EEG Personal Identification Based On Brain Functional Network And Autoregressive Model And Its Application

With the progress of science and technology,intelligent devices are gradually spread all over every corner.How to correctly and effectively identify different individuals and protect important information security is a key and difficult problem that our contemporary society should pay attention to.EEG identification is a promising new biometric identification method for security requirements.Compared with other identification technologies,EEG identification has many significant advantages,such as more secure,more reliable,unimitable,and attached to in live detection.First,the brain interconnects with neurons in different areas of the brain and works together to exchange information.Brain functional networks mine the interactive information among different brain regions,but ignore the individual local brain response activities.Second,at present most of the EEG identification methods of the data set is based on a specific task,participants in a particular induced by external conditions,apply it to real life does not favor the EEG identification application and popularization,has nothing to do so based on task,do not rely on a specific task data set of EEG identification study is very necessary,in view of the above problem,this thesis mainly work arrangement is as follows:1.An EEG identification method based on brain functional network and autoregressive model(BFN-AR)is proposed.For the brain functional network,only the information interaction between any lead electrodes was considered,and the characteristics of a single electrode were ignored.On this basis,the AR model parameter characteristics were added,and the phase characteristics between any two electrodes and the frequency domain information of a single electrode were also considered.Compared with the single brain network attributes and autoregressor model features,BFN-AR can more effectively extract the unique characteristics of subjects with different tasks.Meanwhile,the classification performance of four frequency bands of EEG signals,namely ?,?,? and ?,which are closely related to identity recognition,was further investigated in this thesis.2.A task-independent EEG identification method based on low-rank sparse decomposition is proposed.A low-rank sparse decomposition algorithm was introduced to eliminate task awareness related EEG signals,and the brain network topological attributes and autoregressive model features of inherent background EEG signals were extracted for EEG identification of mixed data sets with different tasks.3.Design and implement an EEG identification system.The system mainly includes three modules: user registration,login and user management.Registration module is mainly to save the basic information of the user to the database;Login module is used to judge whether the login user is legal;The user management module is responsible for managing the user information of the system.Finally,the test results show that the system can effectively identify the visitors,which lays a foundation for the application of the EEG identification system in the actual scene.