Research And System Design Of Identification Algorithm Based On PPG Signal

With the rapid development of economy and technology,information security has become increasingly important.Identity recognition technology is an important way to ensure information security.Traditional identification methods,such as fingerprint identification,identity documents,passwords and so on,have some defects,which can not meet the growing security needs of people.Biometrics is a more secure and reliable way of identification because it can distinguish the individual identity according to the inherent physiological signal characteristics of human body,and the physiological signal is not easy to be stolen or forged.Based on the above analysis,this paper proposes an identity recognition scheme which takes the PPG signal as the research object,and designs a set of networking identity recognition system.In this paper,PPG signal is preprocessed by filtering,feature point marking,single period division and so on.Because the traditional feature extraction method contains incomplete signal information,it may affect the accuracy of recognition.In order to avoid this problem,this paper uses the matched pursuit(MP)sparse decomposition algorithm to restore the PPG signal with a few optimal atoms,and takes the time-frequency characteristic parameters that can determine the atoms as the extracted features,Because the atom can restore PPG signal with high precision,its timefrequency characteristic parameters contain more complete signal information.In order to improve the performance of MP algorithm in processing PPG signal,this paper improves the algorithm in speed and precision according to the characteristics of PPG signal.After the feature extraction is completed,the main features are filtered by the relief algorithm to form the final feature vector.In order to get a higher recognition accuracy,this paper uses machine learning method to build a classification recognition model.Two algorithms,decision tree and support vector machine(SVM),are used to build the recognition model,and the two algorithms are improved according to the characteristics of PPG signal.In the construction of decision tree,firstly,by neglecting the fuzzy attribute,the misjudgment error is reduced and the complexity of decision tree is reduced.Then,a feature importance is added to calculate the information gain using ID3 algorithm,which avoids the misjudgment of information gain caused by the number of feature values,and improves the effectiveness of branch feature selection.In the construction of SVM classification model,due to the nonlinearity of PPG signal characteristics,this paper selects radial basis function for high altitude mapping,and uses crow search algorithm to optimize the kernel parameters and penalty factors.After experimental comparison,the performance index of SVM classification model optimized by Crow search algorithm is the best.Finally,this paper uses Wi Fi module or Internet to realize the remote transmission of data,and completes the realization of algorithm function on the upper computer or cloud computing platform.Finally,the test results are displayed in real time through the upper computer interface or mobile phone and other intelligent terminals to build a networked identity recognition system.