最优区域共空间模式的运动想象脑电信号分类方法

共空间模式(Common Spatial Pattern,CSP)是脑机接口(Brain-Computer Interface,BCI)中一种有效的特征提取方法,然而传统CSP算法并未考虑在提取前剔除可能会影响其性能的不相关的嘈杂通道信号。所以针对不同对象的通道选择问题,提出了一种最优区域共空间模式(ORCSP)特征提取方法。首先通过欧式距离得到每个通道的附近区域,再根据方差比选择可分性最高的区域,然后采用5折交叉验证对区域内通道数目进行寻优,进而得到区分度最高的区域特征,最后使用支持向量机(SVM)进行分类。所提方法在BCI竞赛数据上进行了实验测试,并与同类型的正则化CSP和局部区域CSP算法进行了对比,在BCI Competition Ⅲ Dataset Ⅳ a数据集上达到了89.78%的平均准确率。实验结果验证了所提出方法的有效性。

Optimal Region Common Spatial Pattern for Motor Imagery EEG Classification

Common spatial pattern(CSP)is an effective feature extraction method for the brain-computer interface(BCI).However,the CSP algorithm does not consider eliminating the noisy and uncorrelated channel signals that can affect its performance before extraction.To address the problem,an optimal region common spatial pattern(ORCSP)feature extraction method is proposed for the channel selection of different subjects.Firstly,the local region of each channel s obtained by the Euclidean distance,and the region with the highest separability is selected according to the variance ratio.Then the 5-fold cross-validation is used to optimize the number of channels in the region.Finally,the region features with the highest discrimination are obtained,and the support vector machine(SVM)is used for classification.The proposed method was tested on the BCI Competition data,and compared with the regularized CSP and local area CSP algorithms.Our method has yielded the average accuracy of 89.78%on the BCI Competition Ⅲ Dataset Ⅳ a.The results verify the effectiveness of the proposed method.