Biofusion: Expanding Biometric Horizons with ECG and PCG Integration

In recent years, biometrics has emerged as a highly reliable and authentic method for person identification. We proposed a novel method in this paper for biometric identification based on the fusion of electrocardiogram (ECG) signals and heart sounds in the form of phonocardiogram (PCG) signals. A self-collected dataset from 32 individuals by using BIOPAC MP-36 systems was utilized. Raw signals were preprocessed using 3 types of IIR filters. Subsequently, cepstral features were extracted and fused serially, resulting in combined information from both ECG and PCG signals. Different classifiers were tested for performance comparison. Linear discriminant classifier as well as ensemble subspace discriminant classifiers outperformed the others depicting an accuracy of 100%. The results demonstrate that highly accurate biometric identification can be accomplished by the fusion of ECG and PCG signals. Moreover, the proposed system demonstrated promising accuracy levels for a wide range of signal-to-noise ratios under various noisy conditions.