| Considering most of the patients with severe motor dysfunction still have normal brain function,electroencephalograph(EEG)signals are regarded as a signal source for prosthetic limb control,which can perceive the brain’s movement intention.The electrooculogram(EOG)signal is one of the methods that most widely used currently to measure eye movements,which initiate self-selection goals by controlling eye’s movements,so the EOG signals as human’s intention to control external devices are feasible.Therefore,this thesis designs a system of two kinds of biological signals,which is based on the combination of the EEG with EOG.The ultimate purpose is to improve the recognition of the brain computer interaction equipment and increase the freedom of control the external device.The main work includes two aspects: one is to analyze denoising preprocessing,feature extraction and pattern recognition on brain electrical signals,the other is feature extraction,analysis and research on the EOG signals.When the brain is imagining body movements,EEG appear event related synchronization and event related desynchronization characteristics,for the two kinds of EEG signals,the signal acquisition for motor imagery has been designed and implemented.The preprocessing part adopts the method of band-pass filter and wavelet threshold de-noising,the filtering effect is analyzed by comparing the signal before and after filtering.Based on the characteristics of EEG signals,time-frequency characteristics of FC5 and FC6 lead are analyzed.Finally,Hilbert method is used to extract the feature of the brain signals,and the energy spectrum characteristic values of two kinds of imaginary motions are extracted.In the aspect of pattern recognition,three kinds of common classifier algorithms have been studied,the classification accuracy and the maximum from the area under the receiver operating characteristic(ROC)curve as the evaluation criteria,by comparing the classification performance of the three kinds of pattern recognition,support vector machine is selected as the classification and recognition method of EEG signals.EOG signal is mainly based on the typical waveform characteristics of the signal to identify the different modes of winking.In order to distinguish the natural blink actions and intentional wink movements,the duration and amplitude of the eyelid closure can be used as the evaluation criteria.The recognition of the eye signal is done by designing a threshold detection algorithm which is a kind of mathematical morphology.After processing,peak detection and threshold detection,the wink action of information in the EOG signal can be effectively identified,which can solve the key problem in the EOG identification.At last,a system scheme based on EEG and eye signal is designed.In this thesis,under the assistance of EOG,the control command of EEG signal to confirm and cancel have been studied.From the results of identification,the proposed system is feasible to recognize the human upper limb motion intention with the combination of EEG and EOG. |
