基于脑电信号的情绪分类

Emotion classification based on EEG signal

针对不同情绪的分类提出了基于脑电信号的情绪分类方法。采用了图片和视频2种情绪刺激方式,分别设计了积极和消极2种情绪作为刺激方案,采集受试者受到刺激后的脑电信号。先将信号去噪,再通过共空间模式算法对信号进行特征提取,最后应用支持向量机算法(SVM)将脑电信号进行分类,分类的优劣用分类率表示。通过这种方法实现了不同刺激方式下情绪脑电信号的分类以及对比了不同刺激方式的分类率,发现2种刺激方式都能将情绪分类且分类率均在75%以上,视频刺激以88.97%的信号识别率高于图像刺激79.24%的信号识别率。分类结果表明,有效的情绪刺激方式能够将不同情绪脑电信号很好地分类。情绪分类的研究为抑郁症以及压力比较大的群体的情绪变化提供了较好的参考价值。

In this paper, a method of emotion classification based on Electroencephalograph （EEG） signals is proposed. Two kinds of emotional stimuli including pictures and videos are used to design positive and negative emotions as a stimulus program. Then EEG signals of subjects are collected. After signal denoising, the feature extraction is carried out by the common spatial pattern algorithm. Finally, SVM is used to classify EEG signals. The classification rate indicates the quality of classification. The comparison of different stimulus patterns of EEG in the emotional classification is realized by this method. And the classification rates of different stimulation modes are compared. It is found that the two kinds of stimuli can classify the emotion and the classification rate is above 75%. Also, the recognition rate of 88.97% of the video stimulus is higher than that of the image stimulus with the recognition rate at 79.24%. The results show that the effective emotion stimulation method can classify the EEG signals of different emotions very well. The study of emotion classification provides a good reference value for the depression and the emotional changes of the group with large pressure.