Motor Imagery Recognition Method based on Multi-Classifier Integrated Desision

Motor imagery brain-Computer Interface has a unique advantage in stroke rehabilitation. It encourages the patient to imagine the affected side of motor motion based on plasticity of brain nerve. Motor imagery plays an indispensable role on stroke rehabilitation for its benefit to the reconstruction of brain's damaged nerve. In order to further improve the classification accuracy and reliability, enhance patients' confidence in treatment, at the same time, it also provides more accurate feedback for the online training part of the follow-up rehabilitation system. SVM, KNN (k-nearest neighbor) and LDA (linear discrimination analysis) are used to classify EEG data. For each sampling data, three classification methods are used, each classification method produces a corresponding classification result, and the final classification result is determined by voting mechanism. The comprehensive classification effect of multi-classifier is 2 %~6% higher than that of SVM classifier only.