The Development Of Technologies For Biometrics Based On Multi-biological Signal

With the rapid development of information technology,the issue of information securityhas become increasingly prominent.Biometrics technology with higher safety and convenience has become a development trend.However,in the application of realistic scenarios,the accuracy of biometrics has been affected by several problems with a single biometric system,such as the noise of the collected data,the similarity between features,non-university,and deceptive attacks.The research of multi-mode biometrics technology provided an effective solution to solve many shortcomings of single feature recognition system.In this paper,electrocardiograph(ECG)and photoplethysmography(PPG)are selected for fusion recognition research.The work done is as follows:(1)In view of the current lack of a database of ECG and PPG signals of the same subject,it is difficult to carry out related work.This experiment used a signal acquisition device designed and developed by the laboratory to collect ECG signals and PPG of multiple subjects.Signals,a small dual-mode database needed for subsequent fusion recognition is established.Corresponding pre-processing methods are adopted for the interference of the two signals,laying a foundation for subsequent signal processing.(2)The P-T algorithm and wavelet transform method are used to extract ECG signals.Two methods are used to extract the time-domain and amplitude characteristics of the ECG signal,and the signal is segmented periodically.The collected features are supported by the application.The vector machine classifier performs multi-class recognition and obtains the performance of the two methods.From the recognition rate and the equal error rate,the recognition rate of the P-T algorithm is 96.1%,the equal error rate is 9.68% at the threshold of 0.7368,and the recognition rate of the wavelet transform method is 95.3%,the equal error rate was 12.15% at the threshold of 0.7249.(3)Aiming at the waveform characteristics of the PPG signal,a differential threshold method for feature extraction of the PPG signal is proposed.From the first derivative curve and the second derivative curve of the PPG signal,four time-domain features and seven amplitude features of the signal are obtained.The support vector machine classifier is used to classify and identify the PPG signals.The performance of the single-mode PPG recognition system is experimentally analyzed,and the recognition rate is 91.5%.The equal error rate is 24.3% at the threshold 0.8017.(4)Aiming at the shortcomings of the single-mode ECG signal and PPG signal recognition technology,an improved discriminant correlation analysis method is proposed based on the typical correlation analysis method for feature layer fusion.In this case,a multi-set discriminant correlation analysis method is proposed.The experiments show that the recognition rate of fusion recognition can reach 98.2%.Compared with single-mode recognition,the recognition rate is significantly improved in short-term training.The equal error rate at the threshold of 0.6188 is 8.642%,which proves the effectiveness of the dual-mode fusion recognition system.