A highly generalized ECG signal authentication method

Abstract

The invention discloses a highly generalized electrocardiographic signal identity authentication method, which relates to the field of biological signal identity authentication. Solve the problem that the generalization of existing methods is poor, and for new users, it needs to be retrained. The method adopts a parallel multi-scale one-dimensional residual network architecture, and uses three convolution kernels of different sizes for parallel feature extraction, which can extract features from ECG signals of different lengths, effectively retaining the features contained in ECG signals. Timing improves the effectiveness and accuracy of overall identification. The central objective function and the boundary objective function are used to train the deep neural network, which ensures the inter-class dispersion and intra-class aggregation of the extracted features, so that the similar features are gathered more closely, and the heterogeneous features are scattered more openly. It can improve the matching accuracy, reduce the dependence of the trained model on the training data, and greatly improve the generalization and robustness of the model.

Description

technical field [0001] The invention relates to the fields of pattern recognition, signal processing and identity authentication, in particular to a highly generalized electrocardiographic signal identity authentication method. Background technique [0002] With the rapid development of information technology, informatization and digitization have brought great convenience to human beings, but also magnified the possibility of personal or organizational data security being endangered. Many identity authentication methods have emerged for this reason. In recent years, with the maturity of various technologies and acquisition equipment, many identity authentication methods have begun to shift from traditional biometrics such as fingerprints and faces to more emerging, reliable and secure biometric feature recognition. The ECG signal is unique, alive, unique, and private, and at the same time has a certain degree of stability and collectability. It is a good biosignal identific...

AI Technical Summary

Problems solved by technology

Because the positioning of many reference points is relatively vague and difficult, and the position of the reference points in the ECG signals collected by different devices usually changes greatly, so these methods often have poor generalization. The performance of the data set is quite different, and the accuracy of identification will also be greatly affected

In addition, for unregistered users, these methods usually need to retrain the model, which cannot meet the real-time requirements of identity authentication

Images