Abstract:

To solve the problem of low emotional recognition rate of EEG signals, this paper proposes an EEG emotion recognition method based on 3DC-BGRU. Firstly, the Short-Time Fourier Transform（STFT） is performed on the single-channel EEG signal, and the time-frequency information of the relevant frequency band is extracted to form a two-dimensional time-frequency map, the time-frequency map of multiple channels constitutes a new three-dimensional data form of time, frequency and channels. Then, a novel Convolutional Neural Network（CNN） model is designed to extract deep features of 3D data by 3D convolution. Finally, the Bidirectional Gated Recurrent Unit（BGRU） is designed to process the sequence information of deep features and classify them with Softmax. The experimental results show that the classification recognition rate of the method is improved.

Key words: emotion recognition, Short-Time Fourier Transform（STFT）, three-dimensional data, Convolutional Neural Network（CNN）, Bidirectional Gated Recurrent Unit（BGRU）

摘要：

针对脑电信号情感识别率偏低的问题，提出了一种基于3DC-BGRU的脑电情感识别方法。对单通道脑电信号进行短时傅里叶变换（STFT），提取相关频带的时频信息构成二维时频图，并将多个通道的时频图构成一种全新的时间、频率和通道的三维数据形式，通过三维卷积的方式设计了一种新颖的卷积神经网络（CNN）模型对三维数据进行深层特征提取，设计双向门控循环单元（BGRU）对深层特征的序列信息进行处理并配合Softmax进行分类。实验结果表明该方法分类识别率得到提高。

关键词: 情感识别, 短时傅里叶变换（STFT）, 三维数据, 卷积神经网络（CNN）, 双向门控循环单元（BGRU）