Research And Realization Of Personal Identification Algorithm Using Electrocardiograms

With the development of our society, the recognition system of identification with high efficient and reliability is becoming a great demand. Biometrics refers to automatic personal identification on the basis of some physiological characteristics or some behavioral aspects, and it has major advantages over traditional identification technology. ECG signal has broad application prospects in person identification as a new biological characteristic which exists in live person only. The mechanism of ECG features is complex and not easily imitated. It has high safety factor, and it is more suitable for medical fields than any other methods. The development of portable electrocardiogram data collector can reduce the costs of data acquisition, and make the data acquisition more convenient. The ECG biometrics will be a powerful complement for current biometric identification systems.A novel method of ECG biometrics is proposed in this article based on the background and current ECG human identification research. The software of the ECG biometric has been realized on the Android platform. High recognition rate was obtained from normal and arrhythmia patients. Specific details are as follows:(1) Based on the R wave peaks exacted from the ECG signal, the full heartbeats are exacted to get the fusion feature of ECG. The fusion feature is combined with single average waveforms and wavelet features of the waveforms, and this means the fusion feature contains both the time domain and frequency domain features of ECG signal. The differences of different subjects’feature are strengthened. The result of the experiment on the arrhythmia patients has confirmed the effectiveness of the fusion feature. (2) The PLR-DTW method is used to discriminate individuals. The Dynamic Time Wrapping (DTW) is improved in this article. Adapt from human perception of waveforms to detect the key points, the piecewise linear representation (PLR) of templates are extracted. Meanwhile, the wrapping path of DTW method is limited to avoid useless search of all paths. The experimental results of MIT-BIH database prove the efficiency of the proposed PLR-DTW method while the identification accuracy rate remains high.(3) The arrhythmia patients are added to the experimental subjects. The experimental results show that the identification accuracy rate of the arrhythmia patients has reduced and the causes of the lower accuracy rate were analyzed. The causes could be the adverse effect of arrhythmia on the R peaks detection and the instability of the arrhythmia patients’ECG waveforms. We use one-lead ECG signal obtained from the in-hand physiological data collection terminal to do the experiment, the accuracy rate reached90%, and it shows that using single lead for ECG human identification is feasible.(4) Based on the simulation of ECG identification and the Android platform, the system software is realized. This lift up ECG identification from theory to application. SQLite embedded database is selected to create a template platform for ECG identification, and Java programming language is adopted to realize the system software. Finally the process of identification is demonstrated.