佑民, 牛满宇. 基于最大最小距离的多中心数据综合增强方法[J]. 计算机工程, 2022, 48(6): 174-181.

选择文件类型/文献管理软件名称

选择包含的内容