论文部分内容阅读
针对BCI技术中的脑电信号处理方法和事件相关去同步化的特点,提出了一种结合时、频、空域的特征提取方法。结合CSSD和AAR模型来提取脑电特征,并对基于AAR模型系数的特征提取方法进行了探讨,最终选择卡尔曼平滑算法提取模型系数,然后将提取的特征用简单的线性分类器进行分类。实验结果表明测试集的分类正确率达到了94.08%,而且这种特征提取方法有很好的时间分辨率,适合于在线分类。这是一种正确率高,时间分辨率高,适合在线分类的好方法。
In view of the characteristics of BCI processing and event-related desynchronization, a method of feature extraction combining time, frequency and spatial domain is proposed. Combining the CSSD and AAR models to extract the EEG features, the feature extraction method based on the AAR model coefficients is discussed. Finally, the Kalman smoothing algorithm is selected to extract the model coefficients, and then the extracted features are classified by a simple linear classifier. The experimental results show that the classification accuracy of the test set reaches 94.08%, and this feature extraction method has good time resolution and is suitable for online classification. This is a good way to sort online with the highest accuracy and time resolution.