A Personal Identification Method Based on Surface Electromyography

Xudong Yang,Jingwen Ye,Pan Xu,Yueming Gao

DOI: https://doi.org/10.1109/apcap56600.2022.10069224

2022-01-01

Abstract:Biometric identification technology plays an important role in information security and access control systems. Traditional biometric features such as face, fingerprint, and iris have the risk of being cracked and forged, whereas in vivo biometric features have the advantage of being real-time and difficult to be forged. Therefore, a personal identification method based on surface electromyography (sEMG) signals was proposed and the feasibility together with the effectiveness of sEMG signals for personal identification was validated. Firstly, the sEMG signals of twenty volunteers were collected and the signal segmentation was implemented. Then, the classifier based on feature extraction and support vector machine (SVM), as well as continuous wavelet transform (CWT) and convolutional neural network (CNN) were performed for personal identification, which obtained the accuracy of 94.44% and 95.00%, respectively. The results showed that sEMG signals could effectively personalize identification, which may give insights to improve the security of existing biometric identification systems.