ECG-Based Human Identification System by Adaptive SVD Algorithm

With the growing demand for enhanced security in biometric technology, ECG-based human identification has attracted increasing attention. This study developed a portable ECG acquisition device and proposed an adaptive Singular Value Decomposition (SVD)-based algorithm for feature extraction from ECG signals in the context of identity recognition. Compared with the classical Singular Value Decomposition–Total Least Square (SVD–TLS) algorithm, the proposed adaptive SVD algorithm automatically selects the model and order of the parameterized power spectrum based on the correlogram. This addresses a key limitation of the SVD–TLS algorithm, which suffers from the challenge of selecting an appropriate threshold for the normalized ratio. Additionally, leveraging the frequency-domain characteristics of ECG signals, this method reduces the number of features, thereby enhancing the execution efficiency of the identification process. The structure of this paper is organized as follows: First, multiple cardiac cycles are extracted from the ECG signal. Second, features of each cardiac cycle are obtained using the adaptive SVD algorithm. Finally, an ECG signal analysis software was developed to facilitate the acquisition of various ECG analysis results. To validate the effectiveness of the proposed method, experiments were conducted on the MIT-BIH Arrhythmia database, achieving an identification accuracy of 99.59%. Additionally, ECG data from 40 healthy subjects were collected using the designed portable ECG acquisition device, yielding a recognition rate of 95.03% on the self-constructed database. Experimental results confirm that the proposed method outperforms state-of-the-art approaches in terms of identification performance.