Electroencephalography Synthesis of Hand Movement: Events Features Detection, Classification, for Robotics Applications

摘要

This research is related to Electroencephalography (EEG) analysis for hand movements, and events features detection. The adopted BCI, EEG-Technique, uses non-invasive way in reading the brain signal by sensors (the Electrodes) at the top of the brain. The work was related to samples of 15 subject with 14 Tasks using 64-Channels Electrode, which generates 64 signal from the brain. Given this fact, we have focused on the three most important channels for movements of the right and left hand, which are (C3, C_z, and C_4), representing the main focus of this research. An adopted procedure is: Initially, EEG and subjects experimentation analysis were performed, where we filtered the EEG signal and segmentation of important parts of EEG and noise removal. Secondly, the feature extraction, where we adopted statistical based algorithms to extract specific features which will be used in third and final step of the research The final stage was related to the classifying of the filtered and preprocessed EEG waves and patterns of subjects conducting the same experiment. In this respect, Linear Discremenant Analysis (LDA) classifier was adopted to classifier the EEG signal into three main events, the (Γ_0,Γ_1,Γ_2) corresponding to REST, LEFT, RIGHT hand respectively.