Research And System Design Of Identification Algorithm Based On Multimodal Physiological Signal Fusion

With the vigorous development of the information technology industry and the Internet industry,the daily convenient living and working environment cannot be separated from the personal identification system.At present,traditional identification methods such as certificates and passwords are gradually eliminated because they can not meet people’s daily needs.Biological signal identification technology based on human physiological characteristics has gradually become a research hotspot.Due to the differences between different individuals,and the unique uniqueness,stability,and difficulty in forgery of physiological signals,it has unique advantages to use them for human identity recognition.In this paper,we design and verify a complete set of identity recognition system which can be networked windows,taking the PPG(Photoplethysmography)signal and ECG(electrocardiogram)signal as the objects.Firstly,the PPG signal and ECG signal are cleaned,including signal filtering,denoising and so on;Then,the signal period is segmented into single period signal,and the signal characteristics of PPG signal are extracted from the aspects of time domain,frequency domain and blood flow information parameters;For ECG signal,signal features are extracted from QRS main wave in time domain and frequency domain to form PPG signal and ECG signal feature vectors respectively.Then,in order to enhance the stability of the identity recognition system,a method based on the feature layer fusion of PPG signal and ECG signal is proposed in this paper.Using the principle of minimum correlation between the two modal signals,the unique features of each mode are fully expressed,and the features of each mode complement each other,avoiding the disadvantage of low recognition accuracy of a single modal feature identity recognition system.Because the fusion process only considers the correlation between the two modal signals and ignores the relationship between signal features and individual categories,this paper proposes a neighborhood rough set model based on Cauchy perturbed particle swarm optimization to further reduce the dimension of the fused features and obtain the final fused feature vector.In order to obtain a higher recognition accuracy,this paper uses the artificial bee colony algorithm to optimize the support vector machine,constructs the identity recognition model,and inputs the fused reduced dimension signal features into the model for testing to verify the effectiveness of the proposed algorithm.After many experiments,the identification method based on the fusion of PPG signal and ECG signal proposed in this paper has good performance and can achieve a high correct recognition rate.Finally,this paper realizes the networking of identity recognition system based on the fusion of two modal physiological signals.The Wi-Fi module is used to complete the wireless transmission of data,and the algorithm implementation process is completed on the cloud platform or host computer.A complete set of networked identity recognition system is built through the graphical display interface or other display of recognition prediction results.