Research On Multi-frequency Domain And Multi-feature Identity Recognition Method And System Based On PPG And ECG Signals

In recent years,with the prosperity of emerging technology industries and internet companies,people’s lives have become increasingly convenient and efficient.And this convenient lifestyle is closely linked to personal identification.Nowadays,traditional identity verification methods,such as ID cards and passwords,are gradually being replaced by physiological signal based identity recognition methods due to their inability to meet people’s needs anytime and anywhere.This method has become a hot topic for many researchers in recent years.Physiological signals exhibit differences among different individuals,while possessing unique,stable,and difficult to replicate characteristics,making them very suitable for identity recognition.This study selects PPG and ECG signals generated independently by the human body as biological features and verifies their effectiveness in networked environments.(1)Studied the preprocessing of PPG and ECG signal filtering,noise reduction,and other operations.Perform periodic cutting on each PPG and ECG signal.For two types of single cycle signals,the 20 dimensional features of PPG signals and the 13 dimensional features of ECG signals are extracted from the time domain,frequency domain,and wavelet domain,respectively.Next,these features are constructed into feature vectors for PPG signals and ECG signals,and normalized for each type of feature vector.(2)A feature level weighted average fusion algorithm based on PPG and ECG signals and an improved fast correlation filtering algorithm based on importance factor were studied.By fusing two feature vectors with different dimensions,the complementarity between PPG and ECG signals is fully utilized.The weighted average fusion method improves the reliability of features and enhances the stability of the identity recognition system;Pay attention to the correlation between signal features and individual categories,further reduce the dimensionality of the fused features,and obtain the reduced feature subset.Improve the fast correlation filtering algorithm to further improve performance.(3)We optimized the support vector machine using an improved grey wolf algorithm based on the base weight group wise strategy and constructed an identity recognition model.Input the fused feature subset after dimensionality reduction into the constructed model for training and testing to verify the feasibility of the algorithm proposed in this study;The experimental results show that the identity recognition method proposed in this study based on the multi frequency domain features of PPG and ECG signals exhibits good performance.Compared with traditional identity recognition methods,the algorithm proposed in this paper has significantly improved in speed and recognition accuracy.(4)A networked identity recognition system based on PPG and ECG multi frequency domain features has been studied and implemented.This system utilizes the WI-FI module to achieve wireless data transmission,implements algorithm processes through cloud platforms or upper computers,and displays recognition prediction results through graphical display interfaces or other means,constructing a complete networked identity recognition system.