基于运动观察EEG的运动方向解析

Analysis of motion direction of EEG based on motion observation

与运动想象相比,运动观察的脑电信号没有被试主动思维参与,解析难点在于其信号幅值更弱且难以获取。实验针对运动观察中的方向判别进行了研究,以获取运动观察过程脑电特征明显频段作为切入点,利用运动观察眼动追踪信号确定有效观察任务,绘制脑地形图序列,定位激活的脑区,选出关联通道;然后结合运动观察EEG在特定频段时域上的能量特征较明显的特性,改进CSP算法,基于信号能量特征,利用SVM进行分类识别。实验得到运动方向解析平均80%以上分类准确率,最高在0～4 Hz频段上,达到了86. 28%,实现了运动观察虚拟小车左右转的解析与识别,为复杂运动观察任务EEG的解析与识别提供了有效的方法。

Compared with motion imagination,EEG signal of motion observation was not involved in active thinking,and the difficulty of analysis was that the signal amplitude was weaker and more difficult to obtain.This paper aimed at analyzing the direction in motion observation,took the obvious frequency band of the EEG characteristic of the motion observation process as the entry point,first determined the effective observation task by the motion observation eye tracking signal,then drew the brain topographic map sequence,located the activated brain area,selected the associated channel,and next improved CSP algorithm according to obvious energy characteristics of EEG in the specific frequency band,finally used the SVM to classify and identify based on the signal energy characteristics.The experiment shows that the accuracy of motion resolution about direction is more than 80%,and the highest accuracy is 86.28%in the range of 0~4 Hz,which realizes the analysis and identification of the left and right turn of the virtual car,and provides an effective method for the analysis and identification of complex task of motion observation EEG.