Research Of Inheritance Method For Authentication Based On Electrocardiogram

With the development of science and technology,people's demand for personal information security is gradually climbing.Authentication is applied to the high security scene,such as payment system,immigration clearance authentication system,etc.In recent years,because it has these defects,like gotten,forged,lost and so on,to authenticate by adopting the methods of traditional ID card and password,biometric identification technology is used in authentication frequently for their abilities to overcome these defects.Electrocardiogram(ECG)is a “live” signal and is more difficult to be copied or forged than fingerprints,face and other biological characteristics,which is safer.Therefore,more attention has been paid to ECG's ability in authentication.In order to improve the true positive rate(TPR)of authentication continually under the condition of the zero false positive rate(FPR),the inheritance learning mechanism is introduced for authentication based on ECG.In the paper,the authentication strategy based on ECG tunnel morph is proposed,and the problems such as the formation of ECG tunnel morph,boundary extraction,and authentication method are researched.The strategy inherits the ECG waveform of individual in the form of a tunnel morph,and with the increasing amount of ECG data,the individual characteristics are to be more obvious,which is helpful to improve the true positive rate of authentication.A number of experiments are conducted in MIT-BIH ECG data set and the hand ECG data collected in real scene.Experimental results show that the average true positive rate of authentication is significantly improved with the inheritance learning mechanism is introduced for authentication.In order to reduce the influence of the noise waveform(or singular waveform)for tunnel shape,a template selection method is proposed based on contribution rate of each ECG waveform.According to the contribution rate of ECG waveform,the second-order differential threshold value is used to determine the number of templates and representative ECG templates selection for inheritance learning of the ECG tunnel morph.A number of experiments are conducted in MIT-BIH ECG data set and the hand ECG data collected in real scene.The experiments are conducted in the hand ECG data collected in real scene the average true positive rates of authentication are improved 41.93% and 26.29% respectively with the inheritance learning of ECG tunnel morph.Experimental results that it will help to make the outline of ECG tunnel morph more clear by using the method,and improve the true positive rate of authentication.Due to the differences of individual ECG,ECG data that is collected from the same individual in different states exists differences.In order to improve the true positive rate of authentication continually under the condition of zero false acceptance rate,a method of key feature pant based on the section strategy for authentication is proposed in this paper.The method matches the key feature pant of ECG tunnel morph strictly,and increases inclusive weight of non-critical feature pant.A number of experiments are conducted in MIT-BIH ECG data set and the chest lead ECG data collected in real scene.Comparing with not using section strategy,the experimental results show that the average true positive rates of authentication are increased 15.53% and 14.81% respectively.