Comparison of Two Person Identification Methods using Radar Extracted Heartbeat Signals

Non-contact biometric identification has gained significant attention in recent years due to its flexibility and ability to ensure privacy and confidentiality. Previous research has primarily focused on utilizing cardiac radar signals detected by radars. This paper aims to perform a comparison of two methods for person identification using heartbeat signals extracted from radar data. In the first method, we directly use the heartbeat time series data as the input of a deep learning model for person identification. The second method explores the feasibility of using the spectrogram generated from cardiac radar heartbeat signals combined with another deep learning model to achieve the same object. Furthermore, we examined the robustness of both methods by introducing noise into the signals and assessing their performance under these conditions. Experimental results show that the accuracy of person identification using spectrograms is 98.82%, which is higher than that of the method based on spectrograms. After introducing noise, the accuracy of both methods decreases. However, the method based on spectrograms experiences a more pronounced decline in accuracy than the method using time series data directly.