Design And Implementation Of Brain Computer Interface Keyboard Based On SSVEP

Brain-Computer Interface(Brain-Computer Interface)is a non-muscle communication method,which can operate external devices by brain activity to communicate with the outside world.One of the most widely used applications of BCI research based on electroencephalography(EEG)is the BCI speller,which enables paralyzed people to express their ideas by paying attention to the target.Traditional P300 speller has a certain time delay.This thesis designs an EEG spelling system based on steady-state visual evoked potential(SSVEP),which improves the recognition rate and accuracy,and makes the paralyzed people adapt quickly in the actual application process.The thesis studies the principle of visual stimulus and the selection of visual stimulus types,and selects the parameters such as frequency and color of visual stimuli.The thesis uses Html5 and Python to design a visual stimulation interface based on SSVEP,which can induce EEG signals in the human brain.The front and back ends interactively control the visual stimulation interface to display the spelling results in real time,and output the characters to the upper part of the stimulation interface.The system can realize the addition,deletion and modification of characters.The thesis uses the portable EEG signal acquisition device to obtain EEG signals,filter and preprocess the EEG signals through independent component analysis,extract the preprocessed signals through typical correlation analysis to perform feature extraction,classify the signals,and then classify the signals.The obtained signal is output to the visual stimulation interface,and the keyboard spelling closed loop is completed.The thesis uses independent component analysis to remove EEG artifacts,and uses Fast Fourier Transform(FFT)and Canonical Correlation Analysis(CCA)to classify EEG signals.It's found that CCA is significantly better than FFT.In the online system experiment,the subjects achieved an accuracy rate of 89.14%.It shows that the research carried out in this thesis can process EEG signals well to achieve character spelling and achieve the expected functions.