In recent years, research on emotion recognition has become more and more popular, but there are few studies on emotion recognition based on cerebral blood oxygen signals. Since the electroencephalogram (EEG) is easily disturbed by eye movement and the portability is not high, this study uses a more comfortable and convenient functional near-infrared spectroscopy (fNIRS) system to record brain signals from participants while watching three different types of video clips. During the experiment, the changes in cerebral blood oxygen concentration in the 8 channels of the prefrontal cortex of the brain were collected and analyzed. We processed and divided the collected cerebral blood oxygen data, and used multiple classifiers to realize the identification of the three emotional states of joy, neutrality, and sadness. Since the classification accuracy of the convolutional neural network (CNN) in this research is not significantly superior to that of the XGBoost algorithm, this paper proposes a CNN-Transformer network based on the characteristics of time series data to improve the classification accuracy of ternary emotions. The network first uses convolution operations to extract channel features from multi-channel time series, then the features and the output information of the fully connected layer are input to the Transformer netork structure, and its multi-head attention mechanism is used to focus on different channel domain information, which has better spatiality. The experimental results show that the CNN-Transformer network can achieve 86.7% classification accuracy for ternary emotions, which is about 5% higher than the accuracy of CNN, and this provides some help for other research in the field of emotion recognition based on time series data such as fNIRS.

Blood oxygen content plays an important role in clinical diagnosis and treatment. However, interference signals will be introduced in the process of the signal acquisition, which will affect the accuracy of measurement results. Although the filter can eliminate most of the interference, there will still be a part of the residual interference signal. In order to eliminate these residual interference, this paper proposes a signal denoising method based on ensemble empirical mode decomposition (EEMD), lifting wavelet transform (LWT) and singular spectrum analysis (SSA). Firstly, the measured signal is decomposed by EEMD to obtain multiple modal components (IMF). The high-frequency components are denoised by LWT, and then reconstructed with the low-frequency components. Finally, the SSA is carried out on the reconstructed signal, the appropriate components are selected to reconstruct the signal to obtain the final denoised signal. After testing the EEMD-LWT-SSA method with simulated and measured signals, it can be seen that the method can effectively suppress the residual noise in signals and improve the accuracy of data.

Functional near-infrared spectroscopy (fNIRS) can dynamically respond to the relevant state of brain activity based on the hemodynamic information of brain tissue. The cerebral cortex and gray matter are the main regions reflecting brain activity. As they are far from the scalp surface, the accuracy of brain activity detection will be significantly affected by a series of physiological activities. In this paper, an effective algorithm for extracting brain activity information is designed based on the measurement method of dual detectors so as to obtain real brain activity information. The principle of this algorithm is to take the measurement results of short-distance channels as reference signals to eliminate the physiological interference information in the measurement results of long-distance channels. In this paper, the performance of the proposed method is tested using both simulated and measured signals and compared with the extraction results of EEMD-RLS, RLS and fast-ICA, and their extraction effects are quantified by correlation coefficient (R), root-mean-square error (RMSE), and mean absolute error (MAE). The test results show that even under low SNR conditions, the proposed method can still effectively suppress physiological interference and improve the detection accuracy of brain activity signals.