Exploring Seismocardiogram Biometrics with Wavelet Transform

Seismocardiogram (SCG) has become easily accessible in the past decade owing to the advance of sensor technology. However, SCG biometric has not been widely explored. In this paper, we propose combining wavelet transform together with deep learning models, machine learning classifiers, or distance metric to perform SCG biometric matching tasks. We validate the proposed methods on the publicly available dataset from PhysioNet database. The dataset contains one hour long electrocardiogram, breathing, and SCG data of 20 subjects. We train the models on the first five minute SCG and conduct identification on the last five minute SCG. We evaluate the identification and authentication performance with recognition rate and equal error rate, respectively. Based on the results, we show that wavelet transformed SCG biometric can achieve state-of-the-art performance when combined with deep learning models, machine learning classifiers, or structural similarity metric.