基于小波包分解和近似熵的脑电信号分类

Classification of EEG signals based on wavelet packet decomposition and approximate entropy

为提高脑电信号分类准确率,提出基于小波包分解和近似熵相结合的特征提取方法。该方法利用小波包对信号的低频和高频进行分解,用近似熵对得到的叶子结点进行计算得到特征值,然后将其输入支持向量机进行分类。实验结果表明,该方法在两种思维结合识别中正确率最高达到了97.37%,取得了较好的分类效果。

In order to improve the classification accuracy of EEG signals,we propose a feature extraction method based on wavelet packet decomposition and approximate entropy. The proposed method uses wavelet packet to decompose the low and high frequency of EEG signals,and uses approximate entropy to compute the feature values for the obtained leaf nodes. Obtained feature values are used as the inputs of support vector machine to classify. Experimental results show that the proposed method achieves classification accuracy over 97. 37% in the case of the combination of two mental tasks,which obtains a good classification accuracy.