User psychological state analysis method and system based on intelligent robot

Abstract

The invention relates to the technical field of psychological analysis, in particular to a user psychological state analysis method and system based on an intelligent robot. The method comprises the following steps: carrying out dynamic interaction with a user based on an intelligent robot, collecting a real-time interaction video of the user, and obtaining physiological state parameters of the user through a sensor; performing image optimization on the user real-time interaction video, and constructing a filtering optimization interaction video; performing user dynamic micro-expression analysis on the filtering optimization interaction video to obtain user deep micro-expression state features; performing dynamic limb behavior feature mining on the filtering optimization interaction video to generate personalized limb behavior features of the user; performing time sequence fluctuation trend fitting on the physiological state parameters of the user so as to construct a physiological state fluctuation trend curve; and performing user emotional state analysis on the personalized limb behavior characteristics of the user based on the deep micro-expression state characteristics of the user to generate a user emotional evolution rule. According to the invention, efficient and accurate psychological state analysis is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of psychological analysis, and in particular to a user psychological state analysis method and system based on an intelligent robot. Background Art

[0002] With the rapid development of artificial intelligence and robotics, intelligent user analysis has become the key to improving user experience in various fields. In daily life and work scenarios, the user's emotional state is often an important factor affecting their behavior and decision-making. Therefore, being able to accurately capture and analyze the user's psychological changes and provide valuable basis for optimizing the user experience has become a hot research direction in the industry.

[0003] However, traditional user analysis methods mainly rely on subjective means such as questionnaires and interviews, or are limited to the analysis of single-modal data such as user expressions and voices, which makes it difficult to fully reflect the user's true emotional state. At the same time, these methods usually require manual intervention, are inefficient, and cannot meet the needs of fast-paced modern service scenarios. In addition, most existing emotion recognition technologies lack in-depth understanding and knowledge support of user psychological characteristics, and the analysis results are difficult to interpret and cannot provide effective support for decision-making. With the increasing maturity of intelligent robot technology, using robot platforms to conduct intelligent user psychological state analysis has become a new direction in the industry. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a user psychological state analysis method and system based on an intelligent robot to solve at least one of the above technical problems.

[0005] To achieve the above object, the present invention provides a user psychological state analysis method based on an intelligent robot, comprising the following steps: Step S1: Based on the dynamic interaction between the intelligent robot and the user, the real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; the image of the real-time interaction video of the user is optimized, and a filtering optimized interaction video is constructed; Step S2: Analyze the user's dynamic micro-expression on the filtered optimized interactive video to obtain the user's deep micro-expression state characteristics; Step S3: mining dynamic body behavior features of the filtered optimized interactive video to generate user personalized body behavior features; Step S4: fitting the time series fluctuation trend of the user's physiological state parameters to construct a physiological state fluctuation trend curve; Step S5: analyzing the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution law; Step S6: Based on the physiological state fluctuation trend curve, the personalized psychological state mining is carried out on the user's emotional evolution law to construct a personalized psychological state portrait.

[0006] The present invention dynamically interacts with users through intelligent robots, collects real-time interactive videos and user physiological state parameters, and provides a rich data source for subsequent analysis. By optimizing and processing interactive videos through images, the video quality can be improved, making subsequent analysis more accurate and reliable. Through the analysis of users' dynamic micro-expressions, the user's emotional state can be deeply understood, and the subtle details in the user's emotional changes can be explored, helping to better understand the user's emotional experience. Generating users' personalized body behavior characteristics can help understand users' behavioral habits and characteristics, and provide a more comprehensive information basis for emotional analysis. Constructing a physiological state fluctuation trend curve can reveal the law of changes in the user's physiological state and provide more in-depth background information for emotional analysis. Through the analysis of users' emotional states, the motivations and emotional change laws behind the user's emotions can be understood, providing a basis for personalized psychological state mining. Constructing personalized psychological state portraits can help to deeply understand users' psychological characteristics and needs, and provide a basis for providing personalized services and support.

[0007] Preferably, step S1 comprises the following steps: Step S11: Based on the dynamic interaction between the intelligent robot and the user, real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; Step S12: performing multi-region sampling on the user real-time interactive video to obtain multiple region sampling videos; Step S13: performing local pixel calculation on the plurality of region sampling videos to generate pixel values ​​for each region; Step S14: performing global average brightness calculation according to the pixel values ​​of each area to generate a global average brightness value; Step S15: performing brightness enhancement processing on the user real-time interactive video based on the global average brightness value, thereby obtaining a brightness enhanced interactive video; Step S16: performing edge depth filtering optimization on the brightness enhanced interactive video to construct a filtering optimized interactive video.

[0008] The present invention dynamically interacts with users through intelligent robots, collects real-time interactive videos and user physiological state parameters, and provides data support for subsequent psychological state analysis. Multi-region sampling of real-time interactive videos can more comprehensively capture the user's facial expressions and action information, providing more details for subsequent analysis. Generating pixel values ​​for each region by calculating local pixel points can help analyze the light and dark changes in different regions, providing a basis for subsequent brightness enhancement processing. Calculating the global average brightness based on the pixel values ​​of each region can objectively evaluate the overall brightness level of the video and provide a reference for subsequent brightness adjustment. By performing brightness enhancement processing on the video based on the global average brightness value, the brightness and clarity of the video screen can be improved, which helps to accurately capture user expressions and actions. Edge depth filtering optimization can further improve video quality, enhance edge features, make user expressions and actions clearer and more accurate, and provide a more reliable data basis for psychological state analysis.

[0009] Preferably, the specific steps of step S2 are: Step S21: positioning the facial region of the filtered optimized interactive video to obtain facial region coordinates; Step S22: performing facial region segmentation based on facial region coordinates to extract a real-time facial image of the user; Step S23: performing facial feature point recognition on the user's real-time facial image and marking the user's facial feature points; Step S24: Analyze the temporal changes of the facial feature points of the user to generate a temporal change trajectory of the facial feature points; Step S25: analyzing the user's dynamic micro-expression according to the temporal change trajectory of the facial feature points, thereby generating user's dynamic micro-expression data; Step S26: Perform deep state mining on the user's dynamic micro-expression data to obtain the user's deep micro-expression state features.

[0010] The present invention obtains facial region coordinates through facial region positioning, which is helpful to accurately locate the facial region of the user and provide a basis for subsequent facial feature extraction. Facial region segmentation based on facial region coordinates can effectively extract the facial image of the user and provide a clear input for subsequent facial feature point recognition. By identifying the facial feature points of the user's real-time facial image, the user's facial features can be accurately marked to provide detailed data for subsequent analysis. Time-series change analysis of the user's facial feature points can capture the changes of facial features over time and provide time-series information for subsequent micro-expression analysis. Micro-expression analysis based on the time-series change trajectory of facial feature points can deeply understand the dynamic characteristics of the user's micro-expressions and help reveal the user's emotional state and inner experience. Deep state mining generates the user's deep micro-expression state characteristics, which can more comprehensively understand the user's emotional state and psychological experience and provide a deeper information basis for psychological state analysis.

[0011] Preferably, step S3 specifically comprises the following steps: Step S31: Perform frame-by-frame dynamic analysis on the filtered optimized interactive video to identify the user's dynamic behavior data; Step S32: performing body movement analysis on the user's dynamic behavior data to extract multiple user body movements; Step S33: performing action frequency statistics on the body movements of multiple users to generate body movement frequencies; Step S34: Calculating the duration of the body movements of multiple users to obtain the duration of the body movements; Step S35: mining personalized behavior characteristics of the body movement frequency and the duration of the body movement to generate personalized body behavior characteristics of the user. The present invention identifies the user's dynamic behavior data through frame-by-frame dynamic analysis, so as to capture the user's subtle movement changes and provide detailed data support for subsequent limb movement analysis. By performing limb movement analysis on the user's dynamic behavior data, multiple user limb movements can be extracted to help understand the user's behavior pattern and movement characteristics. By statistically analyzing the frequencies of multiple user limb movements, the occurrence frequencies of different limb movements can be understood to provide a data basis for personalized behavior feature mining. By calculating the duration of multiple user limb movements, the time length information of the limb movements can be obtained to help analyze the continuity and duration of the user's behavior. By analyzing the frequency and duration of limb movements and performing personalized behavior feature mining, the user's personalized limb behavior characteristics can be generated, which helps to deeply understand the user's behavioral habits and behavioral characteristics and provide a more comprehensive information basis for psychological state analysis.

[0012] Preferably, the specific steps of step S4 are: Step S41: performing digital standardization processing on the user's physiological state parameters to generate standardized physiological state parameters; Step S42: extracting the user's real-time heart rate parameters and respiratory rate parameters based on the standardized physiological state parameters; Step S43: performing multi-time point fluctuation analysis on the user's real-time heart rate parameters and respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data; Step S44: Perform time series fluctuation trend fitting on the user's heart rate and respiratory rate fluctuation data over time, so as to construct a physiological state fluctuation trend curve.

[0013] The present invention can eliminate the dimensional influence between parameters by digitally standardizing the physiological state parameters of the user, making different parameters comparable and providing a consistent data basis for subsequent physiological parameter analysis. Extracting the user's real-time heart rate and respiratory rate parameters based on the standardized physiological state parameters can monitor the user's physiological state changes in real time and provide real-time physiological indicator data for psychological state analysis. Multi-time point fluctuation analysis of the user's real-time heart rate and respiratory rate parameters can understand the fluctuation of the user's physiological parameters at different time points and help discover potential physiological change trends. By fitting the time series fluctuation trend of the user's multi-time heart rate and respiratory rate fluctuation data, a fluctuation trend curve of the physiological state can be constructed, which helps to reveal the long-term change trend of the user's physiological state and provide more in-depth physiological data support for psychological state analysis.

[0014] Preferably, the specific steps of step S43 are: Define the sliding window length; Dividing the user's real-time heart rate parameter and respiratory rate parameter into multiple time windows according to the sliding window length, thereby obtaining multiple time window samples; Calculate the average value of multiple time window samples to obtain the average value of the heart rate parameter and respiratory rate parameter of each time window; A multi-time point fluctuation analysis is performed on the average values ​​of the heart rate parameters and the respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data.

[0015] The present invention helps to determine the number of data points contained in each time window by defining the sliding window length, which affects the granularity of subsequent time window division and the time span of sample data. By dividing the user's real-time heart rate and respiratory rate parameters into multiple time windows according to the defined sliding window length, continuous physiological parameter data can be divided into multiple independent samples, which is helpful for analyzing the changing trend of the user's physiological parameters. By calculating the average value of multiple time window samples, the average value of the heart rate and respiratory rate parameters in each time window can be obtained, and the average physiological state data of each time window is provided, which is convenient for subsequent fluctuation analysis. By performing multi-time point fluctuation analysis on the average value of the heart rate and respiratory rate parameters in each time window, the fluctuation of the user's heart rate and respiratory rate in different time windows can be revealed, which helps to understand the changing trend and fluctuation law of the user's physiological parameters, and provide more detailed data support for psychological state analysis.

[0016] Preferably, the specific steps of step S5 are: Step S51: reconstructing the user's deep micro-expression state features through facial animation to generate a deep micro-expression visualization animation; Step S52: Real-time matching of the user's personalized body behavior characteristics based on the deep micro-expression visualization animation to obtain micro-expression-body behavior matching data; Step S53: analyzing the user's emotional state based on the micro-expression-body behavior matching data, thereby generating real-time emotional state data of the user; Step S54: Identify the emotional mutation points of the user's real-time emotional state data and extract the real-time emotional mutation points; Step S55: mining the emotion evolution of the real-time emotion mutation points to generate the user emotion evolution rules. The present invention can convert the user's micro-expression characteristics into a visualized animation by reconstructing the user's deep micro-expression state characteristics through facial animation, thereby helping to understand the subtle changes in the user's emotional expression; based on the deep micro-expression visualized animation, instant matching of the user's personalized body behavior characteristics can establish the connection between micro-expressions and body behaviors, thereby helping to identify the user's emotional expression pattern; by parsing the micro-expression-body behavior matching data, the user's real-time emotional state data can be generated to help identify the user's current emotional state and provide a data basis for emotional analysis; by identifying emotional mutation points of the user's real-time emotional state data, the key points of emotional changes can be extracted to help capture the mutation moment of the user's emotional state and provide important clues for emotional evolution analysis; by performing emotional evolution mining on real-time emotional mutation points, the user's emotional evolution law can be generated to help understand the pattern and law of the user's emotional change and provide a deeper analysis for a deeper understanding of the user's psychological state.

[0017] Preferably, the specific steps of step S6 are: Step S61: performing physiological state fluctuation correlation analysis on the real-time emotion mutation point based on the physiological state fluctuation trend curve, thereby obtaining emotion mutation-physiological state fluctuation correlation data; Step S62: mining the user's emotional evolution rules based on the emotional mutation-physiological state fluctuation correlation data to generate the user's personalized psychological state data; Step S63: Predict the evolution of the user's personalized psychological state data and construct a personalized psychological state portrait.

[0018] The present invention helps to reveal the relationship between emotional mutation and physiological state fluctuation by performing physiological state fluctuation correlation analysis on real-time emotional mutation points based on the physiological state fluctuation trend curve, and helps to understand the physiological mechanism behind emotional changes. Personalized psychological state mining based on emotional mutation-physiological state fluctuation correlation data can deeply explore the user's psychological state change rules, help generate more personalized psychological state data, and provide support for personalized psychological feature analysis. Predicting the situation evolution of the user's personalized psychological state data can construct a personalized psychological state portrait of the user, help predict the trend and change of the user's future psychological state, and provide more accurate personalized services and emotional support for intelligent robots.

[0019] In this specification, a user psychological state analysis system based on an intelligent robot is provided, which is used to execute the user psychological state analysis method based on an intelligent robot as described above, including: The image optimization module is used to dynamically interact with users based on intelligent robots, collect real-time user interaction videos, and obtain user physiological state parameters through sensors; perform image optimization on real-time user interaction videos and construct filter-optimized interaction videos; The micro-expression analysis module is used to analyze the user's dynamic micro-expressions on the filtered and optimized interactive video to obtain the user's deep micro-expression state characteristics; The body behavior module is used to mine dynamic body behavior features of the filtered and optimized interactive video and generate user-personalized body behavior features; The physiological state fluctuation module is used to fit the time series fluctuation trend of the user's physiological state parameters, so as to construct a physiological state fluctuation trend curve; The emotional state analysis module is used to analyze the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution rules; The psychological state portrait module is used to conduct personalized psychological state mining on the user's emotional evolution rules based on the physiological state fluctuation trend curve and construct a personalized psychological state portrait.

[0020] The present invention can improve video quality, reduce noise and distortion, and provide clearer and more stable video input by performing image optimization and filtering processing on real-time user interactive videos, which is helpful to improve the accuracy and reliability of subsequent analysis steps; micro-expression analysis on filtered optimized interactive videos can obtain deep micro-expression state characteristics of users, help identify subtle emotional changes in user emotional expressions, and provide more detailed data support for emotional analysis; by performing body behavior analysis on filtered optimized interactive videos, personalized body behavior characteristics of users can be mined, which helps understand the action habits and personality characteristics of users, and provides more comprehensive information for emotional analysis and user interaction; time series fluctuation trend fitting of user physiological state parameters can construct physiological state fluctuation trend curves, which helps understand the changing laws of user physiological states, and provides physiological data support for emotional analysis; emotional state analysis based on deep micro-expression state characteristics and personalized body behavior characteristics of users can generate the laws of user emotional evolution, help understand the patterns and laws of user emotional changes, and provide support for personalized emotional analysis; personalized psychological state portraits can be constructed according to physiological state fluctuation trend curves and the laws of user emotional evolution, which can deeply mine the psychological state characteristics of users, help understand the psychological characteristics and personalized needs of users, and provide more accurate personalized services and emotional support for intelligent robots. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1A schematic diagram of the steps of a method for analyzing a user's psychological state based on an intelligent robot according to the present invention; Figure 2 Detailed implementation flow chart of step S1; Figure 3 Detailed implementation flow chart of step S2; Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0023] The example of this application provides a method and system. The execution subjects of the method and system include but are not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of this application, and the data processing platform includes but is not limited to: at least one of audio and image management system, information management system, and cloud data management system.

[0024] See also Figures 1 to 4 The present invention provides a user psychological state analysis method based on an intelligent robot, and the user psychological state analysis method based on an intelligent robot comprises the following steps: Step S1: Based on the dynamic interaction between the intelligent robot and the user, the real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; the image of the real-time interaction video of the user is optimized, and a filtering optimized interaction video is constructed; Step S2: Analyze the user's dynamic micro-expression on the filtered optimized interactive video to obtain the user's deep micro-expression state characteristics; Step S3: mining dynamic body behavior features of the filtered optimized interactive video to generate user personalized body behavior features; Step S4: fitting the time series fluctuation trend of the user's physiological state parameters to construct a physiological state fluctuation trend curve; Step S5: analyzing the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution law; Step S6: Based on the physiological state fluctuation trend curve, the personalized psychological state mining is carried out on the user's emotional evolution law to construct a personalized psychological state portrait.

[0025] The present invention dynamically interacts with users through intelligent robots, collects real-time interactive videos and user physiological state parameters, and provides a rich data source for subsequent analysis. By optimizing and processing interactive videos through images, the video quality can be improved, making subsequent analysis more accurate and reliable. Through the analysis of users' dynamic micro-expressions, the user's emotional state can be deeply understood, and the subtle details in the user's emotional changes can be explored, helping to better understand the user's emotional experience. Generating users' personalized body behavior characteristics can help understand users' behavioral habits and characteristics, and provide a more comprehensive information basis for emotional analysis. Constructing a physiological state fluctuation trend curve can reveal the law of changes in the user's physiological state and provide more in-depth background information for emotional analysis. Through the analysis of users' emotional states, the motivations and emotional change laws behind the user's emotions can be understood, providing a basis for personalized psychological state mining. Constructing personalized psychological state portraits can help to deeply understand users' psychological characteristics and needs, and provide a basis for providing personalized services and support.

[0026] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a user psychological state analysis method based on an intelligent robot of the present invention. In this example, the steps of the method include: Step S1: Based on the dynamic interaction between the intelligent robot and the user, the real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; the image of the real-time interaction video of the user is optimized, and a filtering optimized interaction video is constructed; In this embodiment, ensure that the interaction environment between the intelligent robot and the user is suitable, select a place with sufficient light and simple background to facilitate video acquisition and physiological status monitoring, configure the intelligent robot to have a video acquisition function, install a high-definition camera to ensure that the user's facial and body movements can be captured clearly, configure sensors (such as heart rate monitors, breathing sensors, etc.) to obtain the user's physiological status parameters in real time, start the intelligent robot, and start dynamic interaction with the user. During the interaction process, the camera continuously records the user's real-time video and monitors the user's physiological status in real time. The sensor continuously collects physiological parameters such as heart rate, breathing rate, and body temperature, and transmits the data to the processing system in real time. The collected physiological status parameters are stored in the database to ensure the integrity and continuity of the data. In order to be used for subsequent analysis, the collected real-time interactive video is optimized to improve the video quality and usability. The main steps include: De-noising: Use filtering algorithms (such as Gaussian filtering, median filtering, etc.) to remove background noise and image spots to improve image clarity. Apply image enhancement techniques (such as histogram equalization, contrast enhancement, etc.) to increase image brightness and contrast to make facial features more obvious. If there is jitter in the video, apply video stabilization algorithms (such as global motion compensation) to ensure the smoothness of the video stream. The optimized image sequence is synthesized into a filtered optimized interactive video to ensure that the video is smooth and can clearly show the user's dynamic behavior and facial expressions. Record the video metadata, including video length, frame rate, resolution and other information, for subsequent processing and analysis.

[0027] Step S2: Analyze the user's dynamic micro-expression on the filtered optimized interactive video to obtain the user's deep micro-expression state characteristics; In this embodiment, a suitable facial detection algorithm is selected, such as Haar Cascade, Dlib or MediaPipe, to accurately identify the facial area of ​​the user in the video. In each frame, the selected facial detection algorithm is applied to extract the key points of the user's face (such as the corners of the eyes, the corners of the mouth, the eyebrows, etc.). These key points will be used for subsequent micro-expression analysis. A model suitable for micro-expression analysis is selected, such as a convolutional neural network (CNN) based on deep learning or a feature point motion analysis model to ensure that subtle facial changes can be captured. The facial key point data is analyzed, and the slight motion changes of each key point in the video are calculated. Focus on the features of micro-expressions such as frowning and raising the corners of the mouth. The extracted facial feature point motion data is converted into time series data, and each facial The position of the feature points changes in each frame. If there is labeled data, a micro-expression classification model can be trained to identify different types of micro-expressions (such as happiness, sadness, surprise, etc.). If not, a ready-made pre-trained model can be used to input the extracted dynamic features into the micro-expression classification model to obtain the micro-expression state label of each frame, record the user's deep micro-expression state characteristics, summarize the micro-expression state of each frame, and calculate the user's micro-expression feature statistics in the entire video, such as the frequency and duration of each micro-expression, to generate the user's deep micro-expression state feature dataset, including the number of occurrences, intensity and duration of each micro-expression.

[0028] Step S3: mining dynamic body behavior features of the filtered optimized interactive video to generate user personalized body behavior features; In this embodiment, a suitable posture estimation algorithm is selected, such as OpenPose, MediaPipe or AlphaPose, to achieve real-time detection and tracking of the user's limbs. The selected posture estimation algorithm is applied in each frame to extract the user's key limb joints (such as head, shoulders, elbows, knees, etc.), record the coordinates of each key point, calculate the user's motion characteristics in the video based on the extracted key point data, pay attention to the dynamic changes of the limb's motion direction, speed, acceleration, etc., define a series of common limb behaviors (such as waving, nodding, sitting, standing, etc.), and select a suitable behavior recognition algorithm (such as CNN based on deep learning). Or traditional machine learning algorithms) for training or use pre-trained models to convert the extracted limb movement features into time series data, record the start time, end time and duration of each limb movement. If there is labeled data, a limb behavior classification model can be trained to identify the user's different limb behaviors. If there is no labeled data, an existing pre-trained model can be used to input the extracted dynamic limb features into the classification model to obtain the limb behavior state label of each frame, record the user's personalized limb behavior, summarize the limb behavior state of each frame, and calculate the user's personalized limb behavior characteristics in the entire video, such as the frequency, duration and intensity of each limb behavior, to generate the user's personalized limb behavior feature dataset, which includes statistical information, time series characteristics and user behavior habits of each limb behavior.

[0029] Step S4: fitting the time series fluctuation trend of the user's physiological state parameters to construct a physiological state fluctuation trend curve; In this embodiment, it is checked whether there are missing values ​​in the physiological state parameter data, and the missing values ​​are filled (such as using interpolation) or deleted. Smoothing techniques (such as moving average, Savitzky-Golay filter) are applied to reduce noise and make the data smoother to facilitate subsequent trend analysis. The selected fitting model is trained according to historical data to find the best fitting parameters. Trend prediction: the trained model is used to predict future physiological state trends, the correlation coefficient (R²) of the fitting results is calculated, and the model's ability to explain the data is evaluated. Residual analysis is used to check the fitting effect of the model to ensure that the residual conforms to the normal distribution. The original physiological state data is visually compared with the fitting trend curve, and a chart (such as a line chart or a scatter plot) is used to display the fitting effect. Based on the above analysis results, a physiological state fluctuation trend curve is generated. The curve should be able to reflect the trend of the user's physiological state changing over time. The parameters and model information of the trend curve are recorded for subsequent analysis and reference.

[0030] Step S5: analyzing the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution law; In this embodiment, it is ensured that the deep micro-expression features and the body behavior features are aligned in time. The features in the two data sets are merged through timestamps to form a comprehensive data set. Key features such as the intensity and duration of micro-expressions, and the type and frequency of body behaviors are selected from the deep micro-expression and body behavior data. If there is labeled data (such as the user's emotional state label), these data are used for model training. If not, a semi-supervised learning method can be used to divide the integrated feature data set into a training set and a validation set. The training set is used to train the model, and the model parameters are adjusted through the validation set. The performance indicators of the model (such as accuracy, F1-score) are evaluated. The user's deep micro-expression state features and personalized body behavior features are input into the trained emotional state model to perform emotional state analysis. The emotional state (such as happiness, sadness, anger, etc.) and its confidence level analyzed at each time point are recorded. According to the analyzed emotional state, the changes in the user's emotions over time are analyzed. The frequency, duration, transition rate, etc. of the emotional state can be calculated. Cluster analysis, pattern recognition and other methods are used to identify the laws and trends in the evolution of the user's emotions and form a description of the laws of the user's emotional evolution.

[0031] Step S6: Based on the physiological state fluctuation trend curve, the personalized psychological state mining is carried out on the user's emotional evolution law to construct a personalized psychological state portrait.

[0032] In this embodiment, data on the physiological state fluctuation trend curve and the user's emotional evolution law are collected and integrated. These data should contain timestamps for subsequent analysis and comparison to ensure that the physiological state fluctuation trend curve and the emotional evolution law are aligned in time. The timestamps are used to merge the data in the two data sets to form a comprehensive data set. Key features (such as fluctuation amplitude, fluctuation frequency, etc.) are extracted from the physiological state fluctuation trend and important emotional states (such as basic emotional types and their duration) are extracted from the emotional evolution law. If there is labeled data (such as expert evaluation of psychological state), these data are used for model training. If not, an unsupervised learning method can be used to divide the integrated feature data set into a training set and a test set, and the training set is used to train the model, and the test set is used to train the model. Evaluate model performance, input the user's physiological state fluctuation characteristics and emotional evolution characteristics into the trained model, conduct personalized psychological state mining, identify the user's current psychological state, and summarize the user's personalized psychological state characteristics based on the model output to form a user psychological state portrait, including the user's emotional characteristics, coping strategies, stress levels, etc. Visualize the personalized psychological state portrait, use charts (such as radar charts, heat maps, etc.) to show the user's performance in different psychological dimensions, intuitively reflect the user's psychological state, analyze the constructed psychological state portrait, identify key psychological characteristics and trends, understand the relationship between the user's emotional response and physiological state, and provide users with personalized emotion management suggestions or intervention measures based on the psychological state portrait to help users improve their psychological state.

[0033] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: Based on the dynamic interaction between the intelligent robot and the user, real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; Step S12: performing multi-region sampling on the user real-time interactive video to obtain multiple region sampling videos; Step S13: performing local pixel calculation on the plurality of region sampling videos to generate pixel values ​​for each region; Step S14: performing global average brightness calculation according to the pixel values ​​of each area to generate a global average brightness value; Step S15: performing brightness enhancement processing on the user real-time interactive video based on the global average brightness value, thereby obtaining a brightness enhanced interactive video; Step S16: performing edge depth filtering optimization on the brightness enhanced interactive video to construct a filtering optimized interactive video.

[0034] In this embodiment, ensure that the intelligent robot is equipped with a high-definition camera and sensor to perform video acquisition and physiological parameter monitoring (such as heart rate, body temperature, etc.), start the camera, start collecting the user's real-time interactive video, ensure the stability and clarity of the video stream, and at the same time, collect the user's physiological state parameters through the sensor, synchronize them with the video data, define multiple regions of interest (ROI) in the video, such as face, hand, background, etc., extract frames of the defined multiple regions from the real-time video, generate multiple regional sampling videos, ensure that the sampling frequency of each region is consistent, and facilitate subsequent analysis. Analyze the sampling video of each region frame by frame, extract the pixel point information of each region, calculate the brightness value of each pixel point in each region, and use the formula for converting RGB value to grayscale value to record the brightness of each region. The average pixel value of the domain is obtained for subsequent processing. The average pixel values ​​of all regions are added and divided by the number of regions to calculate the global average brightness value. The calculated global average brightness value is recorded to provide a basis for subsequent brightness enhancement processing. A suitable brightness enhancement algorithm is selected, such as contrast stretching, gamma correction, etc., to improve the visual effect of the video. The selected brightness enhancement algorithm is applied to the real-time interactive video to adjust the brightness and contrast to make the video content clearer and more visible. The processed brightness enhanced interactive video is saved as a new file to ensure that the original video data is not overwritten. A suitable edge depth filter is selected, such as Laplace filter or Canny edge detection, to enhance the edge details of the video. The selected filter is applied to the brightness enhanced interactive video to process each frame to improve edge clarity.

[0035] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: positioning the facial region of the filtered optimized interactive video to obtain facial region coordinates; Step S22: performing facial region segmentation based on facial region coordinates to extract a real-time facial image of the user; Step S23: performing facial feature point recognition on the user's real-time facial image and marking the user's facial feature points; Step S24: Analyze the temporal changes of the facial feature points of the user to generate a temporal change trajectory of the facial feature points; Step S25: analyzing the user's dynamic micro-expression according to the temporal change trajectory of the facial feature points, thereby generating user's dynamic micro-expression data; Step S26: Perform deep state mining on the user's dynamic micro-expression data to obtain the user's deep micro-expression state features.

[0036] In this embodiment, a suitable face detection algorithm is selected, such as a Haar cascade classifier, a deep learning model of Dlib or OpenCV (such as MTCNN), the face detection algorithm is applied to each frame of video, the face area is identified, the coordinates of the face area are obtained (usually the coordinates of the upper left corner and the lower right corner of the rectangular frame), the face image in each frame is cropped using the coordinates of the face area, and the cropped image is ensured to be clear and retain sufficient edge information. The extracted face image is preprocessed, such as resizing, normalizing and denoising, to improve the effect of subsequent analysis, and a suitable facial feature point detection algorithm is selected, such as Dlib Facial landmark detection or OpenFace is used to apply a feature point detection algorithm to each frame of the facial image to identify key facial feature points (such as corners of the eyes, corners of the mouth, tip of the nose, etc.), mark the identified feature points on the facial image, and record their coordinates for subsequent analysis. The coordinates of the feature points in each frame are recorded in chronological order to form time series data. Time series analysis methods (such as dynamic time warping) are used to analyze the change trajectory of the feature points, identify the change pattern of the feature points in the time dimension, and visualize the change trajectory of the feature points to more intuitively display the dynamic changes of the face. Appropriate deep data mining methods, such as cluster analysis, principal component analysis (PCA) or deep learning algorithms, are selected to extract deep features from dynamic micro-expression data, perform state analysis, identify potential emotional states and behavioral patterns, organize the analysis results into structured data, and record the user's deep micro-expression state features for subsequent use and analysis.

[0037] In this embodiment, refer to Figure 4 , is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: Perform frame-by-frame dynamic analysis on the filtered optimized interactive video to identify the user's dynamic behavior data; Step S32: performing body movement analysis on the user's dynamic behavior data to extract multiple user body movements; Step S33: performing action frequency statistics on the body movements of multiple users to generate body movement frequencies; Step S34: Calculating the duration of the body movements of multiple users to obtain the duration of the body movements; Step S35: mining personalized behavior characteristics of the body movement frequency and the duration of the body movement to generate personalized body behavior characteristics of the user. In this embodiment, the interactive video after filtering optimization is analyzed frame by frame. In each frame, an action recognition algorithm (such as an optical flow method or a deep learning model) is applied to identify the user's dynamic behavior. A pre-trained action recognition model (such as a model based on a convolutional neural network) is used to process each frame to extract the user's dynamic behavior features. The identified behavior data (such as walking, waving, sitting, etc.) is stored in a structured data set. The identified dynamic behavior data is analyzed, with special attention paid to the extraction of body movements. Posture estimation technology (such as OpenPose) is used to extract the user's dynamic behavior features. or MediaPipe) to extract the user's limb joints, identify multiple specific limb movements such as reaching, turning, and raising legs by analyzing the movement and position changes of the limb joints, count the extracted limb movements, calculate the frequency of occurrence of each limb movement in the entire video, use a counting algorithm to traverse the extracted action data, record the number of occurrences of each action, generate statistical results including each limb movement and its frequency, and form a limb movement frequency data set, calculate the duration of the extracted limb movements, record the time from the start to the end of each action, use timestamps or frame numbers to calculate the duration of each action to ensure accuracy, generate statistical results including each limb movement and its duration, and form a limb movement time length data set, combine the limb movement frequency and time length for personalized analysis, and identify the user's behavioral characteristics. For example, frequent waving may indicate the user's social inclination. Use cluster analysis or principal component analysis (PCA) and other technologies to deeply mine the data, identify the individual behavior patterns of users, and generate a personalized limb behavior feature data set for users, including behavioral characteristics such as frequency and time length, to form a personalized profile of users.

[0038] In this embodiment, step S4 includes the following steps: Step S41: performing digital standardization processing on the user's physiological state parameters to generate standardized physiological state parameters; Step S42: extracting the user's real-time heart rate parameters and respiratory rate parameters based on the standardized physiological state parameters; Step S43: performing multi-time point fluctuation analysis on the user's real-time heart rate parameters and respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data; Step S44: Perform time series fluctuation trend fitting on the user's heart rate and respiratory rate fluctuation data over time, so as to construct a physiological state fluctuation trend curve.

[0039] In this embodiment, the user's physiological state parameters are collected, including heart rate, respiratory rate, blood pressure, etc. Ensure that the data source is stable and the data quality is ensured. Select a suitable standardization method, usually using Min-Max normalization or Z-score standardization to eliminate the influence of different dimensions. Min-Max standardization: scale the data to the range of [0, 1]. Z-score standardization: convert the data into a distribution with a mean of 0 and a standard deviation of 1. Standardize each physiological state parameter to generate a standardized physiological state parameter. Extract the user's real-time heart rate and respiratory rate parameters from the standardized physiological state parameters. Ensure that the extraction process can reflect the user's physiological state in real time. Organize the extracted heart rate and respiratory rate parameters into a structured data format for subsequent analysis. Verify the accuracy of the extracted heart rate and respiratory rate data to ensure that it is consistent with the user's actual physiological state. Select a suitable fluctuation analysis method, such as using moving average, standard deviation or Fourier transform techniques to identify the fluctuation pattern of heart rate and respiratory rate. Perform multi-time point analysis on the user's heart rate and respiratory rate parameters to calculate the fluctuation value of each time period. Record the heart rate and respiratory rate fluctuation data at each time point for subsequent trend analysis. Store the fluctuation data in a structured database to form the user's heart rate and respiratory rate fluctuation data set. Select a suitable trend fitting algorithm, such as linear regression, spline interpolation, or polynomial fitting, to construct a physiological state fluctuation trend curve. Use the selected fitting algorithm to analyze the user's heart rate and respiratory rate fluctuation data and generate a trend curve. Ensure that the fitting results can effectively reflect the changes in the user's physiological state. Visualize the fitting results, such as drawing fluctuation trend curves of heart rate and respiratory rate, for subsequent analysis and understanding.

[0040] In this embodiment, the specific steps of step S43 are: Define the sliding window length; Dividing the user's real-time heart rate parameter and respiratory rate parameter into multiple time windows according to the sliding window length, thereby obtaining multiple time window samples; Calculate the average value of multiple time window samples to obtain the average value of the heart rate parameter and respiratory rate parameter of each time window; A multi-time point fluctuation analysis is performed on the average values ​​of the heart rate parameters and the respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data.

[0041] In this embodiment, the sliding window is a data processing technology used to create multiple overlapping samples in time series data. The window length determines the time span of each sample. According to actual needs and data characteristics, a suitable sliding window length is selected. For example, the window length can be set to 5 seconds, 10 seconds, etc., depending on the frequency of changes in heart rate and respiratory rate and the purpose of analysis. The selected window length is saved as a parameter for subsequent window division and analysis to ensure that the user's real-time heart rate and respiratory rate data have been obtained and sorted. The data should be a time series arranged in chronological order. According to the defined sliding window length, the real-time heart rate and respiratory rate parameters are divided into multiple time windows. The starting point and end point of each window are determined according to the window length. There can be overlap between windows. For example, if the window length is 10 seconds and the data sampling frequency is 1 Hz, each window contains 10 data points, and the windows can overlap by 5. seconds, store the divided time window samples in a structured data format for subsequent processing and analysis, calculate the average heart rate and respiratory rate for each divided time window, store the average heart rate and respiratory rate values ​​of each window in a new data structure to form an average data set of the time window, select an appropriate fluctuation analysis method, for example, you can use standard deviation, coefficient of variation or moving average and other techniques to evaluate the volatility of heart rate and respiratory rate, perform fluctuation analysis on the average heart rate and respiratory rate data of each newly generated time window, calculate the standard deviation of each time window, to evaluate the fluctuation of the user's physiological state in the time period, draw a fluctuation graph, intuitively display the fluctuation of heart rate and respiratory rate, record the results of the fluctuation analysis in the data set, including the fluctuation indicators of each time window (such as standard deviation, coefficient of variation, etc.).

[0042] In this embodiment, the specific steps of step S5 are: Step S51: reconstructing the user's deep micro-expression state features through facial animation to generate a deep micro-expression visualization animation; Step S52: Real-time matching of the user's personalized body behavior characteristics based on the deep micro-expression visualization animation to obtain micro-expression-body behavior matching data; Step S53: analyzing the user's emotional state based on the micro-expression-body behavior matching data, thereby generating real-time emotional state data of the user; Step S54: Identify the emotional mutation points of the user's real-time emotional state data and extract the real-time emotional mutation points; Step S55: mining the emotion evolution of the real-time emotion mutation points to generate the user emotion evolution rules. In this embodiment, deep micro-expression state features are extracted from the user's facial feature point data, and these features generally include subtle changes such as the corners of the mouth rising and the eyebrows raising. Select a suitable animation reconstruction algorithm, such as creating facial animation based on skeletal animation (rigging) or facial expression synthesis technology (such as BlendShapes). According to the extracted micro-expression features, the corresponding facial animation is generated using computer graphics technology. This can be achieved through animation software or custom scripts to ensure that the animation can truly reflect the changes in micro-expressions. The generated deep micro-expression visualization animation is stored as a video file or animation sequence for subsequent use and display. The limb behavior features related to the user's deep micro-expressions can be extracted, and posture estimation technology can be used to identify the user's limb movements, such as gestures, body tilt, etc. Select a suitable matching algorithm (such as dynamic time warping DTW or similarity measurement) to compare the similarity between micro-expression animation and limb behavior features. Match the generated micro-expression visualization animation with the extracted limb behavior features, record the limb behavior features corresponding to each micro-expression, and form a micro-expression-limb behavior matching data set. Determine a suitable emotion recognition model, which may be based on machine learning or deep learning methods, and combine micro-expression and limb behavior features for emotion classification. Using matching data as input, the user's emotional state is analyzed in real time to identify the user's current emotion (such as happiness, sadness, anger, etc.). The analyzed real-time emotional state of the user is stored in the form of structured data, including information such as emotion type and confidence. Select a suitable mutation point detection algorithm, such as CUSUM (Cumulative Sum Control Chart) or moving window method, to identify the key moments of emotional state change. Analyze the user's real-time emotional state data to identify emotional mutation points, that is, the moments when the emotional state changes significantly. Record the identified emotional mutation points to form time series data of emotional changes. Determine a suitable emotional evolution analysis method, such as time series analysis, cluster analysis, or state transition model, to study the change pattern of emotional state over time. Based on the mutation point data, analyze the change trajectory of the user's emotions and explore the patterns and laws of emotional changes, such as the fluctuation range and frequency of emotions. Record the excavated emotional evolution laws as the user's emotional state characteristics to form an analysis report on the user's emotional development trend.

[0043] In this embodiment, the specific steps of step S6 are: Step S61: performing physiological state fluctuation correlation analysis on the real-time emotion mutation point based on the physiological state fluctuation trend curve, thereby obtaining emotion mutation-physiological state fluctuation correlation data; Step S62: mining the user's emotional evolution rules based on the emotional mutation-physiological state fluctuation correlation data to generate the user's personalized psychological state data; Step S63: Predict the evolution of the user's personalized psychological state data and construct a personalized psychological state portrait.

[0044] In this embodiment, the physiological state fluctuation trend curve data and the real-time emotion mutation point data are collected and sorted. These data should have a timestamp to facilitate subsequent analysis. A suitable statistical analysis method is selected, such as correlation analysis (Pearson correlation coefficient or Spearman ranking correlation coefficient) or regression analysis, to evaluate the relationship between emotion mutation and physiological state fluctuation. For each emotion mutation point, the physiological state fluctuation data before and after are analyzed, and the average value, standard deviation and other indicators of the physiological state before and after the emotion mutation point are calculated. Association analysis is performed, and the analysis results are recorded to determine the degree of association between emotion mutation and physiological state fluctuation. A suitable personalized analysis method is selected, such as cluster analysis, principal component analysis (PCA) or decision tree, to identify the pattern of user emotion evolution. The emotion mutation-physiological state fluctuation association data is used to analyze the user's emotion change law. The user's emotion evolution law is combined with his physiological state fluctuation through clustering or classification methods to identify personalized psychological state characteristics. Generate personalized psychological state data of users, including emotion change patterns, key emotional states and their association with physiological states, form a description of the user's psychological state, select appropriate prediction models, such as time series prediction models (ARIMA), machine learning models (such as random forests or support vector machines), deep learning models (such as LSTM), etc., according to the characteristics and needs of the data, use personalized psychological state data of users as input, train the prediction model, and predict the user's future psychological state changes. The model should be able to learn the relationship between the user's emotions and physiological state fluctuations, visualize the prediction results, build a personalized psychological state portrait of the user, and show the user's emotional changes and physiological state in different time periods.

[0045] In this embodiment, a user psychological state analysis system based on an intelligent robot is provided, which is used to execute the user psychological state analysis method based on an intelligent robot as described above, including: The image optimization module is used to dynamically interact with users based on intelligent robots, collect real-time user interaction videos, and obtain user physiological state parameters through sensors; perform image optimization on real-time user interaction videos and construct filter-optimized interaction videos; The micro-expression analysis module is used to analyze the user's dynamic micro-expressions on the filtered and optimized interactive video to obtain the user's deep micro-expression state characteristics; The body behavior module is used to mine dynamic body behavior features of the filtered and optimized interactive video and generate user-personalized body behavior features; Physiological state fluctuation module, used to fit the time series fluctuation trend of the user's physiological state parameters, so as to construct a physiological state fluctuation trend curve; The emotional state analysis module is used to analyze the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution rules; The psychological state portrait module is used to conduct personalized psychological state mining on the user's emotional evolution rules based on the physiological state fluctuation trend curve and construct a personalized psychological state portrait.

[0046] The present invention can improve video quality, reduce noise and distortion, and provide clearer and more stable video input by performing image optimization and filtering processing on real-time user interactive videos, which is helpful to improve the accuracy and reliability of subsequent analysis steps; micro-expression analysis on filtered optimized interactive videos can obtain deep micro-expression state characteristics of users, help identify subtle emotional changes in user emotional expressions, and provide more detailed data support for emotional analysis; by performing body behavior analysis on filtered optimized interactive videos, personalized body behavior characteristics of users can be mined, which helps understand the action habits and personality characteristics of users, and provides more comprehensive information for emotional analysis and user interaction; time series fluctuation trend fitting of user physiological state parameters can construct physiological state fluctuation trend curves, which helps understand the changing laws of user physiological states, and provides physiological data support for emotional analysis; emotional state analysis based on deep micro-expression state characteristics and personalized body behavior characteristics of users can generate the laws of user emotional evolution, help understand the patterns and laws of user emotional changes, and provide support for personalized emotional analysis; personalized psychological state portraits can be constructed according to physiological state fluctuation trend curves and the laws of user emotional evolution, which can deeply mine the psychological state characteristics of users, help understand the psychological characteristics and personalized needs of users, and provide more accurate personalized services and emotional support for intelligent robots.

[0047] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0048] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A user psychological state analysis method based on an intelligent robot, characterized in that: The following steps are involved: Step S1: Based on the dynamic interaction between the intelligent robot and the user, real-time interaction video of the user is collected, and the user's physiological state parameters are obtained through sensors; Perform image optimization on real-time user interaction videos and construct filter-optimized interaction videos; Step S2: Analyze the user's dynamic micro-expression on the filtered optimized interactive video to obtain the user's deep micro-expression state characteristics; Step S3: mining dynamic body behavior features of the filtered optimized interactive video to generate user personalized body behavior features; Step S4: fitting the time series fluctuation trend of the user's physiological state parameters to construct a physiological state fluctuation trend curve; Step S5: analyzing the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution law; Step S6: Based on the physiological state fluctuation trend curve, the personalized psychological state mining is carried out on the user's emotional evolution law to construct a personalized psychological state portrait.

2. The user psychological state analysis method based on an intelligent robot according to claim 1 is characterized in that: The specific steps of step S1 are: Step S11: Based on the dynamic interaction between the intelligent robot and the user, real-time interaction video of the user is collected, and the physiological state parameters of the user are obtained through sensors; Step S12: performing multi-region sampling on the user real-time interactive video to obtain multiple region sampling videos; Step S13: performing local pixel calculation on the plurality of region sampling videos to generate pixel values ​​for each region; Step S14: performing global average brightness calculation according to the pixel values ​​of each area to generate a global average brightness value; Step S15: performing brightness enhancement processing on the user real-time interactive video based on the global average brightness value, thereby obtaining a brightness enhanced interactive video; Step S16: performing edge depth filtering optimization on the brightness enhanced interactive video to construct a filtering optimized interactive video.

3. The user psychological state analysis method based on an intelligent robot according to claim 1 is characterized in that: The specific steps of step S2 are: Step S21: positioning the facial region of the filtered optimized interactive video to obtain facial region coordinates; Step S22: performing facial region segmentation based on facial region coordinates to extract a real-time facial image of the user; Step S23: performing facial feature point recognition on the user's real-time facial image and marking the user's facial feature points; Step S24: Analyze the temporal changes of the facial feature points of the user to generate a temporal change trajectory of the facial feature points; Step S25: analyzing the user's dynamic micro-expression according to the temporal change trajectory of the facial feature points, thereby generating user's dynamic micro-expression data; Step S26: Perform deep state mining on the user's dynamic micro-expression data to obtain the user's deep micro-expression state features.

4. The user psychological state analysis method based on an intelligent robot according to claim 1, characterized in that: The specific steps of step S3 are: Step S31: Perform frame-by-frame dynamic analysis on the filtered optimized interactive video to identify the user's dynamic behavior data; Step S32: performing body movement analysis on the user's dynamic behavior data to extract multiple user body movements; Step S33: performing action frequency statistics on the body movements of multiple users to generate body movement frequencies; Step S34: Calculating the duration of the body movements of multiple users to obtain the duration of the body movements; Step S35: mining personalized behavior characteristics of the body movement frequency and the duration of the body movement to generate personalized body behavior characteristics of the user.

5. The user psychological state analysis method based on an intelligent robot according to claim 1, characterized in that: The specific steps of step S4 are: Step S41: performing digital standardization processing on the user's physiological state parameters to generate standardized physiological state parameters; Step S42: extracting the user's real-time heart rate parameters and respiratory rate parameters based on the standardized physiological state parameters; Step S43: performing multi-time point fluctuation analysis on the user's real-time heart rate parameters and respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data; Step S44: Perform time series fluctuation trend fitting on the user's heart rate and respiratory rate fluctuation data over time, so as to construct a physiological state fluctuation trend curve.

6. The user psychological state analysis method based on an intelligent robot according to claim 5 is characterized in that: The specific steps of step S43 are: Define the sliding window length; Dividing the user's real-time heart rate parameter and respiratory rate parameter into multiple time windows according to the sliding window length, thereby obtaining multiple time window samples; Calculate the average value of multiple time window samples to obtain the average value of the heart rate parameter and respiratory rate parameter of each time window; A multi-time point fluctuation analysis is performed on the average values ​​of the heart rate parameters and the respiratory rate parameters to obtain the user's multi-time heart rate and respiratory rate fluctuation data.

7. The user psychological state analysis method based on an intelligent robot according to claim 1, characterized in that: The specific steps of step S5 are: Step S51: reconstructing the user's deep micro-expression state features through facial animation to generate a deep micro-expression visualization animation; Step S52: Real-time matching of the user's personalized body behavior characteristics based on the deep micro-expression visualization animation to obtain micro-expression-body behavior matching data; Step S53: analyzing the user's emotional state based on the micro-expression-body behavior matching data, thereby generating real-time emotional state data of the user; Step S54: Identify the emotional mutation points of the user's real-time emotional state data and extract the real-time emotional mutation points; Step S55: mining the emotion evolution of the real-time emotion mutation points to generate the user emotion evolution rules.

8. The user psychological state analysis method based on an intelligent robot according to claim 1, characterized in that: The specific steps of step S6 are: Step S61: performing physiological state fluctuation correlation analysis on the real-time emotion mutation point based on the physiological state fluctuation trend curve, thereby obtaining emotion mutation-physiological state fluctuation correlation data; Step S62: mining the user's emotional evolution rules based on the emotional mutation-physiological state fluctuation correlation data to generate the user's personalized psychological state data; Step S63: Predict the evolution of the user's personalized psychological state data and construct a personalized psychological state portrait.

9. A user psychological state analysis system based on an intelligent robot, characterized in that: The method for analyzing the user's psychological state based on an intelligent robot as claimed in claim 1 comprises: The image optimization module is used to dynamically interact with users based on intelligent robots, collect real-time user interaction videos, and obtain user physiological state parameters through sensors; perform image optimization on real-time user interaction videos and construct filter-optimized interaction videos; The micro-expression analysis module is used to analyze the user's dynamic micro-expressions on the filtered and optimized interactive video to obtain the user's deep micro-expression state characteristics; The body behavior module is used to mine dynamic body behavior features of the filtered and optimized interactive video and generate user-personalized body behavior features; Physiological state fluctuation module, used to fit the time series fluctuation trend of the user's physiological state parameters, so as to construct a physiological state fluctuation trend curve; The emotional state analysis module is used to analyze the user's emotional state based on the user's deep micro-expression state characteristics and the user's personalized body behavior characteristics to generate the user's emotional evolution rules; The psychological state portrait module is used to conduct personalized psychological state mining on the user's emotional evolution rules based on the physiological state fluctuation trend curve and construct a personalized psychological state portrait.

Images