Feature Classification And Parameter Optimization Of Brain-computer Interface Based On Chinese Character Silent Reading

The number of Chinese users is the most in the world.Effectively extract EEG characteristics induced by silent reading and use it as control input of brain-computer interface.This way will enrich the paradigm of brain-computer interface and contribute to the rehabilitation of aphasia patients.By analyzing the off-line EEG experimental data,we explore the optimization of EEG classification feature extraction in Chinese character silent reading.This paper is helpful to the theoretical research of language brain-computer interface technology.It also improves the accuracy of language brain-computer interface based on Chinese character silent reading.And we provide new ideas for language rehabilitation training.The details are as follows:Based on the evoked EEG,a language-brain-computer interface experiment for silent reading of Chinese characters is designed in this paper.We collected EEG signals from 9 subjects through off-line experiments.Data features are selected and optimized from time domain,frequency domain and space domain for Chinese character recognition.Event Related Spectral Perturbation is used for time-frequency analysis to obtain the characteristics of each participant.The Common Spatial Pattern is used for spatial analysis to select the best lead group.And we combined the classification results to optimize the selection of electrodes.The results show that spectral energy was dynamic changed in alpha and beta bands while all characters were read silently.Averaged matching accuracy of Chinese characters were improved by used the modified time and frequency range.And the matching accuracy was increased by 3.37% than used the feature from unified time and frequency range.In this paper,pictures,audio and voiced mouth action videos are used as experimental stimulus materials.The EEG generated by the test participants in different Chinese characters is collected,pre-processing and time-frequency analysis are performed.And the EEG topographic map is generated by Common Spatial Pattern.The Fisher classifier is used to obtain three.The correct classification rate of different states under the stimulating materials is optimized from the experimental point of view to extract the EEG classification features when reading Chinese characters.The results showed that with the deepening of the sensory dimension,the brain cortex area of the subject was obviously activated,and the activation range was more concentrated,and the classification accuracy rate was also improved.The audio stimulus is stimulating relative to the picture,and the accuracy of the second classification is increased by an average of 2.5%.The video stimuli increased by 4.3% compared to the picture stimuli.Language-brain-computer interface experiment of off-line Chinese character silent reading was designed by us.This paper based on the language brain-computer interface experiment of offline Chinese character reading,optimizes the extraction of EEG signal classification feature parameters through time,frequency and space.The algorithm dimension is online brain-computer interface.The recognition rate is improved.Based on the principle of mirror neuron system,three kinds of stimulating materials such as pictures,audio and video are compared and analyzed.And the activation state of the brain is analyzed.The practical application of brain-computer interface is supported from the optimization of experimental design.