Algorithm Of Human Identification Based On ECG Signal

Over the past decade,research on the use of ECG signals in biometrics has increased.ECG characteristics are suitable for human identification: it is unique,existing in all living body,and difficult to forge.ECG features have a lot of potential,such as in daily access to access control,and in patient diagnosis and in wearable electronic applications.ECG signal has a unique advantage and a good application prospect in the identification technology.This paper studys the method of identification of ECG signals based on multi-fiducial,the denoising knowledge of ECG signals based on wavelet transform,and uses the method to denoise multi-lead ECG signals.Paper searches ECG feature points by wavelet transform,and extracts 25 amplitude and interval features of three-lead ECG signal as characteristics,then use KNN classifier for classification and identification.Paper also extracts 20 amplitude and interval features of two-lead ECG signal as well as 15 amplitude and interval features of one-lead ECG signal.This paper compares the three-lead,two-lead,single-lead ECG identification and authentication.This paper presents a new ECG identification method based on ECG vector ring reconstruction including single-fiducial and non-fiducial.The ECG vector loop is reconstructed in spaces based on single-fiducial by using three-lead ECG,and the EN distance of the vector loop is used to identify with the KNN classifier.The ECG vector loop is reconstructed in two-demension based on non-fiducial by two-lead ECG,the Jaccard similarity coefficient of the vector ring is used to identify with the KNN classifier.And the influence of parameter changing on recognition rate is discussed.Then,paper converts the standard 12-lead into an orthogonal ECG vector using the Dower conversion matrix and reconstruct vector ring in space.The Jaccard similarity coefficient of the vector ring is used to identify people.Paper chooses 2-3 leads from the multi-lead to rebuild the space vector ring through the Dower conversion matrix,then compares the results of combination.Single-lead ECG signal is used for phase space reconstruction.The Jaccard similarity coefficient is used to identify the space.The influence of parameter changing on recognition rate is discussed.The minimum Manhattan distance is used to authenticate.Paper applicates algorithm in multiple database and compares with authoritative algorithm.