基于脑网络节点波动的EEG情感识别方法

An EEG emotion recognition method based on node fluctuations in brain networks

为探究大脑情感处理机制,提出了一种基于脑网络的节点波动脑区选择方法。通过采集处理情感脑区的脑电信号,基于锁相值构建脑网络,计算脑网络的节点波动指标,提取其差分熵特征,建立121500个特征数据集,分类识别脑区情感。结果表明,在不同情感状态下,高频波段大脑的左右颞区能量增强,提取相应脑区的差分熵特征,选用支持向量机进行分类,分类结果为88.27%。在使用12通道数的情况下,文中所选择的脑区差分熵的识别率明显提高了1.26%。

This paper describes the exploration of the mechanism of brain emotion processing and proposes a method of nodal fluctuation brain region selection based on brain network.The study works by constructing a brain network based on the phase-locked value by collecting and processing the EEG signal in the affective brain area;calculating the node fluctuation index;extracting its differential entropy feature;establishing 121500 feature data sets to classify and identify the emotion in the brain area.The results show that the energy of the left and right temporal region of the brain in the high frequency band is enhanced under different emotional states.The differential entropy feature of the corresponding brain region is extracted and the classification result is 88.27%by using support vector machine.When the number of channels is 12,the recognition rate of the brain region differential entropy selected in this paper obviously increases by 1.26%.