A method and system for emotion computing based on wearable physiological signal microstate

Abstract

The application provides a kind of based on wearable physiological signal microstate emotional computing method and system, comprising: obtaining multi-channel physiological signal by wearable acquisition device;After extracting the features of multi-channel human physiological signal, dimension reduction is carried out, the feature dimension is compressed, and the physiological feature categories are obtained by clustering analysis on the compressed dimension features, and the microstate division of physiological signal is carried out;Based on the microstate division result of the physiological signal, the mapping of original physiological signal and microstate category is carried out, the continuous physiological signal is converted into microstate sequence, and the microstate sequence feature is extracted;The microstate sequence feature is input into the preset emotional state prediction model, and the emotional computing result is generated by the emotional state prediction model.The application solves the problem that the prior art cannot accurately analyze human emotional state from physiological signal.

Description

Technical Field

[0001] This invention relates to the field of physiological signal analysis technology, and in particular to an emotion computing method and system based on wearable physiological signal microstates. Background Technology

[0002] Affective computing technology refers to the technology of using computers to collect, analyze, and identify emotion-related individual characteristics. The development of affective computing increasingly requires the interdisciplinary integration of psychology, cognitive science, and other disciplines. Among emotion-related individual characteristics, physiological signals are objective and difficult to forge, making them highly valuable for research. Measurement of physiological signals based on wearable sensing devices is a relatively convenient method, extending physiological measurement to a wide range of daily life scenarios while maintaining low user workload during continuous long-term measurements. In recent years, affective computing methods have increasingly emphasized the measurement and analysis of wearable physiological signals, including heart rate and electrodermal conductivity (EDS).

[0003] To date, emotion computing methods based on wearable physiological signals still face challenges, one significant being the lack of attention to the temporal dynamics of emotional physiological responses. Most existing studies typically adopt a static perspective, simply comparing emotional states with baselines or averaging physiological signal characteristics across different emotional states. However, functionalist theory and extensive empirical research demonstrate that changes in physiological responses induced by different emotions affect multiple bodily systems and exhibit unique temporal patterns. Therefore, recording multimodal physiological changes and analyzing their temporal dynamics will help accurately identify individual emotional states, bringing incremental value to current emotion computing based on physiological signals.

[0004] Microstate analysis is a crucial method for observing the temporal dynamics of physiological signals. Recent laboratory studies have revealed that the microstate patterns of multichannel recorded electroencephalograms (EEGs) can reflect temporal dynamics and can be used in the field of EEG-based emotion computation. Specifically, researchers clustered the spatial features of EEG signals within a 60–120 ms time window, finding that the spatial patterns exhibit multiple categories, termed EEG microstates (commonly four categories: A, B, C, and D). Each microstate possesses inherent stability, corresponding to a set of cross-channel activation patterns in the EEG signals. Based on this method, the dynamic changes of a relatively long physiological signal can be characterized by combinations of many microstates. By calculating the proportion, duration, and transition probability of each microstate, studies have found significant differences in the EEG microstate characteristics of individuals under different emotional states. Microstate analysis of multichannel EEG signals can effectively identify individual emotional states.

[0005] However, while the effectiveness of the microstate method has been validated in EEG analysis, its application in other physiological signals is currently rare. Physiological signals measured by wearable devices are closely related to central nervous system signals such as EEG. Wearable measurement devices can typically monitor multimodal or multichannel physiological signals, thus providing a data foundation for microstate analysis. Compared to traditional static analysis, microstate analysis based on wearable physiological signals can obtain higher-dimensional temporal information, improving the effectiveness and accuracy of emotion recognition. Furthermore, the microstate method holds promise for revealing the characteristics of specific temporal and cross-modal changes in different emotion categories, providing valuable insights for future research in emotion computing. Summary of the Invention

[0006] This invention provides an emotion computing method and system based on wearable physiological signal microstates, which solves the problem that existing technologies are unable to accurately analyze human emotional states from physiological signals.

[0007] This invention provides an emotion computing method based on wearable physiological signal microstates, comprising:

[0008] Acquire multi-channel physiological signals using wearable data acquisition devices;

[0009] After extracting features from multi-channel human physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories and perform micro-state division of physiological signals.

[0010] Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, the continuous physiological signals are transformed into microstate sequences, and the microstate sequence features are extracted.

[0011] The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results.

[0012] According to the present invention, an emotion computing method based on wearable physiological signal microstates is provided, wherein acquiring multi-channel physiological signals through a wearable acquisition device specifically includes:

[0013] Multi-channel human peripheral physiological signals are acquired using wearable acquisition devices at a sampling rate greater than 1 Hz.

[0014] The collected multi-channel human peripheral physiological signals are aggregated and then time-aligned.

[0015] According to the present invention, an emotion computing method based on wearable physiological signal microstates is provided, wherein the extraction of features from multi-channel human physiological signals is followed by dimensionality reduction, feature compression, and cluster analysis of the compressed features to obtain physiological feature categories. Specifically, this includes:

[0016] Acquire multi-channel human physiological signals, segment them according to time windows, and extract standardized physiological features;

[0017] Based on standardized physiological characteristics, all physiological characteristics are summarized and dimensionality reduced. Principal component analysis is used to extract the feature dimensions with higher weights, and cluster analysis is performed on the dimensionality-reduced physiological characteristics.

[0018] According to the present invention, an emotion computing method based on wearable physiological signal microstates includes, in which the clustering analysis of compressed features is performed to obtain physiological feature categories, and the microstates of physiological signals are divided, specifically including:

[0019] Check the distribution of cluster analysis results, exclude clusters with too small a sample size or those concentrated on only a few individuals, and perform a stability check on the cluster results.

[0020] The effective feature categories are named as corresponding microstates, the microstate division is completed, and the physiological characteristics of the cluster center point mapped to the original feature space are marked.

[0021] According to the present invention, an emotion computing method based on wearable physiological signal microstates is provided. The method involves mapping the original physiological signals to microstate categories based on the microstate segmentation results of the physiological signals, transforming continuous physiological signals into microstate sequences, and extracting microstate sequence features. Specifically, this includes:

[0022] After completing the microstate category classification, the mapping method from the original physiological signal to the microstate category is obtained;

[0023] Based on the mapping method from the original physiological signal to the microstate category, long continuous physiological signals are transformed into microstate sequences, and the features of the microstate sequences are extracted.

[0024] According to the present invention, an emotion computing method based on wearable physiological signal microstates is provided, wherein the microstate sequence features are input into a preset emotion state prediction model, and the emotion state prediction model is used to generate emotion computing results, specifically including:

[0025] Micro-state sequence features are input into a pre-defined emotion state prediction model;

[0026] The emotion state prediction model maps micro-state sequence features to emotion labels and outputs emotion calculation results.

[0027] The emotional labels are derived from emotionally induced material labels, subjects' self-reports, and comprehensive behavioral performance.

[0028] This invention also provides an emotion computing system based on wearable physiological signal microstates, the system comprising:

[0029] The signal acquisition module is used to acquire multi-channel physiological signals through wearable acquisition devices;

[0030] The micro-state partitioning module is used to extract features from multi-channel human physiological signals, reduce the dimensionality, compress the feature dimensions, and perform cluster analysis on the compressed features to obtain physiological feature categories, thereby performing micro-state partitioning of physiological signals.

[0031] The sequence feature extraction module is used to map the original physiological signal to the microstate category based on the microstate segmentation result of the physiological signal, transform the continuous physiological signal into a microstate sequence, and extract the microstate sequence features.

[0032] The sentiment computing module is used to input the micro-state sequence features into a preset sentiment state prediction model, and generate sentiment computing results through the sentiment state prediction model.

[0033] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the emotion computing method based on wearable physiological signal microstates as described above.

[0034] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the emotion computing method based on wearable physiological signal microstates as described above.

[0035] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the emotion computing method based on wearable physiological signal microstates as described above.

[0036] This invention provides an emotion computing method and system based on wearable physiological signal microstates. By leveraging the temporal dynamic characteristics of emotional physiological responses and utilizing microstate clustering analysis methods that have been preliminarily validated in electroencephalogram (EEG) signals, an automated emotion state prediction model is established. Compared to traditional static wearable emotion computing, this invention considers the temporal process of emotional responses, potentially bringing new incremental improvements to the identification of emotion-specific physiological response patterns, thereby enhancing the accuracy and effectiveness of emotion recognition. Furthermore, the method based on wearable physiological measurements has lower user compliance, which helps promote the practical application of wearable emotion computing technology and brings new growth points to the advancement of the field of emotional intelligence. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0038] Figure 1 This is a flowchart illustrating an emotion computing method based on wearable physiological signal microstates provided by the present invention.

[0039] Figure 2 This is an architecture diagram of an emotion computing system based on wearable physiological signal microstates provided by the present invention;

[0040] Figure 3 This is a schematic diagram of the module connection of an emotion computing system based on wearable physiological signal microstates provided by the present invention;

[0041] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0042] Figure label:

[0043] 110: Signal acquisition module; 120: Microstate partitioning module; 130: Sequence feature extraction module; 140: Sentiment computing module.

[0044] 410: Processor; 420: Communication interface; 430: Memory; 440: Communication bus. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0046] The following is combined with Figure 1 - Describes an emotion computing method based on wearable physiological signal microstates according to the present invention, comprising:

[0047] S100: Acquire multi-channel physiological signals through wearable acquisition devices;

[0048] In this invention, acquiring effective physiological signals using wearable measurement devices is fundamental to subsequent microstate analysis. Since the latency and duration of common peripheral physiological responses typically range from several seconds to tens of seconds, the physiological measurement devices required by this method should have a sampling rate of at least 1 Hz. The measured signal categories include: pulse wave, conductance of skin, electrocardiogram, respiration, and skin temperature. Simultaneously, the signals from each channel should be time-aligned when summarizing. Regarding data acquisition, wearable devices meeting the measurement requirements are selected. Considering that wearable devices are more frequently used in daily measurements, they need to possess effective measurement capabilities for everyday scenarios. Specifically, the most important aspects are: signal quality and signal-to-noise ratio, anti-interference effect (e.g., resistance to motion interference), data transmission or storage method, sampling rate, and continuous measurement time.

[0049] S200: After extracting the features of multi-channel human physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is performed on the compressed features to obtain physiological feature categories and perform micro-state division of physiological signals.

[0050] Specifically, it includes:

[0051] Acquire multi-channel human physiological signals, segment them according to time windows, and extract standardized physiological features;

[0052] Based on standardized physiological characteristics, all physiological characteristics are summarized and dimensionality reduced. Principal component analysis is used to extract the feature dimensions with higher weights, and cluster analysis is performed on the dimensionality-reduced physiological characteristics.

[0053] Check the distribution of cluster analysis results, exclude clusters with too small a sample size or those concentrated on only a few individuals, and perform a stability check on the cluster results.

[0054] The effective feature categories are named as corresponding microstates, the microstate division is completed, and the physiological characteristics of the cluster center point mapped to the original feature space are marked.

[0055] In this invention, preliminary analysis of physiological signals to determine microstate categories forms the basis for subsequent feature extraction and emotion computation. Since the response speed of peripheral physiological responses is typically slower than that of EEG signals, changing on a timescale of seconds, the time window used to calculate peripheral physiological response features should be no less than 0.1 seconds to capture dynamic patterns of emotion-related physiological responses. After extracting features from multi-channel physiological signals, the data is dimensionality-reduced (e.g., using principal component analysis) to compress feature dimensions, and the resulting samples are clustered to obtain several major physiological feature categories, referred to as microstates. Microstates should meet basic constraints, such as the overall proportion of each microstate not being too low (e.g., <0.1%), and having a relatively uniform distribution across individuals. Microstates that do not meet these conditions are considered for merging with other microstates or recalculation.

[0056] In the micro-state segmentation stage, the preprocessed physiological signals are first segmented according to time windows, and their main features are extracted. The time window is no less than 0.1s. Common physiological features include, but are not limited to, time-domain features and frequency-domain features, such as statistical indicators like mean, standard deviation, skewness, and kurtosis, and frequency-domain attributes like peak frequency and harmonics. In addition, waveform characteristics of short-term physiological changes, as well as methods such as autoregressive coefficients and wavelet decomposition coefficients, can also be used to extract higher-order continuous signal features. Simultaneously, physiological signals themselves may have inherent individual differences, which are eliminated through intra-individual normalization. After extracting standardized physiological features, all samples are summarized and dimensionality reduced. Principal component analysis (PCA) is used to extract feature dimensions with higher weights. The explained variance of the dimensionality reduction results is no less than 95%, and the purpose of dimensionality reduction is to reduce the complexity of subsequent cluster analysis. Finally, k-means and cluster density peak methods are used to perform cluster analysis on the dimensionality-reduced physiological features. k-means can specify the number of cluster centroids, so by traversing different numbers of clusters, the effectiveness of the clustering results is evaluated to determine the final clustering result. Indicators for evaluating clustering effectiveness include, but are not limited to, silhouette coefficient and sum of squared clustering errors.

[0057] The distribution of clustering results is examined, excluding clusters with too small a sample size or those concentrated on only a few individuals. Since k-means clustering has a degree of randomness, the stability of the clustering results should be checked. Subsequently, the effective feature categories are named the corresponding microstates, and the physiological characteristics mapping their cluster centers to the original feature space are indicated.

[0058] S300. Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, the continuous physiological signals are transformed into microstate sequences, and the microstate sequence features are extracted.

[0059] Specifically, after completing the classification of microstate categories, the mapping method from the original physiological signal to the microstate category is obtained;

[0060] Based on the mapping method from the original physiological signal to the microstate category, long continuous physiological signals are transformed into microstate sequences, and the features of the microstate sequences are extracted.

[0061] In this invention, after classifying microstates, a mapping method from the original physiological signal to microstate categories can be obtained. Based on this, longer continuous physiological signals can be transformed into microstate sequences, and microstate sequence features can be extracted. These microstate sequence features include: the total proportion of each microstate, the average duration of each microstate's continuous occurrence, and the transition probability between pairs of microstates.

[0062] S400. Input the micro-state sequence features into a preset emotion state prediction model, and generate emotion calculation results through the emotion state prediction model.

[0063] Specifically, the micro-state sequence features are input into a pre-defined emotional state prediction model;

[0064] The emotion state prediction model maps micro-state sequence features to emotion labels and outputs emotion calculation results.

[0065] The emotional labels are derived from emotionally induced material labels, subjects' self-reports, and comprehensive behavioral performance.

[0066] In this invention, during practical emotion computing applications, wearable devices can be used to collect users' physiological responses under different emotional states. The micro-state features of the wearable physiological signals are then correlated with emotion tags obtained through subjective reports and other methods. Based on this correlation, an emotion state prediction model can be established. By optimizing the model parameters on the training set and testing the stability of the micro-state segmentation and the accuracy of the emotion computing model using an external test set, the robustness and effectiveness of the model can be ensured.

[0067] This method calculates the proportion, duration, and transition probability of each microstate, and inputs these microstate features into the emotion state prediction model to train the emotion recognition model. Emotional labels for physiological data can come from emotion-induced material labels, subject self-reports, and comprehensive behavioral performance. Considering the potential categorical attributes or continuous numerical ratings of the labels, the prediction model will selectively construct classification and regression models. The accuracy of the prediction model requires additional validation / testing. Test samples should be strictly independent of existing samples, ensuring they are not involved in cluster analysis, model training, or other steps. The model's performance is evaluated based on the accuracy of the test results, and then optimized.

[0068] This invention provides an emotion computing method based on wearable physiological signal microstates. Based on the temporal dynamic characteristics of emotional physiological responses, and utilizing a microstate clustering analysis method that has been preliminarily validated in electroencephalogram (EEG) signals, an automated emotion state prediction model is established. Compared to traditional static wearable emotion computing, this invention considers the temporal process of emotional responses, potentially bringing new incremental improvements to the identification of emotion-specific physiological response patterns, thereby enhancing the accuracy and effectiveness of emotion recognition. Furthermore, the wearable physiological measurement-based method has lower user compliance, which helps promote the practical application of wearable emotion computing technology and brings new growth points to the advancement of the field of emotional intelligence.

[0069] refer to Figure 2 and Figure 3The present invention also discloses an emotion computing system based on wearable physiological signal microstates, the system comprising:

[0070] The signal acquisition module 110 is used to acquire multi-channel physiological signals through a wearable acquisition device;

[0071] The micro-state partitioning module 120 is used to extract features of multi-channel human physiological signals, reduce the dimensionality, compress the feature dimension, and perform cluster analysis on the compressed features to obtain physiological feature categories, thereby performing micro-state partitioning of physiological signals.

[0072] The sequence feature extraction module 130 is used to map the original physiological signal to the microstate category based on the microstate division result of the physiological signal, transform the continuous physiological signal into a microstate sequence, and extract the microstate sequence features.

[0073] The emotion computing module 140 is used to input the micro-state sequence features into a preset emotion state prediction model and generate emotion computing results through the emotion state prediction model.

[0074] Among them, the signal acquisition module acquires multi-channel human peripheral physiological signals through wearable acquisition devices at a sampling rate greater than 1Hz;

[0075] The collected multi-channel human peripheral physiological signals are aggregated and then time-aligned.

[0076] The micro-state segmentation module acquires multi-channel human physiological signals, segments them according to time windows, and extracts standardized physiological features.

[0077] Based on standardized physiological characteristics, all physiological characteristics are summarized and dimensionality reduced. Principal component analysis is used to extract the feature dimensions with higher weights, and cluster analysis is performed on the dimensionality-reduced physiological characteristics.

[0078] Check the distribution of cluster analysis results, exclude clusters with too small a sample size or those concentrated on only a few individuals, and perform a stability check on the cluster results.

[0079] The effective feature categories are named as corresponding microstates, the microstate division is completed, and the physiological characteristics of the cluster center point mapped to the original feature space are marked.

[0080] The sequence feature extraction module, after completing the microstate category classification, obtains the mapping method from the original physiological signal to the microstate category;

[0081] Based on the mapping method from the original physiological signal to the microstate category, long continuous physiological signals are transformed into microstate sequences, and the features of the microstate sequences are extracted.

[0082] The emotion computing module inputs micro-state sequence features into a preset emotion state prediction model;

[0083] The emotion state prediction model maps micro-state sequence features to emotion labels and outputs emotion calculation results.

[0084] The emotional labels are derived from emotionally induced material labels, subjects' self-reports, and comprehensive behavioral performance.

[0085] This invention provides an emotion computing system based on wearable physiological signal microstates. Based on the temporal dynamic characteristics of emotional physiological responses, and utilizing a microstate clustering analysis method that has been preliminarily validated in electroencephalogram (EEG) signals, an automated emotion state prediction model is established. Compared to traditional static wearable emotion computing, this invention considers the temporal process of emotional responses, potentially bringing new incremental improvements to the identification of emotion-specific physiological response patterns, thereby enhancing the accuracy and effectiveness of emotion recognition. Furthermore, the wearable physiological measurement-based method has lower user compliance, which helps promote the practical application of wearable emotion computing technology and brings new growth points to the advancement of the field of emotional intelligence.

[0086] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440, wherein the processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute an emotion computing method based on wearable physiological signal microstates, the method including: acquiring multi-channel physiological signals through a wearable acquisition device;

[0087] After extracting features from multi-channel human physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories and perform micro-state division of physiological signals.

[0088] Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, the continuous physiological signals are transformed into microstate sequences, and the microstate sequence features are extracted.

[0089] The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results.

[0090] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0091] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being able to be stored on a non-transitory computer-readable storage medium, the computer program being executed by a processor, the computer being able to execute an emotion computing method based on wearable physiological signal microstates provided by the above methods, the method including: acquiring multi-channel physiological signals through a wearable acquisition device;

[0092] After extracting features from multi-channel human physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories and perform micro-state division of physiological signals.

[0093] Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, the continuous physiological signals are transformed into microstate sequences, and the microstate sequence features are extracted.

[0094] The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results.

[0095] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform an emotion computing method based on wearable physiological signal microstates provided by the above methods, the method comprising: acquiring multi-channel physiological signals through a wearable acquisition device;

[0096] After extracting features from multi-channel human physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories and perform micro-state division of physiological signals.

[0097] Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, the continuous physiological signals are transformed into microstate sequences, and the microstate sequence features are extracted.

[0098] The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An emotion computing method based on wearable physiological signal microstates, characterized in that, include: Multi-channel human peripheral physiological signals are acquired through wearable acquisition devices; After extracting features from multi-channel human peripheral physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories and perform micro-state classification of physiological signals. Based on the microstate segmentation results of the physiological signals, the original physiological signals are mapped to microstate categories, and the continuous physiological signals are transformed into microstate sequences. Microstate sequence features are extracted. Among them, the microstate sequence features include: the total proportion of each microstate, the average duration of each microstate, and the transition probability between pairs of microstates. The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results. The process of clustering the compressed features to obtain physiological feature categories and performing micro-state division of physiological signals specifically includes: checking the distribution of clustering analysis results, excluding clustering results with too few samples or concentrated on only a few individuals, and checking the stability of the clustering results; naming the effective feature categories as corresponding micro-states, completing the micro-state division, and indicating the physiological features mapped from the cluster center point to the original feature space. The process of mapping the original physiological signals to microstate categories based on the microstate classification results of the physiological signals, transforming continuous physiological signals into microstate sequences, and extracting features from the microstate sequences specifically includes: after completing the microstate category classification, obtaining the mapping method from the original physiological signals to the microstate categories; transforming longer continuous physiological signals into microstate sequences based on the mapping method from the original physiological signals to the microstate categories, and extracting features from the microstate sequences.

2. The emotion computing method based on wearable physiological signal microstates according to claim 1, characterized in that, The acquisition of multi-channel human peripheral physiological signals through wearable acquisition devices specifically includes: Multi-channel human peripheral physiological signals are acquired using wearable acquisition devices at a sampling rate greater than 1 Hz. The collected multi-channel human peripheral physiological signals are aggregated and then time-aligned.

3. The emotion computing method based on wearable physiological signal microstates according to claim 1, characterized in that, After extracting features from multi-channel human peripheral physiological signals, dimensionality reduction is performed to compress the feature dimensions. Then, cluster analysis is conducted on the compressed features to obtain physiological feature categories. Specifically, this includes: Acquire multi-channel human peripheral physiological signals, segment them according to time windows, and extract standardized physiological features; Based on standardized physiological characteristics, all physiological characteristics are summarized and dimensionality reduced. Principal component analysis is used to extract the feature dimensions with higher weights, and cluster analysis is performed on the dimensionality-reduced physiological characteristics.

4. The emotion computing method based on wearable physiological signal microstates according to claim 1, characterized in that, The micro-state sequence features are input into a preset emotion state prediction model, and the emotion state prediction model generates emotion calculation results, specifically including: Micro-state sequence features are input into a pre-defined emotion state prediction model; The emotion state prediction model maps micro-state sequence features to emotion labels and outputs emotion calculation results. The emotion labels are derived from emotion-inducing material labels, subjects' self-reports, and comprehensive behavioral patterns.

5. An emotion computing system based on wearable physiological signal microstates, characterized in that, The system includes: The signal acquisition module is used to acquire multi-channel human peripheral physiological signals through wearable acquisition devices; The micro-state partitioning module is used to extract features from multi-channel human peripheral physiological signals, reduce the dimensionality of the features, compress the feature dimensions, and perform cluster analysis on the compressed features to obtain physiological feature categories, thereby performing micro-state partitioning of physiological signals. The sequence feature extraction module is used to map the original physiological signal to the microstate category based on the microstate classification result of the physiological signal, transform the continuous physiological signal into a microstate sequence, and extract the microstate sequence features; wherein, the microstate sequence features include: the total proportion of each microstate, the average duration of each microstate, and the transition probability of switching between pairs of microstates. The emotion computing module is used to input the micro-state sequence features into a preset emotion state prediction model, and generate emotion computing results through the emotion state prediction model. The micro-state partitioning module is also used to: check the distribution of cluster analysis results, exclude cluster results with too few samples or concentrated on only a few individuals, and perform stability checks on the cluster results; name the effective feature categories as corresponding micro-states, complete the micro-state partitioning, and indicate the physiological characteristics of the cluster center points mapped to the original feature space. The sequence feature extraction module is further configured to: after completing the microstate category classification, obtain the mapping method from the original physiological signal to the microstate category; based on the mapping method from the original physiological signal to the microstate category, convert the long continuous physiological signal into a microstate sequence, and extract the features of the microstate sequence.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the emotion computing method based on wearable physiological signal microstates as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the emotion computing method based on wearable physiological signal microstates as described in any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the emotion computing method based on wearable physiological signal microstates as described in any one of claims 1 to 4.

Images