System for detecting mental diseases through face recognition based on artificial intelligence technology

Abstract

The invention provides a system for detecting mental diseases through face recognition based on an artificial intelligence technology, and the system comprises a collection unit which collects the original image data of a face through a plurality of cameras, and the original image data comprises static image data and dynamic image data; the static image data and the dynamic image data are respectively sent to a static analysis module and a dynamic analysis module; the static analysis module comprises a recognition unit and a first partition unit, the recognition unit obtains static key point data according to the static image data and recognizes facial expressions according to the static key point data, and the first partition unit divides a human face into a plurality of areas according to the static key point data; the dynamic analysis module is used for analyzing dynamic key point data obtained through dynamic image data under a preset condition, and the dynamic key point data comprises pupil action track parameters, mouth action track parameters and face jittering parameters.

Description

Technical Field

[0001] The present invention belongs to the field of recognition and detection, and in particular relates to a system for detecting mental illness by face recognition based on artificial intelligence technology. Background Art

[0002] With the rapid development of artificial intelligence technology, face recognition technology has been widely used in many fields, including security monitoring, identity authentication and human-computer interaction. In the field of medical health, especially in the identification and monitoring of mental illness, face recognition technology has also begun to show its potential. Studies have shown that facial expressions and emotional states are closely related to an individual's mental health. By analyzing facial features and expression changes, the risk of mental illness can be identified and assessed to a certain extent. However, there are still some technical shortcomings and challenges in the current AI-based facial recognition systems for detecting mental illness. Existing facial recognition technology may lead to a decrease in recognition accuracy under changes in lighting, angles, and facial expressions. The facial expressions of patients with mental illness may change due to the disease state, and may not even be obvious in some cases. This makes it possible for traditional facial recognition algorithms to fail to accurately capture key emotional signals when dealing with these complex situations, thus affecting the system's judgment and recognition results.

[0003] Current face recognition systems often rely on a large amount of labeled data for training, but in the field of mental illness, relevant labeled data is relatively scarce. The manifestations of many mental illnesses are highly individual, and emotional expressions may also vary in different cultural backgrounds. This makes it difficult for existing systems to obtain sufficient samples during the training phase, which affects the generalization ability and accuracy of the algorithm. Existing technologies are still insufficient in the depth and detail of sentiment analysis. Many systems rely solely on expression recognition technology and fail to combine other physiological signals (such as heart rate, skin galvanic response, etc.) for more comprehensive sentiment analysis. This one-sidedness may lead to misjudgment of the patient's mental state and affect subsequent treatment decisions. Existing face recognition technology is also controversial in terms of ethics and privacy protection. Patients with mental illness are usually very sensitive to personal privacy, and the application of face recognition technology may lead to infringement of their privacy rights, leading to patients' resistance to technology. If this privacy issue is not properly resolved, it will affect patients' acceptance and the popularization and application of the system. Many existing face recognition detection systems lack sufficient real-time and intelligence and cannot provide immediate feedback and support in clinical settings.

[0004] The identification and intervention of mental illness usually requires a rapid response, but traditional systems often cannot meet this requirement, thus affecting the efficiency and effectiveness of clinical decision-making. Therefore, the development of a system for detecting mental illness by face recognition based on artificial intelligence technology aims to overcome various defects in current technology by introducing more advanced algorithms and multimodal data analysis. Such a system will be able to more accurately identify and analyze the facial expressions and emotional states of patients, provide effective support for the early identification and intervention of mental illness, and thus improve the quality of life and treatment effects of patients. Summary of the invention

[0005] The present invention proposes a system for detecting mental illness by face recognition based on artificial intelligence technology, aiming to solve multiple technical challenges in the detection of mental illness. First, traditional methods for diagnosing mental illness usually rely on clinical interviews and questionnaires, which have the problems of subjectivity and dependence on professional knowledge, and are easily affected by the experience and judgment of doctors, leading to misdiagnosis or missed diagnosis. By introducing face recognition technology and dynamic behavior analysis, this system provides an objective, data-driven evaluation method, which can reduce the influence of human factors on the diagnosis results to a certain extent.

[0006] The technical solution of the present invention is implemented as follows: a system for detecting mental illness by face recognition based on artificial intelligence technology, comprising a collection unit, which collects original picture data of a face through a plurality of cameras, wherein the original picture data comprises static image data and dynamic image data, and the static image data and the dynamic image data are respectively sent to a static analysis module and a dynamic analysis module; the static analysis module comprises a recognition unit and a first partitioning unit, wherein the recognition unit obtains static key point data according to the static image data, and recognizes facial expressions according to the static key point data, and the first partitioning unit divides the face into a plurality of regions according to the static key point data; the dynamic analysis module obtains dynamic key point data through the dynamic image data under preset conditions, wherein the dynamic key point data comprises pupil motion trajectory parameters, mouth motion trajectory parameters and facial jitter parameters; The dynamic analysis module includes a pupil analysis unit, a mouth analysis unit, a face analysis unit, a storage unit and a disease classification unit. The dynamic key point data includes pupil motion trajectory parameters, mouth motion trajectory parameters and face jitter parameters. The pupil motion trajectory parameters are obtained by the pupil analysis unit, the mouth motion trajectory parameters are obtained by the mouth analysis unit, and the face jitter parameters are obtained by the face analysis unit. The pupil analysis unit, the mouth analysis unit and the face analysis unit store the obtained parameters in the storage unit. The storage unit pre-stores several types of mental illnesses. The disease classification unit calls the parameter data in the storage unit, analyzes the parameter data and the The correspondence between the types of mental illness; the pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; the pupil analysis unit includes an eye flipping subunit, a blinking subunit, a pupil offset subunit, and a pupil shape change subunit, the eye flipping subunit identifies the eyelid movement frequency and movement amplitude, the blinking subunit identifies the blinking movement frequency, the pupil offset subunit identifies the pupil and eye socket offset frequency, the pupil shape change subunit identifies the pupil shape change frequency, and the eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters are obtained according to the movement frequency, movement amplitude, offset frequency, and shape change frequency. The movement amplitude is obtained by measuring the minimum movement depth and movement displacement length of the eyelids. The eye-rolling subunit determines the movement frequency, and calculates the movement depth and movement displacement length, and obtains an eyelid-rolling coefficient based on the minimum movement depth and movement displacement length. The blinking parameter is obtained based on the blinking movement frequency and movement amplitude. The movement amplitude includes the blinking movement depth and movement displacement length. The movement frequency is recorded after being determined by the blinking subunit, and a blinking coefficient is obtained through the blinking frequency, blinking movement depth and movement displacement length. The pupil shape change is calculated based on the initial image of the pupil and the pupil movement trajectory parameters. The eyelid-rolling coefficient is obtained based on the blinking parameters and the eyelid-rolling parameters, and the disease is preliminarily classified based on the eyelid-rolling coefficient.

[0007] Compared with existing technologies, this AI-based facial recognition system for detecting mental illness shows many significant differences and innovations. From a technical architecture perspective, existing mental illness detection systems rely heavily on traditional questionnaires and clinical observations, lacking data-driven intelligent analysis capabilities. This system uses the collaborative work of multiple cameras to achieve comprehensive acquisition of static and dynamic images. Combining static and dynamic analysis modules, it can capture facial expressions and emotional changes of mental patients in different time periods and situations. This multi-dimensional data collection method significantly improves the accuracy and sensitivity of emotion recognition. Secondly, existing technologies often only focus on changes in facial expressions when analyzing emotions, while ignoring the important information conveyed by eye and mouth movements.

[0008] In contrast, this system uses pupil analysis units, mouth analysis units and face analysis units to deeply analyze the eye and mouth dynamics of mental patients, extract important features such as pupil movement trajectory, mouth movement trajectory and facial jitter parameters, and ensure a comprehensive assessment of the mental state of mental patients. This detailed analysis method is not common in traditional systems, and greatly enhances the system's ability to identify emotions and assess mental states. Furthermore, existing systems often lack real-time feedback when processing data, resulting in the inability to apply analysis results in a timely manner. This system, by building a real-time monitoring mechanism, records the expressions and behavioral changes of mental patients in a timely manner, thereby improving the system's adaptability and response speed to dynamic environments. Existing technologies often rely on fixed rules and models when classifying diseases, lacking flexibility and intelligence. This system, through an analysis method based on machine learning, can continuously optimize and adjust the classification model based on real-time data to ensure the accuracy and reliability of classification. This adaptive capability enables the system to maintain efficient diagnostic capabilities in a changing environment. Existing facial recognition technology has problems with accuracy and robustness. This system, by using advanced algorithms and data processing technology, can more accurately capture the facial features and emotional states of mental patients, ensuring the reliability of the diagnostic process.

[0009] The accuracy and real-time performance of existing technologies in emotion recognition and facial expression analysis are still insufficient. Traditional facial expression recognition algorithms are often unable to cope with complex changes in emotional states, while this system, through the combination of static and dynamic analysis modules, can fully capture facial expressions and movement trajectories, thereby improving the accuracy of emotion recognition. Furthermore, many existing emotion analysis systems lack in-depth analysis of dynamic behaviors, especially careful observation of eye and mouth movements, which is crucial for judging mental states. This system achieves multi-dimensional behavior analysis through pupil analysis, mouth analysis, and extraction of facial jitter parameters, ensuring a comprehensive assessment of the user's mental state. Finally, existing systems usually lack sufficient sample data and algorithm support in disease classification and diagnosis, resulting in unsatisfactory classification results. By integrating big data and machine learning technology, this system can establish a parameter database based on a variety of mental illness types, improving the accuracy and reliability of disease detection.

[0010] As a preferred embodiment, the pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; the mouth movement trajectory parameters include mouth movement frequency, mouth movement depth, and mouth movement direction parameters.

[0011] As a preferred embodiment, the recognition unit includes a first recognition subunit, a second recognition subunit and a third recognition subunit. The first recognition subunit determines whether the mouth is open according to the mouth shape and records the direction of the mouth opening. The second recognition subunit determines the lip movement according to the mouth contour and records the lip movement trajectory. The third recognition subunit determines whether there is saliva according to the mouth contour. The data obtained by the first recognition subunit, the second recognition subunit and the third recognition subunit are stored and recorded by the data storage subunit, and the facial jitter parameters are obtained by the static analysis module.

[0012] As a preferred embodiment, the facial analysis unit includes a shaking subunit and a facial feature change subunit. The shaking subunit determines the shaking of the face by identifying a static image to obtain a facial shaking parameter. The facial feature change subunit records the change parameters of other facial features and stores them in the storage subunit.

[0013] As a preferred embodiment, the disease classification unit includes a storage module and a classification module, the recognition unit obtains static key point data based on the static image data, and recognizes facial expressions based on the static key point data, the storage unit contains a number of facial expressions that reflect specific diseases, and the facial expressions are divided into several different levels. The current facial expression is compared with the facial expressions of different levels in the storage unit to determine which level of facial expressions in the corresponding storage unit the current facial expression belongs to.

[0014] As a preferred implementation, the following steps are adopted for specific operations: static images and dynamic images of a human face are collected by several cameras; the static analysis module obtains static key point data according to the static image data, and recognizes facial expressions according to the static key point data, and the first partition unit divides the human face into several areas according to the static key point data; the dynamic analysis module analyzes the pupil movement trajectory parameters, mouth movement trajectory parameters and facial jitter parameters in the dynamic image data under preset conditions, and obtains blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters respectively; the blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters are stored, and a reaction level is obtained according to the blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters; based on the reaction level, a reaction level factor is obtained, and the type of mental illness is determined according to the level factor.

[0015] After adopting the above technical scheme, the beneficial effect of the present invention is that by introducing static and dynamic analysis modules, the system can realize comprehensive emotion recognition and improve the accuracy of mental state assessment. The comprehensive analysis of facial expressions, movement trajectories and behavior patterns of mental patients enables the system to better capture subtle emotional changes, thereby providing a more scientific psychological assessment in the early stage of intervention. This scientific assessment method not only helps to improve the self-cognition of mental patients, but also the real-time data collection and analysis capabilities of the system ensure that the emotional fluctuations of mental patients can be timely identified during the intervention process, helping the intervention personnel to better understand the psychological state of mental patients. This real-time feedback mechanism makes the intervention process more humane, helps to reduce the tension of mental patients, and improves the comfort and effectiveness of intervention. In addition, the system can timely identify potential mental health problems through monitoring and analysis of abnormal behaviors, providing a basis for subsequent psychological counseling and intervention. This early identification mechanism, by building a rich parameter database and intelligent algorithm, the system can continuously optimize its analysis model to ensure that it can maintain efficient working ability in different environments. This intelligent self-learning ability enables the system to show higher flexibility and adaptability in application, further enhancing its market competitiveness. This system not only improves the scientificity and fairness of intervention, but also helps promote the further development of mental health management and bring more value to mental patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0017] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example like Figure 1 As shown, a system for detecting mental illness by face recognition based on artificial intelligence technology includes a collection unit, which collects original picture data of a face through a plurality of cameras, wherein the original picture data includes static image data and dynamic image data, and the static image data and the dynamic image data are respectively sent to a static analysis module and a dynamic analysis module; the static analysis module includes a recognition unit and a first partitioning unit, wherein the recognition unit obtains static key point data according to the static image data, and recognizes facial expressions according to the static key point data, and the first partitioning unit divides the face into a plurality of areas according to the static key point data; the dynamic analysis module obtains dynamic key point data through the dynamic image data under preset conditions, wherein the dynamic key point data includes pupil motion trajectory parameters, mouth motion trajectory parameters and facial jitter parameters; The dynamic analysis module includes a pupil analysis unit, a mouth analysis unit, a face analysis unit, a storage unit and a disease classification unit. The dynamic key point data includes pupil motion trajectory parameters, mouth motion trajectory parameters and face jitter parameters. The pupil motion trajectory parameters are obtained by the pupil analysis unit, the mouth motion trajectory parameters are obtained by the mouth analysis unit, and the face jitter parameters are obtained by the face analysis unit. The pupil analysis unit, the mouth analysis unit and the face analysis unit store the obtained parameters in the storage unit. The storage unit pre-stores several types of mental illnesses. The disease classification unit calls the parameter data in the storage unit, analyzes the parameter data and the The correspondence between the types of mental illness; the pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; the pupil analysis unit includes an eye flipping subunit, a blinking subunit, a pupil offset subunit, and a pupil shape change subunit, the eye flipping subunit identifies the eyelid movement frequency and movement amplitude, the blinking subunit identifies the blinking movement frequency, the pupil offset subunit identifies the pupil and eye socket offset frequency, the pupil shape change subunit identifies the pupil shape change frequency, and the eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters are obtained according to the movement frequency, movement amplitude, offset frequency, and shape change frequency. The movement amplitude is obtained by measuring the minimum movement depth and movement displacement length of the eyelids. The eye-rolling subunit determines the movement frequency, and calculates the movement depth and movement displacement length, and obtains an eyelid-rolling coefficient based on the minimum movement depth and movement displacement length. The blinking parameter is obtained based on the blinking movement frequency and movement amplitude. The movement amplitude includes the blinking movement depth and movement displacement length. The movement frequency is recorded after being determined by the blinking subunit, and a blinking coefficient is obtained through the blinking frequency, blinking movement depth and movement displacement length. The pupil shape change is calculated based on the initial image of the pupil and the pupil movement trajectory parameters. The eyelid-rolling coefficient is obtained based on the blinking parameters and the eyelid-rolling parameters, and the disease is preliminarily classified based on the eyelid-rolling coefficient.

[0020] The core of this technology is to use artificial intelligence technology to perform face recognition analysis to detect potential mental illness. The workflow of the entire system can be divided into several main steps, including data acquisition, static and dynamic analysis, parameter calculation, data storage and classification, etc. The following is a detailed description of the specific operation steps and working principles.

[0021] 1. Data Collection The system first collects raw image data of human faces through several cameras, which are divided into static images and dynamic images. Static image data can be photos of users' expressions in specific situations, while dynamic image data are facial movements and expression changes of users in a natural state. Step 1.1: The camera performs face recognition to ensure that the captured image is clear and meets the system requirements. Step 1.2: The captured static and dynamic image data are transmitted to the static analysis module and the dynamic analysis module respectively.

[0022] 2. Static Analysis The work of the static analysis module is mainly to identify facial expressions and their key points through static image data. Step 2.1: The recognition unit uses machine learning algorithms to analyze static images and extract static key point data, including the coordinates of facial features such as eyes, mouth, nose, etc. Step 2.2: Based on the static key point data, the system recognizes the user's facial expressions, such as happiness, sadness, anger, etc. Step 2.3: The first partitioning unit divides the face into several areas (such as eye area, mouth area, etc.) for subsequent analysis.

[0023] 3. Dynamic Analysis The dynamic analysis module analyzes the user's facial dynamic performance, including the movement trajectory of the eyes and mouth. Step 3.1: Under the set conditions, the dynamic analysis module uses the dynamic image data to obtain dynamic key point data. Step 3.2: The system analyzes the movement trajectory of the pupil, the movement trajectory of the mouth, and the jitter parameters of the face respectively. Pupil analysis unit: collects data on eyelid flipping, blinking, pupil displacement and shape changes. Mouth analysis unit: analyzes the degree and frequency of mouth opening and closing. Face analysis unit: monitors the overall jitter of the face.

[0024] 4. Parameter calculation During the dynamic analysis, the system calculates multiple parameters that help further classify mental illnesses. Step 4.1: The pupil analysis unit calculates eyelid flipping parameters, blinking parameters, pupil-orbital offset parameters, and pupil shape change parameters. Eyelid flipping parameters: The eyelid flipping coefficient is calculated by measuring the eyelid movement depth and displacement length. Blinking parameters: The blinking frequency and amplitude are recorded to obtain the blinking coefficient. Pupil shape change: The pupil shape change is calculated based on the initial image and motion trajectory parameters.

[0025] 5. Data storage and classification After completing the calculation of all parameters, the system stores the data and performs a preliminary classification of mental illness. Step 5.1: The storage unit stores all the obtained parameter data and pre-stores characteristic data of various types of mental illness. Step 5.2: The disease classification unit calls the data in the storage unit and analyzes the correspondence between the current parameter data and the pre-stored mental illness characteristics. Step 5.3: By comparing the dynamic and static parameters of the current user, the system determines the possible mental illness risk of the user and generates a corresponding analysis report.

[0026] 6. Result feedback Finally, the system will feed back the analysis results to the user or a professional psychologist for further evaluation and treatment. Step 6.1: Generate a detailed analysis report, including the user's facial expressions, key parameters, and possible types of mental illness. Step 6.2: The system can provide advice and guide the user to seek professional psychological counseling or treatment.

[0027] Through the above steps, this technology has achieved in-depth analysis of human faces, and used artificial intelligence technology to conduct preliminary detection and classification of mental illness, providing new tools and methods for mental health management. The key to the entire system lies in efficient data collection, accurate image analysis and scientific parameter calculation, all of which provide strong support for the early detection and intervention of mental illness.

[0028] The pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; the mouth movement trajectory parameters include mouth movement frequency, mouth movement depth, and mouth movement direction parameters.

[0029] The recognition unit includes a first recognition subunit, a second recognition subunit and a third recognition subunit. The first recognition subunit determines whether the mouth is open according to the mouth shape and records the direction of the mouth opening. The second recognition subunit determines the lip movement according to the mouth contour and records the lip movement trajectory. The third recognition subunit determines whether there is saliva according to the mouth contour. The data obtained by the first recognition subunit, the second recognition subunit and the third recognition subunit are stored and recorded by the data storage subunit, and the facial jitter parameters are obtained by the static analysis module.

[0030] The facial analysis unit includes a shaking subunit and a facial features change subunit. The shaking subunit determines the shaking of the face by identifying a static image and obtains a facial shaking parameter. The facial features change subunit records the change parameters of other facial features and stores them in the storage subunit.

[0031] The disease classification unit includes a storage module and a classification module. The recognition unit obtains static key point data based on the static image data and recognizes facial expressions based on the static key point data. The storage unit contains several facial expressions that reflect specific diseases. Facial expressions are divided into several different levels. The current facial expression is compared with facial expressions of different levels in the storage unit to determine which level of facial expressions in the corresponding storage unit the current facial expression belongs to.

[0032] The following steps are adopted for specific operation: static images and dynamic images of human faces are collected through several cameras; the static analysis module obtains static key point data according to the static image data, and recognizes facial expressions according to the static key point data, and the first partition unit divides the human face into several areas according to the static key point data; the dynamic analysis module analyzes pupil movement trajectory parameters, mouth movement trajectory parameters and face jitter parameters in the dynamic image data under preset conditions, and obtains blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters respectively; the blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters are stored, and a reaction level is obtained according to the blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters; based on the reaction level, a reaction level factor is obtained, and the type of mental illness is determined according to the level factor.

[0033] The artificial intelligence-based face recognition system described in this technology has multiple preferred implementations and significant innovations, especially in the detection of mental illness. The following will analyze each preferred implementation in detail, discussing its differences from the prior art and its characteristics.

[0034] The pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters. In the prior art, eye tracking technology mainly focuses on changes in blinking frequency and pupil size, and often ignores important parameters such as eyelid flipping and pupil shape changes. These parameters can not only provide richer eye movement information, but also help identify the user's emotions or mental state. For example, eyelid flipping may be associated with anxiety or uneasiness, while changes in pupil shape may reveal the user's physiological reactions or fluctuations in mental state. By introducing these detailed parameters, the system can capture eye performance more comprehensively, thereby improving the detection accuracy of mental illness.

[0035] The mouth movement trajectory parameters include the frequency of mouth movement, the depth of mouth movement, and the direction parameters of mouth movement. Compared with the prior art, the existing system usually only focuses on the opening and closing state of the mouth, while ignoring the dynamic characteristics such as the frequency and depth of mouth movement. The frequency of mouth movement can reflect the user's tension, while the depth and direction of the movement can provide clues to the user's emotional changes. For example, frequent mouth movements may be related to an anxious, nervous or uneasy psychological state. By comprehensively analyzing these additional parameters, the system can more accurately assess the user's psychological state.

[0036] The recognition unit includes a first recognition subunit, a second recognition subunit and a third recognition subunit, which are respectively used to judge the mouth movement, lip movement and the appearance of saliva. The prior art often uses a single recognition module to process mouth movements, while the present system can capture different mouth movements more comprehensively by subdividing the recognition unit into multiple subunits. This modular design allows the system to independently analyze and record the characteristics of each mouth movement, thereby improving the overall recognition accuracy. In addition, this subdivision structure can support more complex sentiment analysis. For example, the appearance of saliva may be associated with anxiety or tension.

[0037] The static analysis module includes a recognition unit, a first partition unit, a pupil analysis unit, a mouth analysis unit, a face analysis unit, a storage unit, and a disease classification unit. The static analysis module in the prior art is usually relatively simple and often only focuses on the recognition of facial expressions. However, by introducing multiple analysis units, this system can perform more in-depth analysis. For example, the pupil analysis unit can provide detailed information about the state of the eyes, while the mouth analysis unit can analyze subtle changes in the mouth. This multi-dimensional analysis method not only improves the accuracy of recognition, but also provides more data support for the subsequent classification of mental illnesses.

[0038] The dynamic key point data includes pupil motion trajectory parameters, mouth motion trajectory parameters and facial jitter parameters. The existing technology usually only focuses on one aspect of performance during dynamic analysis, while the design of this system allows simultaneous analysis of multiple dynamic parameters. This comprehensive dynamic analysis can capture the complex behavior patterns of users in specific situations, thereby providing a more comprehensive mental state assessment. For example, facial jitter parameters can indicate the user's nervous state, while the dynamic changes of the pupils and mouth can reveal the user's emotional response.

[0039] The facial analysis unit includes a shaking subunit and a facial features change subunit, which record facial shaking and other facial features change parameters respectively. In the prior art, facial analysis often focuses only on expression recognition, while this system can more comprehensively evaluate the user's emotional state by introducing the analysis of facial features changes. For example, facial features changes can reveal subtle changes in the user's anxiety, anger, happiness and other emotions. This meticulous analysis method enables the system to better capture the user's true emotions, thereby improving the accuracy of mental illness detection.

[0040] The structure of the disease classification unit is set by the disease classification unit, including a storage module and a classification module, which can judge the type of mental illness according to the level of facial expression. The existing technology often adopts a simple classification method and lacks in-depth analysis of the details of facial expressions. However, this system can provide a more accurate classification of mental illness by integrating the levels and characteristics of facial expressions. For example, different levels of facial expressions can indicate different psychological states, and the system can judge the user's mental health status based on the changes in these levels. This detailed classification method improves the scientificity and rationality of mental illness detection.

[0041] The operation is carried out in specific steps, covering the entire process from data collection to mental illness judgment. The existing technology is often simple in operation steps and lacks systematicity. However, this system ensures that the data processing and analysis of each link are scientific and systematic through clear operation steps. For example, from camera data collection, static analysis to dynamic analysis, and then to the final mental illness judgment, each step has been rigorously designed to ensure the accuracy of the data and the effectiveness of the analysis. This systematic operation process improves the reliability of the overall detection.

[0042] In summary, this technology has shown obvious innovation and improvement in the definition of eye and mouth movement parameters, the structure of the recognition unit, the composition of the static analysis module, the acquisition of dynamic key point data, the function of the facial analysis unit, the structure of the disease classification unit, and the specific operation steps. These preferred implementations not only improve the accuracy of mental illness detection, but also provide a more scientific and effective means for future mental health management. Through more detailed parameter analysis and systematic operation procedures, this technology will have broad prospects and social value in practical applications.

[0043] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for detecting mental illness by face recognition based on artificial intelligence technology, characterized in that: It includes a collection unit, which collects original picture data of a human face through a plurality of cameras, wherein the original picture data includes static image data and dynamic image data, and respectively sends the static image data and the dynamic image data to a static analysis module and a dynamic analysis module; the static analysis module includes a recognition unit and a first partitioning unit, wherein the recognition unit obtains static key point data according to the static image data and recognizes facial expressions according to the static key point data, and the first partitioning unit divides the human face into a plurality of areas according to the static key point data; The dynamic analysis module analyzes the dynamic image data under preset conditions to obtain dynamic key point data, wherein the dynamic key point data includes pupil motion trajectory parameters, mouth motion trajectory parameters and facial jitter parameters; The dynamic analysis module includes a pupil analysis unit, a mouth analysis unit, a face analysis unit, a storage unit and a disease classification unit. The pupil movement trajectory parameters are obtained by the pupil analysis unit, the mouth movement trajectory parameters are obtained by the mouth analysis unit, and the face shaking parameters are obtained by the face analysis unit. The pupil analysis unit, the mouth analysis unit and the face analysis unit store the obtained parameters in the storage unit. The storage unit pre-stores several types of mental illness. The disease classification unit calls the parameter data in the storage unit and analyzes the corresponding relationship between the parameter data and the type of mental illness. The pupil motion trajectory parameters include eyelid flipping parameters, eye blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; The pupil analysis unit includes an eye-rolling subunit, a blinking subunit, a pupil shift subunit and a pupil shape change subunit, wherein the eye-rolling subunit identifies the eyelid movement frequency and movement amplitude, the blinking subunit identifies the blinking movement frequency, the pupil shift subunit identifies the pupil and eye socket shift frequency, and the pupil shape change subunit identifies the pupil shape change frequency, and obtains eyelid rolling parameters, blinking parameters, pupil and eye socket shift parameters and pupil shape change parameters according to the movement frequency, movement amplitude, shift frequency and shape change frequency respectively; The movement amplitude is obtained by measuring the minimum movement depth and movement displacement length of the eyelids. The eye-rolling subunit determines the movement frequency, and calculates the movement depth and the movement displacement length, and obtains an eyelid-rolling coefficient according to the minimum movement depth and the movement displacement length. The blinking parameter is obtained according to the movement frequency and movement amplitude of the blinking. The movement amplitude includes the movement depth and the movement displacement length of the blinking. The movement frequency is recorded after being determined by the blinking subunit, and a blinking coefficient is obtained through the blinking frequency, the blinking movement depth and the movement displacement length. The pupil shape change is calculated according to the initial image of the pupil and the movement trajectory parameters of the pupil. An eyelid-turning coefficient is obtained according to the blinking parameter and the eyelid-turning parameter, and a preliminary classification of the disease is performed based on the eyelid-turning coefficient.

2. The system for detecting mental illness by face recognition based on artificial intelligence technology according to claim 1 is characterized in that: The pupil movement trajectory parameters include eyelid flipping parameters, blinking parameters, pupil and eye socket offset parameters, and pupil shape change parameters; the mouth movement trajectory parameters include mouth movement frequency, mouth movement depth, and mouth movement direction parameters.

3. The system for detecting mental illness by face recognition based on artificial intelligence technology according to claim 1 is characterized in that: The recognition unit includes a first recognition subunit, a second recognition subunit and a third recognition subunit. The first recognition subunit determines whether the mouth is open according to the mouth shape and records the direction of the mouth opening. The second recognition subunit determines the lip movement according to the mouth contour and records the lip movement trajectory. The third recognition subunit determines whether there is saliva according to the mouth contour. The data obtained by the first recognition subunit, the second recognition subunit and the third recognition subunit are stored and recorded by the data storage subunit, and the facial jitter parameters are obtained by the static analysis module.

4. The system for detecting mental illness by face recognition based on artificial intelligence technology according to claim 1 is characterized in that: The facial analysis unit includes a shaking subunit and a facial features change subunit. The shaking subunit determines the shaking of the face by identifying a static image and obtains a facial shaking parameter. The facial features change subunit records the change parameters of other facial features and stores them in the storage subunit.

5. The system for detecting mental illness by face recognition based on artificial intelligence technology according to claim 1 is characterized in that: The disease classification unit includes a storage module and a classification module. The recognition unit obtains static key point data based on static image data and recognizes facial expressions based on the static key point data. The storage unit contains several facial expressions that reflect specific diseases. Facial expressions are divided into several different levels. The current facial expression is compared with facial expressions of different levels in the storage unit to determine which level of facial expressions in the corresponding storage unit the current facial expression belongs to.

6. The system for detecting mental illness by face recognition based on artificial intelligence technology according to claim 1 is characterized in that: The specific operation is performed by the following steps: a static image and a dynamic image of a human face are collected by a plurality of cameras; the static analysis module obtains static key point data according to the static image data, and recognizes facial expressions according to the static key point data, and the first partition unit divides the human face into a plurality of regions according to the static key point data; the dynamic analysis module analyzes pupil motion trajectory parameters, mouth motion trajectory parameters and facial jitter parameters in the dynamic image data under preset conditions, and obtains blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters respectively; storing blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters, and obtaining a reaction level according to the blinking parameters, eyelid flipping parameters, pupil and eye socket offset parameters, and pupil shape change parameters; Based on the reaction levels, reaction level factors are obtained, and the types of mental illness are determined according to the level factors.

Images