Asynchronous electroencephalogram signal emotion recognition method and system based on high emotion discrimination information

Abstract

The invention belongs to the technical field of signal processing, and discloses an asynchronous electroencephalogram signal emotion recognition method and system based on high emotion discrimination information, and the method comprises the steps: obtaining asynchronous electroencephalogram signals under different emotions; obtaining a high-emotion frequency band of the preprocessed electroencephalogram signal, and extracting various electroencephalogram features from a high-emotion segment; matrix conversion is carried out on the multiple electroencephalogram features, and discriminative emotion information of the multiple electroencephalogram features is obtained; fusing the emotion information with the discriminative electroencephalogram features to obtain final fusion information; and finally, inputting the fused information and the emotion label into machine learning for training to obtain an emotion recognition model. According to the asynchronous electroencephalogram emotion recognition method based on the high emotion discrimination information, the distribution difference between different emotion categories of different subjects can be mined, and the generalization of emotion recognition is improved.

Description

Technical Field

[0001] The present invention relates to the field of signal processing technology, and in particular to an asynchronous electroencephalogram signal emotion recognition method and system based on high emotion discrimination information. Background Art

[0002] In recent years, with the widespread application of EEG signals in the field of emotion recognition, it has shown great potential in emotional computing, mental health assessment, human-computer interaction, etc. However, the current EEG signal data is mainly acquired based on synchronous acquisition technology, resulting in high power consumption of acquisition equipment. In addition, due to individual differences and the non-stationarity of EEG signals, the generalization ability of traditional emotion recognition models is insufficient, making it difficult to achieve stable and accurate emotion recognition.

[0003] At present, EEG data used for emotion recognition is mainly acquired through synchronous acquisition, and there has been no research using asynchronous acquisition technology for cross-subject emotion recognition. In addition, in order to address the problem of poor generalization, most studies focus on data distribution under the same emotion to improve cross-subject emotion recognition performance, but fail to fully consider the distribution differences between different emotion categories of different subjects.

[0004] Therefore, in order to solve the above problems, the present invention proposes an asynchronous EEG signal emotion recognition method and system based on high emotion discrimination information. Summary of the invention

[0005] The purpose of the present invention is to provide an asynchronous EEG signal emotion recognition method based on high emotion discrimination information, comprising the following steps:

[0006] S1. Acquire original EEG signal data, preprocess the original EEG signal data, and obtain preprocessed EEG signal data, wherein the original EEG signal data is EEG signal data stimulated by different EEG experimental procedures;

[0007] S2. extracting high-emotion EEG signal segments from the preprocessed EEG signal data, and performing feature extraction on the high-emotion EEG signal segments to obtain EEG signal features, wherein the EEG signal features include differential entropy features and power spectral density features;

[0008] S3, performing matrix conversion on the EEG signal features to obtain emotion discrimination information corresponding to multiple features, and fusing the emotion discrimination information corresponding to the multiple features to obtain a discrimination information set;

[0009] S4, inputting the discriminant information set and the corresponding emotion label into the machine learning model for training to obtain an emotion recognition model;

[0010] S5. Input the data of the new subject into the emotion recognition model to obtain the emotion recognition result.

[0011] Preferably, in S1, under the stimulation of different emotions, asynchronous EEG signals are acquired using asynchronous acquisition technology SARADC.

[0012] Preferably, the preprocessing of the original EEG signal data to obtain preprocessed EEG signal data specifically includes:

[0013] Using a denoising module to filter and remove artifacts from the original EEG signal data to obtain denoised EEG signal data, and extract EEG signal data of different frequency bands;

[0014] The EEG signal data of different frequency bands are the pre-processed EEG signal data.

[0015] Preferably, extracting high-emotion EEG signal segments from the preprocessed EEG signal data in S2 specifically includes:

[0016] Segmenting the preprocessed EEG signal data, wherein each segment has a length of L seconds and has no overlap;

[0017] Since EEG signals contain multiple channels, where c i Represents the segmented data from the i-th EEG signal channel;

[0018] Calculate C L The Pearson correlation p between the mid-segments and the eigenvalue λ of p are calculated C and the total correlation strength H;

[0019] in I represents the number of channels;

[0020] Calculate the average H value of all segments, and select segments with H greater than the average H value as high-emotion segments.

[0021] Preferably, the plurality of EEG signal features are matrix-converted respectively, and emotion discrimination information corresponding to the plurality of features is extracted respectively, specifically including:

[0022] Extract the difference θ of the same emotion category among different subjects IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS ;

[0023] Using the transformation matrix W = Φ T B maps the features from RKHS to q-dimensional space to achieve feature transformation and obtain the difference θ of the same emotion category among different subjects. IDS , Differences in different emotion categories among different subjects θ IDD, category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS The transformation matrix. Where Φ = [φ(x1), φ(x2), ..., φ(x m )] T represents a typical feature set, φ(x m ) represents the sample x m In the feature representation of RKHS, B is the coefficient matrix;

[0024] Calculate the difference θ of the same emotion category among different subjects IDS The characteristic distance Differences in different emotion categories among different subjects θ IDD The characteristic distance Category difference θ CDD The characteristic distance The divergence between emotion classes of different subjects θ IDBS The characteristic distance Emotional divergence θ of different subjects IDWS The characteristic distance

[0025] Preferably,

[0026]

[0027] Among them, C cla represents the number of emotion categories, represents the number of any two combinations of emotion categories. μ represents the mean in the Reproducing Kernel Hilbert Space (RKHS), and u represents the average value of a certain type of emotion data. Represents the average value of a certain type of sentiment data in RKHS, represents the square norm in RKHS. x represents the EEG feature, N represents the number of source subjects, The number of samples representing a certain type of emotion, represents the average value from all source subjects;

[0028] in,

[0029] in,

[0030]

[0031] Maximization Then find B. Where η Q η G η Fη P is the equilibrium parameter;

[0032] Get the first q eigenvalues B is the corresponding eigenvector;

[0033] The differential entropy feature and the power spectral density feature are passed through B to obtain the converted differential entropy feature and the power spectral density feature.

[0034] Preferably, the emotion discrimination information corresponding to the multiple features is fused to obtain a discrimination information set, including:

[0035] The converted differential entropy feature discriminative information and the power spectrum density feature discriminative information are fused to obtain fused discriminative information;

[0036] The final set of discriminant information and the corresponding emotion labels are input into a machine learning model for training to obtain an emotion recognition model, which specifically includes:

[0037] Input the fused features and emotion labels into the spiking neural network to obtain a training module;

[0038] The data of the new subject is input into the emotion recognition model to obtain the emotion recognition results, which mainly include:

[0039] Obtain the fused discriminant information of the new subject;

[0040] The fused discrimination information of the new subject is input into the spiking neural network to obtain an emotion recognition result.

[0041] Asynchronous EEG signal emotion recognition system based on high emotion discrimination information, including:

[0042] Data processing module: acquiring raw EEG signal data, preprocessing the raw EEG signal data, and obtaining preprocessed EEG signal data, wherein the raw EEG signal data is EEG signal data stimulated by different EEG experimental procedures;

[0043] Feature extraction module: extracting high-emotion EEG signal segments from the preprocessed EEG signal data, and performing feature extraction on the high-emotion EEG signal segments to obtain EEG signal features, wherein the EEG signal features include differential entropy features and power spectral density features;

[0044] Information fusion module: Perform matrix conversion on EEG signal features to obtain emotion discrimination information corresponding to multiple features, and fuse the emotion discrimination information corresponding to multiple features to obtain a discrimination information set;

[0045] Training module: used to input the fused information and emotion labels into the machine learning model to obtain the emotion recognition model;

[0046] The testing module is used to input the EEG signal characteristics of a new subject into the emotion recognition model to obtain the emotion recognition result.

[0047] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, the asynchronous electroencephalogram signal emotion recognition method based on high emotion discrimination information is implemented.

[0048] A computer-readable storage medium storing computer-executable instructions; the computer-executable instructions are configured to execute the above-mentioned asynchronous EEG signal emotion recognition method based on high emotion discrimination information.

[0049] Therefore, the present invention adopts the above-mentioned asynchronous EEG signal emotion recognition method and system based on high emotion discrimination information to explore the distribution differences between different emotion categories of different subjects and improve the generalization of emotion recognition.

[0050] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of the asynchronous EEG signal emotion recognition method based on high emotion discrimination information provided by the present invention;

[0052] Figure 2 A flowchart of an implementation of an asynchronous EEG signal emotion recognition method based on high emotion discrimination information provided by an embodiment of the present invention;

[0053] Figure 3 A stimulation flow chart provided by an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the structure of the asynchronous EEG signal emotion recognition system based on high emotion discrimination information provided by the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0056] The embodiment of the present invention discloses an EEG signal emotion recognition method, that is, an asynchronous EEG signal emotion recognition method based on high emotion discriminant information. The method comprises adopting the successive approximation register analog-to-digital conversion (SARADC) technology to obtain the original asynchronous EEG signal; then, extracting the high emotion segment of the EEG signal, and extracting the differential entropy feature and power spectrum density feature of the segment; then, performing matrix conversion on the EEG features of the segment respectively, and extracting the discriminant information by mining the information of the same emotion category difference of different subjects, the difference of different emotion categories of different subjects, the category difference, the divergence between emotion categories of different subjects, and the divergence within emotion categories of different subjects; secondly, using the fusion method to fuse the power spectrum density discriminant information and the differential entropy discriminant information, and inputting the fused information and the emotion label into the pulse neural network model to obtain the emotion recognition model; finally, inputting the data of the new subject into the emotion recognition model to obtain the emotion recognition result.

[0057] like Figure 1 As shown, the asynchronous EEG signal emotion recognition method based on high emotion discrimination information provided by the present invention includes:

[0058] S1, obtaining raw EEG signal data (obtaining raw EEG signal data in an asynchronous acquisition mode, i.e., SARADC), preprocessing the raw EEG signal data to obtain preprocessed EEG signal data, wherein the raw EEG signal data is EEG signal data stimulated by different EEG experimental procedures;

[0059] The denoising module is used to filter and remove artifacts from the original EEG signal data to obtain denoised EEG signal data;

[0060] The denoised EEG signal data is divided into frequency bands, including delta (1-3 Hz), theta (4-7 Hz), alpha (8-13 Hz), beta (14-30 Hz), and gamma (31-50 Hz), to obtain preprocessed EEG signal data.

[0061] S2. extracting high-emotion EEG signal segments from the preprocessed EEG signal data, and performing feature extraction on the high-emotion EEG signal segments to obtain EEG signal features, wherein the EEG signal features include differential entropy features and power spectral density features;

[0062] Segmenting the preprocessed EEG signal data, each segment being 1 second long and without overlap;

[0063] Since EEG signals contain multiple channels, where c iRepresents the segmented data from the i-th EEG signal channel;

[0064] Calculate C L The Pearson correlation p between the mid-segments and the eigenvalue λ of p are calculated C and the total correlation strength H;

[0065] in I represents the number of channels;

[0066] Calculate the average H value of all segments, and select segments with H greater than the average H value as high-emotion segments.

[0067] S3, performing matrix conversion on the EEG signal features to obtain emotion discrimination information corresponding to multiple features, and fusing the emotion discrimination information corresponding to the multiple features to obtain a discrimination information set;

[0068] Extract the difference θ of the same emotion category among different subjects IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS ,

[0069] in:

[0070]

[0071] Among them, C cla represents the number of emotion categories, represents the number of any two combinations of emotion categories. μ represents the mean in the Reproducing Kernel Hilbert Space (RKHS), and u represents the average value of a certain type of emotion data. Represents the average value of a certain type of sentiment data in RKHS, represents the square norm in RKHS; x represents the EEG feature, N represents the number of source subjects, The number of samples representing a certain type of emotion, Represents the average value from all source subjects.

[0072] Using the transformation matrix W = Φ T B maps the features from RKHS to q-dimensional space to achieve feature transformation and obtain the difference θ of the same emotion category among different subjects. IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWSThe transformation matrix. Where Φ = [φ(x1), φ(x2), ..., φ(x m )] T represents a typical feature set, φ(x m ) represents the sample x m In the feature representation of RKHS, B is the coefficient matrix.

[0073] Calculate the difference θ of the same emotion category among different subjects IDS The characteristic distance Differences in different emotion categories among different subjects θ IDD The characteristic distance Category difference θ CDD The characteristic distance The divergence between emotion classes of different subjects θ IDBS The characteristic distance Emotional divergence θ of different subjects IDWS The characteristic distance

[0074] in,

[0075] in

[0076]

[0077] Maximization Then find B. Where η Q η G η F η P is a balance parameter.

[0078] Get the first q eigenvalues

[0079] The differential entropy feature and the power spectral density feature are passed through B to obtain the converted differential entropy feature and the power spectral density feature.

[0080] The emotion discrimination information corresponding to the multiple features is integrated to obtain a final discrimination information set, which specifically includes:

[0081] The CCA algorithm is used to fuse the converted differential entropy features and power spectrum density features;

[0082] S4, inputting the discriminant information set and the corresponding emotion label into the machine learning model for training to obtain an emotion recognition model;

[0083] Emotion labels are the classification and identification of emotional states, which help us better understand and describe our emotions, as well as communicate our emotions with others. They include positive emotions (joy, gratitude, confidence, love, satisfaction), negative emotions (anger, fear, sadness, anxiety, depression), and neutral emotions (calmness, curiosity, boredom, contempt, surprise). Emotion labels are a complex and diverse system that covers all aspects of human emotions. By accurately identifying and applying these emotion labels, we can better understand the emotional world of ourselves and others, and promote more effective communication and healthier interpersonal relationships.

[0084] Inputting the final set of discriminant information and the corresponding emotion labels into a machine learning model for training to obtain an emotion recognition model;

[0085] Calculating the average value of the final discriminant information set;

[0086] Compare the size of each element of the final discrimination information set with the average value, when the element is greater than the average value, the element becomes 1, when it is less than the negative value of the average value, it becomes -1, otherwise it becomes 0, and a pulse code is obtained;

[0087] Inputting the pulse code and the corresponding emotion label into a spiking neural network to obtain an emotion recognition model;

[0088] S5. Input the data of the new subject into the emotion recognition model to obtain the emotion recognition result.

[0089] Obtain the fused discriminant information of the new subject;

[0090] Obtaining pulse coding of the fused discrimination information of the new subject;

[0091] The pulse code is input into the emotion recognition model to obtain an emotion recognition result.

[0092] The following uses the above-mentioned asynchronous EEG signal emotion recognition method based on high emotion discrimination information to identify emotions as an example to illustrate the superiority of the technical solution provided by the present invention. In the specific application process, changes to the numerical values ​​all fall within the protection scope of the present invention.

[0093] like Figure 2 As shown in Figure 1, the process of emotion recognition for new subjects is as follows:

[0094] First, the EEG experiment was designed using the experimental design module, including selecting subjects and stimuli. The selected subjects all had normal hearing and were right-handed. The subjects all understood the purpose, process, and precautions of the experiment in advance. The selected stimuli included emotional videos and emotional music, and they evaluated their emotions after each stimulus.

[0095] The stimulation is played through the experimental playback module, and the raw EEG signal data is collected through the asynchronous acquisition method SARADC EEG acquisition module.

[0096] The EEG denoising module is used to filter out 50Hz power frequency, electrooculogram and other artifacts.

[0097] The frequency band extraction module is used to extract delta (1-3Hz), theta (4-7Hz), alpha (8-13Hz), beta (14-30Hz), and gamma (31-50Hz) frequency band data.

[0098] The high-emotion segment module is used to extract high-emotion segments in each frequency band.

[0099] The feature extraction module is used to extract the differential entropy features and power spectral density features of the EEG signal, and the EEG features of all frequency bands are horizontally spliced.

[0100] Select q as 3, use the feature conversion module to extract discriminative information, and obtain differential entropy discriminative information and power spectrum density discriminative information.

[0101] The information fusion module is used to fuse the differential entropy discriminant information and the power spectrum density discriminant information using the canonical correlation analysis fusion algorithm to obtain the fused discriminant information.

[0102] The emotion labels and the fused discriminative information are input into the pulse neural network model training to obtain the emotion recognition model.

[0103] Input the data of the new subject into the emotion recognition model to obtain the emotion recognition results.

[0104] Figure 3 To stimulate the process, the specific operations are:

[0105] The subjects were placed in a sound-shielded room. During the experiment, the lighting conditions were kept consistent and the indoor temperature was kept comfortable. Absolute silence was maintained during the experiment, and the volume was adjusted to a comfortable range for people to hear the sound and remained consistent.

[0106] The subjects were asked to reduce unnecessary movements and maintain a comfortable sitting posture during the experiment.

[0107] The experimenter started the whole experiment after the subject confirmed that it was okay to start.

[0108] This specific implementation method selected 12 subjects, all of whom had normal hearing and were right-handed.

[0109] According to the method of the present invention, the EEG signal data is first obtained, and then multi-feature high-emotion discrimination information is obtained, and then an emotion recognition model is obtained, and finally the emotion of a new subject is recognized through the emotion recognition model.

[0110] Corresponding to the above-mentioned asynchronous EEG signal emotion recognition method based on high emotion discrimination information, the present invention also provides an asynchronous EEG signal emotion recognition system based on high emotion discrimination information, such as Figure 4 As shown, the system includes: an asynchronous EEG signal data acquisition module, a data processing module, a high-emotion segment extraction module, a feature extraction module, a feature conversion module, an information fusion module, a training module, and a testing module.

[0111] The asynchronous EEG signal data acquisition module is used to acquire raw EEG signal data. The raw EEG signal data is the EEG signal data stimulated by different EEG experimental programs.

[0112] The data processing module is used to preprocess the original asynchronous EEG signal data to obtain preprocessed EEG signal data;

[0113] The high-emotion segment extraction module is used to extract high-emotion segments from the processed EEG signal data.

[0114] The feature extraction module is used to extract various EEG signal features.

[0115] The feature conversion module is used to convert the multiple EEG signal features to obtain discriminative emotion information corresponding to the multiple EEG signal features.

[0116] The information fusion module is used to fuse the discriminative emotion information corresponding to multiple EEG signal features to obtain fused information.

[0117] The training module is used to input the fused information and emotion labels into the spiking neural network model to obtain an emotion recognition model;

[0118] The testing module is used to input the EEG signal characteristics of a new subject into the emotion recognition model to obtain the emotion recognition result.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. Asynchronous EEG signal emotion recognition method based on high emotion discrimination information, characterized in that: The following steps are involved: S1. Acquire original EEG signal data, preprocess the original EEG signal data, and obtain preprocessed EEG signal data, wherein the original EEG signal data is EEG signal data stimulated by different EEG experimental procedures; S2. extracting high-emotion EEG signal segments from the preprocessed EEG signal data, and performing feature extraction on the high-emotion EEG signal segments to obtain EEG signal features, wherein the EEG signal features include differential entropy features and power spectral density features; S3, performing matrix conversion on the EEG signal features to obtain emotion discrimination information corresponding to multiple features, and fusing the emotion discrimination information corresponding to the multiple features to obtain a discrimination information set; S4, inputting the discriminant information set and the corresponding emotion label into the machine learning model for training to obtain an emotion recognition model; S5. Input the data of the new subject into the emotion recognition model to obtain the emotion recognition result.

2. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 1 is characterized in that: In S1, under the stimulation of different emotions, asynchronous EEG signals are acquired using asynchronous acquisition technology SARADC.

3. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 1 is characterized in that: The preprocessing of the original EEG signal data to obtain preprocessed EEG signal data specifically includes: Using a denoising module to filter and remove artifacts from the original EEG signal data to obtain denoised EEG signal data, and extract EEG signal data of different frequency bands; The EEG signal data of different frequency bands are the pre-processed EEG signal data.

4. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 3 is characterized in that: In S2, high-emotion EEG signal segments are extracted from the preprocessed EEG signal data, specifically including: Segmenting the preprocessed EEG signal data, wherein each segment has a length of L seconds and has no overlap; Since EEG signals contain multiple channels, where c i Represents the segmented data from the i-th EEG signal channel; Calculate C L The Pearson correlation p between the mid-segments and the eigenvalue λ of p are calculated C and the total correlation strength H; in, I represents the number of channels; Calculate the average H value of all segments, and select segments with H greater than the average H value as high-emotion segments.

5. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 3 is characterized in that: The EEG signal features are transformed into matrices respectively, and the emotion discrimination information corresponding to various features is extracted respectively, including: Extract the difference θ of the same emotion category among different subjects IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS ; Using the transformation matrix W = Φ T B maps the features from RKHS to q-dimensional space to achieve feature transformation and obtain the difference θ of the same emotion category among different subjects. IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS The transformation matrix of Where Φ=[φ(x1),φ(x2),...,φ(x m )] T represents a typical feature set, φ(x m ) represents the sample x m In the feature representation of RKHS, B is the coefficient matrix; Calculate the difference θ of the same emotion category among different subjects IDS The characteristic distance Differences in different emotion categories among different subjects θ IDD The characteristic distance Category difference θ CDD The characteristic distance The divergence between emotion classes of different subjects θ IDBS The characteristic distance Emotional divergence θ of different subjects IDWS The characteristic distance 6. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 3 is characterized in that: Differences in the same emotion category among different subjects θ IDS , Differences in different emotion categories among different subjects θ IDD , category difference θ CDD , the divergence between emotion classes of different subjects θ IDBS , the divergence within emotion categories of different subjects θ IDWS The expressions are: Among them, C cla represents the number of emotion categories, represents the number of any two combinations of emotion categories, μ represents the mean in the Reproducing Kernel Hilbert Space (RKHS), and u represents the average value of a certain type of emotion data. Represents the average value of a certain type of sentiment data in RKHS, represents the square norm in RKHS; x represents the EEG feature, N represents the number of source subjects, The number of samples representing a certain type of emotion, represents the average value from all source subjects; in, in Maximization Then find B; where η Q η G η F η P is the equilibrium parameter; Get the first q eigenvalues ​​▽=(λ1,...,λ q ), B is the corresponding eigenvector; The differential entropy feature and the power spectral density feature are passed through B to obtain the converted differential entropy feature and the power spectral density feature.

7. The asynchronous EEG signal emotion recognition method based on high emotion discrimination information according to claim 3 is characterized in that: The emotional discrimination information corresponding to multiple features is integrated to obtain a set of discrimination information, including: The converted differential entropy feature discriminative information and the power spectrum density feature discriminative information are fused to obtain fused discriminative information; The final set of discriminant information and the corresponding emotion labels are input into a machine learning model for training to obtain an emotion recognition model, which specifically includes: Input the fused features and emotion labels into the spiking neural network to obtain a training module; The data of the new subject is input into the emotion recognition model to obtain the emotion recognition results, which mainly include: Obtain the fused discriminant information of the new subject; The fused discrimination information of the new subject is input into the spiking neural network to obtain an emotion recognition result.

8. Asynchronous EEG signal emotion recognition system based on high emotion discrimination information, characterized in that: include: Data processing module: acquiring raw EEG signal data, preprocessing the raw EEG signal data, and obtaining preprocessed EEG signal data, wherein the raw EEG signal data is EEG signal data stimulated by different EEG experimental procedures; Feature extraction module: extracting high-emotion EEG signal segments from the preprocessed EEG signal data, and performing feature extraction on the high-emotion EEG signal segments to obtain EEG signal features, wherein the EEG signal features include differential entropy features and power spectral density features; Information fusion module: Perform matrix conversion on EEG signal features to obtain emotion discrimination information corresponding to multiple features, and fuse the emotion discrimination information corresponding to multiple features to obtain a discrimination information set; Training module: used to input the fused information and emotion labels into the machine learning model to obtain the emotion recognition model; Testing module: used to input the EEG signal features of a new subject into the emotion recognition model to obtain emotion recognition results.

9. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it implements the asynchronous EEG signal emotion recognition method based on high emotion discrimination information as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions; the computer-executable instructions are configured to execute the above-mentioned asynchronous EEG signal emotion recognition method based on high emotion discrimination information.

Images