Psychological state and personality analysis system and method based on non-perception technology

Abstract

The invention provides a psychological state and personality analysis system and method based on a non-perception technology. Comprising a data acquisition module, a feature extraction module, a psychological state analysis module, a psychological portrait module and a behavior prediction module. The device integrates high-precision computer vision, deep learning, millimeter wave radar, an rPPG sensor, infrared thermal imaging and voice recording equipment into a whole, and high-precision real-time evaluation on the psychological state of an individual is achieved; a multi-modal fusion technology is adopted, so that the psychological state and personality characteristics can be analyzed with high precision without active cooperation of a user; edge calculation and deep learning are combined, rapid data processing is achieved, and the real-time performance of psychological state analysis is improved; the invention also relates to a method of the system. According to the system, personality characteristics, emotional states, defense mechanisms, psychological demands and psychological-driven feature fusion algorithms are combined, the scientificity of psychological state and personality analysis is ensured, and the reliability and applicability of evaluation are improved.

Description

Technical Field

[0001] This invention relates to the field of non-perceptual technology; and more particularly to a psychological state and personality analysis system and method based on non-perceptual technology. Background Technology

[0002] In recent years, with the development of artificial intelligence, computer vision, and psychological models, the analysis of mental states based on behavioral and physiological signals has gradually become a research hotspot. However, existing systems still have the following shortcomings:

[0003] (1) The information source is singular, mainly relying on a single modality (such as facial expression recognition), making it difficult to establish a complete psychological profile of a person.

[0004] (2) Existing methods rely too little on psychological models, resulting in a high false alarm rate and a lack of systematic psychological benchmarks.

[0005] (3) Computer vision still suffers from low accuracy and poor robustness in judging psychological states.

[0006] To address the above issues, this invention proposes a psychological state and personality analysis system based on non-perceptive technology. Without requiring active user cooperation, it utilizes multimodal behavioral analysis technology combined with psychological models to achieve accurate psychological state and personality analysis, providing support for security monitoring, school bullying early warning, recruitment screening, and personalized recommendations. Summary of the Invention

[0007] The purpose of this invention is to provide a system and method for analyzing psychological states and personality based on non-perceptual technology.

[0008] This invention is achieved through the following technical solution:

[0009] This invention relates to a psychological state and personality analysis system based on non-perceptual technology, comprising: a data acquisition module, a feature extraction module, a psychological state analysis module, a psychological profiling module, and a behavior prediction module;

[0010] in,

[0011] The data acquisition module is responsible for collecting multimodal data and transmitting it to the feature extraction module;

[0012] The feature extraction module is responsible for preprocessing the data, labeling features, and analyzing behavioral patterns to generate standardized psychological feature vectors and provide real-time psychological features to the psychological state analysis module. The extracted psychological features cover multi-dimensional information such as facial movements, micro-expressions, breathing rate, HRV index, and posture behavior patterns, providing data support for subsequent analysis.

[0013] The psychological state analysis module uses deep learning models (such as DNN and LSTM) and psychological modeling algorithms to identify an individual's current state, assess the individual's current psychological state, and provide data to the psychological profiling module to update the individual's long-term psychological state characteristics. The psychological state analysis module can identify psychological states such as anxiety, stress, depression, and aggression, and output corresponding identification labels and probability levels. This module provides an accurate assessment of an individual's real-time psychological state and forms the basis for behavioral judgment.

[0014] The psychological state analysis module strictly follows the established algorithm logic during the overall modeling process. It combines deep learning models with psychological modeling algorithms to achieve multi-level and systematic identification of psychological states and personality tendencies. The specific process is as follows:

[0015] First, the psychological state analysis module processes and identifies multimodal data of individuals using deep learning models. This includes a facial micro-expression recognition model based on visual information (identifying seven basic emotions: happiness, surprise, sadness, anger, fear, disgust, and contempt), an emotion mapping model based on physiological data (such as the nonlinear relationship between heart rate variability and emotional state), and a model for recognizing body movements and behavioral trajectories (collecting action responses and spatial trajectories through on-site experiments to establish psychological-behavioral pattern datasets and behavioral spatiotemporal trajectory datasets). Based on this, the psychological modeling algorithm analyzes psychological states and personality according to the mapping relationship between observable behavior, physiological data, and psychological traits, focusing on five core dimensions: personality traits, emotional state, defense mechanisms, psychological needs, and psychological driving characteristics. During the modeling process, emotion recognition results are combined with behavioral tendencies, and inferences are made using contextual information (such as social and work settings). For example, "disgust + avoidance behavior" in a social setting can be inferred as social phobia, while in a non-social setting it may point to resistance and destructive behavior towards tasks or the environment.

[0016] Personality trait analysis is based on internationally recognized models such as DSM-5, the Big Five personality traits, and Eysenck's personality theory. It identifies characteristics through the coupling of behavior and emotion; for example, individuals with antisocial personality tendencies exhibit strong anger and aggressive behavior during conflict. The core logic of the model lies in comprehensively judging psychological states and personality tendencies by considering the frequency, context, and time span of psychological behaviors. It also uses thresholds to quantify the explanatory power of personality traits; for instance, if a certain behavioral characteristic appears in 40% of the overall test and lasts for more than six months, its explanatory power for personality tendencies can reach 80%. Ultimately, it achieves systematic modeling and dynamic reasoning of an individual's short-term psychological state and long-term personality tendencies.

[0017] The psychological profiling module is responsible for storing and updating the baseline of an individual's psychological state, providing personalized data support for the behavior prediction module;

[0018] The behavior prediction module combines psychological state analysis and individual profiling to predict future changes in an individual's psychological state and provide risk warnings or personalized suggestions.

[0019] Preferably, the data acquisition module is used to acquire raw data of facial expressions, postures, behaviors, and physiological signals; thereby enabling the establishment of a multi-dimensional psychological state analysis system.

[0020] The feature extraction module preprocesses, labels, and analyzes the collected data; this module includes physiological signal analysis and behavioral pattern recognition; it employs a multimodal data fusion algorithm, combined with deep learning (CNN, LSTM) and Bayesian networks for data analysis; and it proposes a temporal feature modeling method to extract trends in psychological state changes and improve data prediction capabilities.

[0021] The psychological state analysis module analyzes an individual's current psychological state characteristics in real time based on extracted features, including stress and mood fluctuations.

[0022] The Psychological Profiling Module receives long-term time-series data from the Psychological State Analysis Module, models the individual's behavior and emotional trajectory, and constructs a personalized psychological profile. This module comprehensively uses clustering algorithms and psychological standards, such as the Big Five personality model and defense mechanism modeling, to generate user-specific psychological tags, emotional baselines, and behavioral preferences. The profile results also support the improvement of the accuracy of psychological state analysis and behavior prediction.

[0023] in,

[0024] Behavioral trajectory modeling comprises two parts: first, limb movement modeling, which guides individuals to complete specific actions according to designated instructions through on-site experiments, constructing a psychological-behavioral pattern dataset; second, action trajectory modeling, which combines behavioral movement and spatial location information to establish a spatiotemporal trajectory dataset of behavior, reflecting the dynamic behavioral patterns of individuals in different situations. Emotional fluctuation trajectory modeling focuses on the identification of seven basic emotions (happiness, surprise, sadness, anger, fear, disgust, and contempt), relying on psychologically recognized facial micro-expression recognition technology for accurate capture, while integrating physiological data (such as heart rate and respiratory rate) to improve the accuracy and stability of emotional state recognition.

[0025] The modeling process involves synchronizing and serializing raw behavioral labels (such as aggression, avoidance, and concealment) from multimodal data sources with emotion recognition results (such as happiness, anger, and sadness) during the data organization and preprocessing stages, forming behavioral and emotional time trajectory datasets on a daily, weekly, and monthly basis. Secondly, time-series modeling algorithms are used to model the evolutionary trends of behavior and emotion. Combining behavioral and emotional trajectories, a joint psychological trajectory model is constructed to analyze the coupling patterns between behavior and emotion, such as whether a certain emotional state (e.g., aversion) frequently occurs before a specific behavior (e.g., avoidance), forming a stable psychological response pattern. For example: Negative emotion (primarily aversion) + suspicious behavior (concealment, avoidance) = social phobia or destructive behavior. Explanation: Suspicious behavior occurring in social situations manifests as social phobia: aversion to social interaction, avoidance of contact, and exhibiting aloof behavior (e.g., refusing eye contact). Suspicious behavior occurring in work or non-social situations manifests as aversion to work or the environment, requiring attention to whether the work motive is impure. Based on this model, key psychological profile feature indicators can be extracted, such as: emotional baseline, dominant emotion, emotional fluctuation index, normal behavioral trajectory, and behavior-emotion consistency index.

[0026] The psychological profiling module performs long-term modeling of psychological state characteristics to form a personalized profile and provides a historical psychological state baseline.

[0027] The behavior prediction module, based on an individual's current state and profile information, predicts the individual's future psychological trends and behavioral patterns; it uses a Transformer neural network and a psychological decision tree to predict the individual's short-term and long-term psychological change trends.

[0028] Preferably, the data acquisition module integrates an RGB camera, a surveillance camera, a millimeter-wave radar, an rPPG sensor, an infrared thermal imaging camera, and a voice recording device into one device; this device has the following advantages:

[0029] (1) The device can simultaneously collect visible light, infrared thermal imaging and millimeter wave signals, and perform data fusion to improve the accuracy of psychological state recognition and realize non-contact and imperceptible psychological state assessment.

[0030] (2) The device has an embedded psychological analysis algorithm and has independent computing and data transmission capabilities;

[0031] (3) Edge computing technology is adopted to enable the camera to have local preprocessing capabilities, reduce data transmission burden, reduce data transmission and computing latency, and improve real-time performance; adaptive filtering + neural network pruning technology is adopted to optimize model calculation efficiency and improve the low power consumption operation capability of the device.

[0032] (4) The device uploads the processed data to the central analysis system via wireless network or wired connection, supporting remote monitoring, data storage and in-depth analysis; or performs data preprocessing at the smart camera end to reduce bandwidth usage and improve system real-time performance.

[0033] (5) The device has an adaptive learning algorithm that can dynamically adjust the data processing strategy to adapt to different scenario requirements, such as intelligent security, mental health assessment, behavior prediction, personalized interaction and other application scenarios.

[0034] RGB camera: Captures facial expressions and micro-expression features;

[0035] Surveillance cameras: acquire data on human posture, gestures, and behavior;

[0036] Millimeter-wave radar: non-contact acquisition of users' heart rate and respiratory rate, enhancing physiological signal analysis capabilities;

[0037] rPPG sensor: detects changes in facial hemodynamics, calculates heart rate, and infers heart rate variability (HRV) for stress and mood fluctuation assessment;

[0038] Infrared thermal imaging camera: monitors temperature changes in different parts of the body, detects blood circulation patterns, and assists in psychological stress assessment.

[0039] Preferably, the method for calculating heart rate using rPPG is as follows:

[0040] (1) Extract the ROI (forehead and cheek area);

[0041] (2) Apply color space conversion to the RGB image signal (from RGB to CHROM channel);

[0042] (3) Noise is filtered out by bandpass filtering (0.7–4Hz);

[0043] (4) Apply FFT analysis to find the peak of the main frequency;

[0044] (5) Formula for calculating heart rate (HR):

[0045] HR = f peak ×60 (unit: times / minute)

[0046] Among them, f peak It is the peak value of the main frequency.

[0047] Preferably, the feature extraction module includes physiological signal analysis and behavioral pattern recognition;

[0048] Among them, behavior pattern recognition is face and expression recognition, which uses deep learning algorithms to analyze facial expressions and micro-expressions;

[0049] The physiological signal analysis combines millimeter-wave radar and rPPG technology to detect individual physiological status in real time.

[0050] Physiological signal analysis includes: posture and movement analysis, skin temperature and hemodynamic monitoring and analysis, multimodal fusion analysis, and combined speech analysis;

[0051] Preferably, the posture and motion analysis specifically includes:

[0052] (1) The skeleton point detection algorithm tracks human joints in real time and evaluates behavioral patterns;

[0053] (2) Motion vector analysis identifies abnormal postures, stressful behaviors, and potential aggressive actions;

[0054] (3) Based on labeled video training, optimize the action recognition accuracy to ensure that the algorithm can accurately distinguish between normal and abnormal behavior;

[0055] The skin temperature and hemodynamic monitoring and analysis: Infrared thermal imaging was used to analyze changes in psychological stress;

[0056] The multimodal fusion analysis integrates physiological signals and behavioral characteristics to improve the accuracy of psychological state assessment.

[0057] The combination of speech analysis: by analyzing pitch, speech rate, and intonation fluctuations, it helps to judge psychological state and improves the accuracy of psychological state assessment.

[0058] Taking HRV (heart rate variability) as an example:

[0059] Use two independent signal streams: rPPG(t) and Rader(t).

[0060] Use a sliding window for timing alignment and interpolation matching to ensure consistency at different time points:

[0061] rPPG(t i ),Radar(t i for i = 1, ..., N

[0062] Introducing the cross-correlation function:

[0063]

[0064] Determine and correct the time delay between the two signals;

[0065] Adaptive confidence fusion algorithm: A modal confidence evaluation model based on signal-to-noise ratio (SNR), motion state, and illumination conditions, as shown in the formula below:

[0066] HR = w1·HRrPPG + w2·HRRadar

[0067] Where: w1 + w2 = 1

[0068] Anomaly detection and compensation mechanism: When a channel detects a decrease in the quality of physiological signals (such as during periods without RR), another channel can serve as a compensation source;

[0069] HRV calculation optimization: Hybrid sampling enhances stability, as shown in the following formula:

[0070] RR fused (t)=α·RR rPPG (t)+(1-α)·RR Radar (t).

[0071] Preferably, the working principle of the psychological profiling module includes:

[0072] (1) Establish an individual psychological state feature vector by fusing behavioral and physiological data;

[0073] (2) Data-driven feature clustering algorithm is used to classify the psychological states of different individuals;

[0074] (3) Combine time series modeling to predict future behavioral trends of individuals and provide personalized intervention suggestions;

[0075] (4) The results of user psychological state analysis are displayed through a visualization system, which is applicable to security, mental health management and personalized service recommendations.

[0076] Among them, the psychological profile of the group:

[0077] (a) Analyze the overall psychological state by examining the behavioral patterns, emotional trends, and interaction methods of individual group members;

[0078] (b) Use social psychology theories to assess group dynamics (such as emotional synchronization and collective pressure).

[0079] Preferably, the behavior prediction module employs behavior prediction based on a deep learning model to analyze future trends in psychological state changes; it combines Bayesian networks and reinforcement learning algorithms to optimize the short-term and long-term prediction capabilities of psychological states; application scenarios include: security monitoring, intelligent human-computer interaction, mental health management, and personalized recommendations.

[0080] The psychological state analysis module involved in this invention has the following functions:

[0081] (1) Establish psychological baseline: Combine individual historical data to construct a long-term psychological state baseline to avoid false alarms;

[0082] (2) Enhanced analytical accuracy: Psychological models provide stable feature extraction standards, improving the reliability of AI reasoning;

[0083] (3) Constructing a baseline of an individual's long-term psychological state: through time series data analysis.

[0084] (4) Achieve multi-dimensional personality profile: adopt a psychology-driven feature fusion model, including personality traits, emotional state, defense mechanisms, psychological needs, etc.

[0085] (5) False alarms are eliminated: The anomaly detection algorithm is optimized by utilizing psychological defense mechanisms, the Big Five personality traits (OCEAN), and psychological state baselines to reduce false alarms.

[0086] (6) Improve the accuracy of predicting the trend of psychological state changes: adopt a psychological state prediction model based on multilayer neural network (DNN) and long short-term memory network (LSTM).

[0087] (7) Improve the accuracy of personalized behavior trend prediction: optimize the psychological state prediction algorithm through reinforcement learning.

[0088] This invention also relates to a method for the aforementioned psychological state and personality analysis system based on non-perceptual technology, comprising the following steps:

[0089] Step 1: The data acquisition module collects multimodal data and transmits it to the feature extraction module;

[0090] Step 2: The feature extraction module preprocesses the data, performs feature annotation and behavioral pattern analysis, generates standardized psychological feature vectors, and provides real-time psychological features to the psychological state analysis module.

[0091] Step 3: The psychological state analysis module uses deep learning models and psychological modeling algorithms to determine the individual's current state, assess the individual's current psychological state, and provide data to the psychological profile module to update the individual's long-term psychological state characteristics.

[0092] Step 4: The psychological profiling module stores and updates the baseline of an individual's psychological state, providing personalized data support for the behavior prediction module.

[0093] Step 5: The behavior prediction module combines psychological state analysis and individual profiles to predict future changes in an individual's psychological state and provides risk warnings or personalized suggestions.

[0094] The present invention has the following advantages:

[0095] (1) The system involved in this invention integrates high-precision computer vision, deep learning, millimeter-wave radar, rPPG technology and infrared thermal imaging technology to achieve high-precision real-time assessment of individual psychological state.

[0096] (2) Non-perceptible detection: Using multimodal fusion technology, psychological state and personality characteristics can be analyzed with high precision without the user's active cooperation.

[0097] (3) High accuracy: It integrates multiple sensors such as high-definition RGB camera, millimeter-wave radar, rPPG sensor, and infrared thermal imaging to improve recognition accuracy.

[0098] (4) Strong real-time performance: Combining edge computing and deep learning enables rapid data processing and improves the real-time performance of psychological state analysis.

[0099] (5) Multi-dimensional integration: It not only detects facial expressions, but also combines physiological signals, posture, voice and other factors to achieve a more comprehensive psychological assessment.

[0100] (6) Automated analysis: Based on deep learning models, reduce manual intervention and improve the automation level of the system.

[0101] (7) Professional psychological model support: This system combines personality traits, emotional states, defense mechanisms, psychological needs, and psychologically driven feature fusion algorithms to ensure the scientific nature of psychological state and personality analysis, and improve the reliability and applicability of the assessment.

[0102] (8) Personalized prediction: Using time series modeling and reinforcement learning techniques, it is possible to predict an individual's future psychological trends.

[0103] (9) Wide range of applications: It can be used in multiple fields such as security monitoring, mental health management, recruitment screening, and behavior prediction.

[0104] (10) Data security and privacy protection: Reduce the transmission of sensitive data and improve privacy protection capabilities through local edge computing. Attached Figure Description

[0105] Figure 1 This is a schematic diagram of the framework of the psychological state and personality analysis system based on non-perceptual technology involved in this invention;

[0106] Figure 2 This is a flowchart of the psychological state and personality analysis system based on non-perceptual technology involved in this invention;

[0107] Figure 3 This is a schematic diagram of the hardware structure of the psychological state and personality analysis system based on non-perceptual technology involved in this invention. Detailed Implementation

[0108] The present invention will now be described in detail with reference to specific embodiments. It should be noted that the following embodiments are merely further illustrations of the present invention, but the scope of protection of the present invention is not limited to the following embodiments.

[0109] Example 1

[0110] This embodiment relates to a psychological state and personality analysis system based on non-perceptual technology. See Figure 1 As shown, it includes: a data acquisition module, a feature extraction module, a psychological state analysis module, a psychological profiling module, and a behavior prediction module;

[0111] Among them, the data acquisition module, feature extraction module, psychological state analysis module, psychological profiling module, and behavior prediction module form a closed loop through data flow and algorithm-driven processes, realizing the whole process analysis from the acquisition of raw behavioral signals to the prediction of psychological trends.

[0112] The system's entry point is the data acquisition module, which integrates multiple non-contact sensor devices, including RGB cameras, surveillance cameras, millimeter-wave radar, rPPG sensors, and infrared thermal imaging cameras, to achieve simultaneous acquisition of multimodal information such as facial expressions, postures, physiological signals, and temperature distribution. This module provides the system with basic sensing capabilities and features non-contact, automated acquisition, and high coverage.

[0113] The collected raw data is transmitted to the feature extraction module, which performs preprocessing, feature annotation, and behavioral pattern analysis on different modal data, including signal cleaning, temporal alignment, and feature recognition, and finally generates standardized psychological feature vectors. The extracted features cover multi-dimensional information such as facial movements, micro-expressions, respiratory rate, HRV index, and posture behavior patterns, providing data support for subsequent analysis.

[0114] After obtaining the psychological feature vector, the psychological state analysis module uses deep learning models (such as DNN and LSTM) and psychological modeling algorithms to determine the individual's current state. The model can identify psychological states such as anxiety, stress, depression, and aggression, and output the corresponding identification labels and probability levels. This module provides an accurate assessment of the individual's real-time psychological state and forms the basis for behavioral judgment.

[0115] The individual psychological profiling module receives long-term time-series data from the psychological state analysis module, models the individual's behavior and emotional trajectory, and constructs a personalized psychological profile. This module comprehensively uses clustering algorithms and psychological standards, such as the Big Five personality model and defense mechanism modeling, to generate user-specific psychological tags, emotional baselines, and behavioral preferences. This profiling result also supports improvements in the accuracy of psychological state analysis and behavior prediction.

[0116] Based on the integration of an individual's current state and profile information, the behavior prediction module uses methods such as temporal modeling and reinforcement learning to predict the user's possible behavioral trends and psychological change paths. The system can determine whether an individual is about to enter a high-risk state, whether there is an aggressive development trend, or whether intervention suggestions are needed, providing a preliminary judgment basis for scenarios such as security, mental health management, and education.

[0117] The five modules mentioned above have clear data channels and functional collaboration, constructing a full-link psychological analysis system from bottom-level perception to top-level decision-making, realizing highly real-time, seamless, and intelligent psychological state recognition and prediction functions.

[0118] The hardware of the psychological state and personality analysis system based on non-perceptual technology involved in the embodiments of this invention is described in [link to hardware description]. Figure 3 As shown,

[0119] This device is used to achieve real-time assessment of an individual's psychological state and prediction of behavioral trends under non-contact and non-cooperative conditions. The overall structure includes: the device body and multiple integrated functional modules, specifically including a data acquisition module, a data processing module, and a data storage and transmission module. The modules are connected and work together through the device's internal communication bus or high-speed interface to form a localized intelligent psychological analysis system.

[0120] The device itself serves as the carrier structure for the intelligent terminal, embedding various functional sub-modules and achieving integrated functionality through a unified power supply, heat dissipation, and packaging system. This device supports deployment in multiple forms, including wall-mounted, ceiling-mounted, and mobile terminal configurations. It is waterproof, dustproof, and resistant to electromagnetic interference, making it suitable for various environments such as parks, security systems, campuses, and enterprises.

[0121] The device's data acquisition module collects multimodal raw information and includes five non-contact sensors and a voice analysis component: an RGB camera, an infrared thermal imaging device, a millimeter-wave radar, an rPPG sensor, and a voice analysis module. The RGB camera acquires individual facial images and posture information; the infrared thermal imaging device records changes in body surface temperature distribution to aid in psychological stress recognition; the millimeter-wave radar non-contactly collects respiratory rate and heart rate signals; the rPPG sensor estimates HRV (Heart Rate of Change) by analyzing blood flow changes in facial images; and the voice analysis module analyzes an individual's emotional vocal characteristics based on factors such as tone, pitch, and speech rate. These sensors work collaboratively to ensure data synchronization and multi-angle coverage, thus providing high-dimensional input for psychological state modeling.

[0122] The multimodal signals acquired by the data acquisition module are directly transmitted to the data processing module. The data processing module possesses local intelligent analysis capabilities and consists of an edge computing chip, a deep learning inference engine, and a psychological state analysis algorithm unit. The edge computing chip performs preprocessing operations such as filtering, temporal alignment, and feature extraction on the input data, reducing redundancy and improving processing efficiency. The deep learning inference engine deploys a lightweight neural network model to fuse and recognize facial expressions, postures, and physiological signals, outputting structured psychological features. The psychological state analysis algorithm module combines these features, making judgments based on psychological theories and machine learning models, and outputs real-time psychological state results, such as emotion category (pleasure, tension, anger, etc.), state level (low risk, medium risk, high risk), and behavioral characteristics (avoidance, confrontation, aggression, etc.).

[0123] After data processing, the relevant results are stored locally and transmitted remotely through the data storage and transmission module. This module consists of three parts: a wireless data transmission unit, a local data caching unit, and a central analysis server interface. The wireless module supports multiple communication methods such as 4G / 5G and Wi-Fi, pushing structured result data to the cloud platform or remote monitoring center in real time; the local caching unit is used for short-term data storage, supporting offline storage, automatic overwrite, and breakpoint resume mechanisms; the central analysis interface can be linked with the backend platform to achieve long-term updates of individual profiles, identification of group psychological trends, and platform-level linkage control.

[0124] The various modules of this intelligent monitoring device work together to form a complete closed loop from data acquisition, edge analysis, structured output to remote communication. It features high real-time performance, high accuracy, strong adaptability, and convenient deployment, and is suitable for psychological state monitoring, behavioral risk warning, and personalized human-computer interaction systems in various scenarios.

[0125] The process of psychological state analysis in this invention is described in [link to invention]. Figure 2 As shown:

[0126] It is divided into six consecutive processing nodes, namely data acquisition, data preprocessing, feature extraction, multimodal fusion, psychological state modeling and recognition result output. The modules are connected in sequence according to the information processing logic, forming the core link of the system's psychological recognition engine.

[0127] The system first acquires raw data such as facial images, body posture, physiological signals, and thermal characteristics using non-intrusive acquisition devices deployed on-site (e.g., RGB cameras, millimeter-wave radar, rPPG modules, and infrared thermal imaging). This raw data contains noise and exhibits asynchronous behavior, requiring it to be uniformly processed in the data preprocessing stage.

[0128] During the data preprocessing stage, the system uses filtering algorithms, motion correction, time synchronization, and spatial normalization techniques to adjust the consistency of multi-source data. The processed data retains the original fluctuation characteristics while possessing a unified temporal structure, meeting the input requirements for subsequent modeling.

[0129] The processed data is then input into the feature extraction module, where the system extracts core psychological indicators based on an algorithm model, such as heart rate variability (HRV), micro-expression intensity, postural stability, and degree of abnormal movement. These features represent an individual's psychological representation in the current time period and are key variables for the system to determine emotional state.

[0130] The extracted single-modal features are fed into a multimodal fusion module. This module adaptively selects an appropriate fusion strategy based on parameters such as current scene illumination, motion amplitude, and signal quality. It generates a highly reliable unified psychological feature vector through weighted fusion, confidence scoring, and complementary modality compensation. The fusion result eliminates single-modal fluctuation errors, improving the overall stability and accuracy of the model.

[0131] The fused features are then fed into the psychological state modeling module. The system uses deep neural networks (such as CNN and LSTM) and psychological auxiliary models (such as emotion-cognition mapping models and personality bias correction models) to classify the input into a state and assess the risk, thereby obtaining the current individual's psychological state label and risk level, such as "moderate anxiety" or "increased aggression".

[0132] Finally, the analysis results are structured and generated by the output module, making them available for use by the behavior prediction engine, individual profiling system, or remote platform. The entire process supports millisecond-level real-time performance and can run on edge devices, representing the core technological path for the system to achieve accurate psychological state recognition.

[0133] The system involved in this invention has the following advantages:

[0134] (1) The system integrates high-precision computer vision, deep learning, millimeter-wave radar, rPPG technology and infrared thermal imaging technology to achieve high-precision real-time assessment of individual psychological state;

[0135] (2) Non-perceptible detection: Using multimodal fusion technology, psychological state and personality characteristics can be analyzed with high precision without the user's active cooperation.

[0136] (3) High accuracy: It integrates multiple sensors such as high-definition RGB camera, millimeter-wave radar, rPPG technology, and infrared thermal imaging to improve recognition accuracy.

[0137] (4) Strong real-time performance: Combining edge computing and deep learning enables rapid data processing and improves the real-time performance of psychological state analysis.

[0138] (5) Multi-dimensional integration: It not only detects facial expressions, but also combines physiological signals, posture, voice and other factors to achieve a more comprehensive psychological assessment.

[0139] (6) Automated analysis: Based on deep learning models, reduce manual intervention and improve the automation level of the system.

[0140] (7) Professional psychological model support: This system combines personality traits, emotional states, defense mechanisms, psychological needs, and psychologically driven feature fusion algorithms to ensure the scientific nature of psychological state and personality analysis, and improve the reliability and applicability of the assessment.

[0141] (8) Personalized prediction: Using time series modeling and reinforcement learning techniques, it is possible to predict an individual's future psychological trends.

[0142] (9) Wide range of applications: It can be used in multiple fields such as security monitoring, mental health management, recruitment screening, and behavior prediction.

[0143] (10) Data security and privacy protection: Reduce the transmission of sensitive data and improve privacy protection capabilities through local edge computing.

[0144] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A psychological state and personality analysis system based on non-perceptual technology, characterized in that, include: Data acquisition module, feature extraction module, psychological state analysis module, psychological profiling module, and behavior prediction module; in, The data acquisition module is responsible for collecting multimodal data and transmitting it to the feature extraction module; The feature extraction module is responsible for preprocessing the data, labeling features, and analyzing behavioral patterns to generate standardized psychological feature vectors and provide real-time psychological features to the psychological state analysis module. The psychological state analysis module uses deep learning models and psychological modeling algorithms to identify an individual's current state, assess the individual's current psychological state, and provide data to the psychological profile module to update the individual's long-term psychological state characteristics. The psychological profiling module is responsible for storing and updating the baseline of an individual's psychological state, providing personalized data support for the behavior prediction module; The behavior prediction module combines psychological state analysis and individual profiling to predict future changes in an individual's psychological state and provide risk warnings or personalized suggestions.

2. The psychological state and personality analysis system based on non-perceptual technology as described in claim 1, characterized in that, The source of the multimodal data is the raw data of facial expressions, postures, behaviors, and physiological signals. The feature extraction module performs preprocessing, feature annotation, and behavioral pattern analysis on the collected data; this module includes physiological signal analysis and behavioral pattern recognition. Psychological state analysis module: Based on extracted psychological features, it analyzes the individual's current psychological state characteristics in real time; among which, psychological state characteristics include stress and emotional fluctuations; The psychological profiling module receives long-term time-series data from the psychological state analysis module, models the individual's behavior and emotional trajectory, constructs a personalized psychological profile, and provides a historical psychological state baseline. The behavior prediction module combines psychological state analysis and individual profiling to predict an individual's future psychological trends and behavioral patterns.

3. The psychological state and personality analysis system based on non-perceptual technology as described in claim 1, characterized in that, The data acquisition module is a device that integrates an RGB camera, a surveillance camera, a millimeter-wave radar, an rPPG sensor, an infrared thermal imaging camera, and a voice recording device. RGB camera: Captures facial expressions and micro-expression features; Surveillance cameras: acquire data on human posture, gestures, and behavior; Millimeter-wave radar: non-contact acquisition of users' heart rate and respiratory rate, enhancing physiological signal analysis capabilities; rPPG sensor: detects changes in facial hemodynamics, calculates heart rate, infers heart rate variability, and is used for stress and mood fluctuation assessment; Infrared thermal imaging camera: monitors temperature changes in different parts of the body, detects blood circulation patterns, and assists in psychological stress assessment.

4. The psychological state and personality analysis system based on non-perceptual technology as described in claim 3, characterized in that, The method for calculating heart rate using the rPPG sensor is as follows: (1) Extract ROI; (2) Apply color space conversion to the RGB image signal; (3) Noise is filtered out by bandpass filtering at 0.7–4 Hz; (4) Apply FFT analysis to find the peak of the main frequency; (5) Formula for calculating heart rate: HR = f peak ×60 Among them, f peak It is the peak value of the main frequency.

5. The psychological state and personality analysis system based on non-perceptual technology as described in claim 2, characterized in that, The feature extraction module includes physiological signal analysis and behavioral pattern recognition; Among them, behavior pattern recognition is face and expression recognition, which uses deep learning algorithms to analyze facial expressions and micro-expressions; The physiological signal analysis combines millimeter-wave radar and rPPG technology to detect individual physiological status in real time. Physiological signal analysis includes: posture and motion analysis, skin temperature and hemodynamic monitoring and analysis, multimodal fusion analysis, and combined speech analysis.

6. The psychological state and personality analysis system based on non-perceptual technology as described in claim 5, characterized in that, The posture and motion analysis specifically includes: (1) The skeleton point detection algorithm tracks human joints in real time and evaluates behavioral patterns; (2) Motion vector analysis identifies abnormal postures, stressful behaviors, and potential aggressive actions; (3) Based on labeled video training, optimize the action recognition accuracy to ensure that the algorithm can accurately distinguish between normal and abnormal behavior; The skin temperature and hemodynamic monitoring and analysis: Infrared thermal imaging was used to analyze changes in psychological stress; The multimodal fusion analysis integrates physiological signals and behavioral characteristics to improve the accuracy of psychological state assessment. The combination of speech analysis: by analyzing pitch, speech rate, and intonation fluctuations, it helps to judge psychological state and improves the accuracy of psychological state assessment.

7. The psychological state and personality analysis system based on non-perceptual technology as described in claim 1, characterized in that, The working principle of the psychological profiling module includes: (1) Establish an individual psychological state feature vector by fusing behavioral and physiological data; (2) Data-driven feature clustering algorithm is used to classify the psychological states of different individuals; (3) Combine time series modeling to predict future behavioral trends of individuals and provide personalized intervention suggestions; (4) The results of user psychological state analysis are displayed through a visualization system, which is applicable to security, mental health management and personalized service recommendations.

8. The psychological state and personality analysis system based on non-perceptual technology as described in claim 1, characterized in that, The behavior prediction module uses a deep learning model to predict behavior and analyze future trends in psychological state. It combines Bayesian networks and reinforcement learning algorithms to optimize the short-term and long-term prediction capabilities of psychological states.

9. A method for a psychological state and personality analysis system based on non-perceptual technology as described in claim 1, characterized in that, Includes the following steps: Step 1: The data acquisition module collects multimodal data and transmits it to the feature extraction module; Step 2: The feature extraction module preprocesses the data, performs feature annotation and behavioral pattern analysis, generates standardized psychological feature vectors, and provides real-time psychological features to the psychological state analysis module. Step 3: The psychological state analysis module uses deep learning models and psychological modeling algorithms to determine the individual's current state, assess the individual's current psychological state, and provide data to the psychological profile module to update the individual's long-term psychological state characteristics. Step 4: The psychological profiling module stores and updates the baseline of an individual's psychological state, providing personalized data support for the behavior prediction module. Step 5: The behavior prediction module combines psychological state analysis and individual profiles to predict future changes in an individual's psychological state and provides risk warnings or personalized suggestions.

Images