Dry contact fingertip ECG-based authentication system using time, frequency domain features and support vector machine

Acquiring fingertip ECG (electrocardiogram) signal using dry contact electrodes is challenging due to the presence of noise and interference by EMG (electromyogram) potentials. In this paper, we propose a method for using the fingertip ECG signal for biometric authentication. The noisy segments of the signal are segmented out using a variance-based heuristic and the clean signal is used for subsequent processing. By applying baseline correction and band pass filtering, the filtered signal is used for beat feature extraction. The features are used to train a support vector machine (SVM) classifier. Experimental results are presented to show the optimum filter parameters and feature sets for best classification performance. The performance of the proposed method with the optimum parameters was evaluated on a public domain CYBHi dataset with 126 subjects and the beat level EER of 3.4% was obtained.