A multifunctional safe identity recognition and access control method and system

Abstract

The application relates to the technical field of access control, and provides a multifunctional safe identity recognition and access control method and system, which comprises the following steps: continuously monitoring an operator after the operator completes identity confirmation and establishes an access session, obtaining interactive action information of the operator and the safe, and according to the interactive action information and preset task authority, when abnormal operation information which does not match the task authority is identified, non-contactly obtaining body state information of the operator, comparing the body state information with pre-established personal physiological reaction ranges of the operator in different situations, judging whether the operator is in an externally threatened state, if yes, silently sending alarm information and performing a deceptive operation, and if not, according to continuous operation behavior track information of the operator, distinguishing whether the abnormal operation information is an unintentional mistake or a malicious attempt, and taking corresponding guiding or rejecting measures. The application can overcome the limitation of static authorization.

Description

Technical Field

[0001] This application relates to the field of access control technology, and more specifically, to a multifunctional safe identification and access control method and system. Background Technology

[0002] In high-security safe access control scenarios, such as data centers or financial institutions, traditional methods typically rely on one-time identity verification, such as biometrics or password authentication, before access begins, tightly binding access permissions to personal identity. However, this mechanism has significant shortcomings in practical applications. It not only faces the risk of identity theft, but more importantly, it cannot flexibly adjust access permissions based on the operator's real-time behavior or changes in the context. This static authorization method proves inadequate in the face of emergencies or potential external threats to operators, thus posing a potential threat to overall security.

[0003] Specifically, when an authenticated and authorized operator exhibits abnormal behavior during a task—for example, attempting to access sensitive storage cells outside their authorized scope—existing systems often simply deny access and log the event, failing to promptly and accurately determine the true intent behind such abnormal behavior. The system struggles to distinguish whether this is due to unintentional misbehavior under emergency or high-pressure circumstances, malicious probing exploiting an emergency, or involuntary behavior under external coercion. Traditional access control systems lack the comprehensive control capability to assess an operator's trustworthiness based on their real-time behavior and even changes in physical condition throughout the authorized session, and to take differentiated responses accordingly. This limitation prevents the system from effectively intervening when incidents occur and from addressing complex situations where authorized identities are controlled or abused by others, thus leaving serious security vulnerabilities in high-security vault access control. Summary of the Invention

[0004] This application provides a multifunctional safe identification and access control method and system to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, this application discloses a multi-functional safe identification and access control method, including: After the operator completes identity verification and establishes an access session, continuous monitoring of the operator is initiated. During continuous monitoring, information on the interaction between the operator and the safe is acquired, and abnormal operation information that does not conform to the preset task permissions is identified based on the interaction information and the preset task permissions. In response to the detection of abnormal operation information, the system obtains the operator's physical status information non-contactly and compares the physical status information with the pre-established range of the operator's personal physiological response in different situations; Based on the comparison results, determine whether the operator is under external threat; if the operator is determined to be under external threat, silently send an alarm message and perform a deceptive operation. If it is determined that the operator is not under external threat, then based on the operator's continuous operational behavior trajectory information, it is determined whether the abnormal operation information is unintentional or malicious, and corresponding guidance or rejection measures are taken.

[0006] Secondly, this application also discloses a multi-functional safe identification and access control system, including: The monitoring startup module is used to start continuous monitoring of operators after they have completed identity verification and established an access session. The interaction action acquisition and recognition module is used to acquire interaction action information between the operator and the safe during continuous monitoring, and to identify abnormal operation information that does not conform to the task permissions based on the interaction action information and the preset task permissions. The body status acquisition and comparison module is used to acquire the operator's body status information non-contactly in response to the identification of abnormal operation information, and compare the body status information with the pre-established range of the operator's personal physiological response in different situations; The external threat assessment and alarm execution module is used to determine whether the operator is under external threat based on the comparison results; if the operator is determined to be under external threat, an alarm message is sent silently and a deceptive operation is performed. The intent differentiation and action execution module is used to determine whether the abnormal operation information is unintentional or malicious if it is determined that the operator is not under external threat, based on the operator's continuous operation behavior trajectory information, and to take corresponding guidance or rejection measures.

[0007] Compared with the prior art, this application has at least the following beneficial effects: This application effectively solves the technical problems in the prior art that it is impossible to flexibly adjust access permissions according to the real-time behavior and situational changes of operators, and that it is difficult to accurately judge the true intention of abnormal behavior. It significantly improves the intelligence, security and robustness of high-security safe access control, and effectively addresses complex security challenges such as identity theft, external threats and internal malicious attempts. Attached Figure Description

[0008] Figure 1 A flowchart illustrating a multifunctional safe identification and access control method provided in this application; Figure 2 This application provides a schematic diagram of the structure of a multifunctional safe identification and access control system. Detailed Implementation

[0009] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0010] like Figure 1 As shown, this application proposes a multi-functional safe identification and access control method, including: After the operator completes identity verification and establishes an access session, continuous monitoring of the operator is initiated. During continuous monitoring, information on the interaction between the operator and the safe is acquired, and abnormal operation information that does not conform to the preset task permissions is identified based on the interaction information and the preset task permissions. In response to the detection of abnormal operation information, the system obtains the operator's physical status information non-contactly and compares the physical status information with the pre-established range of the operator's personal physiological response in different situations; Based on the comparison results, determine whether the operator is under external threat; if the operator is determined to be under external threat, silently send an alarm message and perform a deceptive operation. If it is determined that the operator is not under external threat, then based on the operator's continuous operational behavior trajectory information, it is determined whether the abnormal operation information is unintentional or malicious, and corresponding guidance or rejection measures are taken.

[0011] Continuous monitoring refers to the system's ongoing observation and data collection of an operator's behavior within the safe's operating area after initial authentication and the establishment of an access session, rather than a one-time verification at the start of access. Interaction action information refers to all data captured and recorded by the system during operator interaction with the safe, such as hand gestures, touch behaviors, and operation sequences. Task permissions refer to the legal operating scope and accessible storage cells within the safe, pre-defined for specific operators or roles. Abnormal operation information refers to any interaction actions inconsistent with preset task permissions, such as attempting to open unauthorized storage cells. Physical status information refers to physiological indicators of the operator obtained non-contactly, such as heart rate, breathing rhythm, and skin temperature, reflecting the operator's psychological and physiological stress levels. Personal physiological response range refers to a pre-established reference baseline based on the operator's historical data and different situations (such as normal work, mild stress, high tension, etc.) used to assess whether their current physiological state is abnormal. External threat status refers to situations where the operator performs operations involuntarily under external pressure or coercion. Deceptive actions refer to situations where the system, while sending alerts, takes actions that appear normal but are actually designed to delay or mislead threat actors, such as slow response or displaying false error messages. Continuous operational behavior trajectory information refers to a series of actions performed by an operator over a period of time, along with their sequence and characteristics; analyzing these trajectories can infer the operator's intentions. Unintentional errors refer to abnormal behavior caused by operator negligence, misunderstanding, or operational mistakes, rather than malicious intent. Malicious attempts refer to actions by an operator that consciously attempt to violate security regulations or obtain unauthorized information. Guidance or denial measures refer to corrective or defensive actions taken by the system based on the assessment results, such as providing operational prompts, restricting access, or directly denying operation.

[0012] The multifunctional safe identification and access control method of this application initiates continuous monitoring of the operator after the operator completes identity verification and establishes an access session. For example, cameras, infrared sensors, or pressure sensors can be deployed in the safe's operating area to capture the operator's activities in real time. These sensors can continuously record the operator's position, posture, hand movements, and other information within the operating area. Alternatively, sensors can be integrated into a smart device worn by the operator (such as a smartwatch) to acquire their location and activity data in real time, with the operator's authorization, and transmit this data to the safe control system for continuous monitoring.

[0013] During continuous monitoring, the system acquires information on the operator's interactions with the safe and identifies abnormal operations that do not conform to the pre-defined task permissions. For example, the system can record the time, location, and type of each action the operator takes, such as touching the safe panel, opening a storage cell, or entering a password. This interaction information can include the operator's hand movement trajectory on the safe panel, the duration of touching a specific area, and the physical interaction with each storage cell. The system can also use deep learning models to analyze the video stream, identifying the operator's fine hand movements, such as finger pointing and pressing force, and using these as interaction information. Pre-defined task permissions define all operations that the operator is allowed to perform in the current session, such as only being able to access specific storage cells or only being able to operate within a specific time period. The system compares the real-time acquired interaction information with these pre-defined task permissions. If the operator attempts to access a storage cell outside their authorized scope or performs an operation sequence that does not conform to the authorized task, the system will identify it as abnormal operation information.

[0014] In response to the detection of abnormal operation information, the system non-contactly acquires the operator's physical condition information and compares it with a pre-established range of personal physiological responses for different situations. For example, when the system detects that the operator is attempting to open an unauthorized storage compartment, it immediately activates non-contact physiological sensors, such as infrared thermal imagers, millimeter-wave radar, or laser Doppler vibrometers, to acquire information such as the operator's heart rate, breathing rhythm, and skin temperature. These sensors can collect data from a distance without direct contact with the operator. Alternatively, the system can utilize a miniature microphone array embedded in the operating area to capture the operator's faint breathing and heartbeat sounds, and extract physiological indicators through acoustic analysis. The acquired physical condition information is then compared with the operator's pre-established range of personal physiological responses. This range is based on historical physiological data from different situations, such as normal work, mild stress, and high tension, and is used to assess the degree of deviation from the current physiological state.

[0015] Based on the comparison results, the system determines whether the operator is under external threat. If the operator is determined to be under external threat, a silent alarm message is sent, and deceptive actions are performed. For example, if the comparison results show that the operator's physiological indicators, such as heart rate, breathing rhythm, and skin temperature, significantly deviate from their normal working range and tend towards a highly tense or fearful physiological response pattern, the system will determine that the operator may be under external threat. In this case, the system will immediately and silently send an alarm message to the security center, ensuring that the threat attacker cannot detect it. At the same time, the system will perform deceptive actions, such as deliberately causing a certain authorized storage compartment of the safe to open slowly, or displaying a seemingly normal system maintenance prompt on the screen, to delay time and mislead the threat attacker, buying valuable time for security personnel to intervene.

[0016] If the system determines that the operator is not under external threat, it will distinguish between unintentional misoperation and malicious intent based on the operator's continuous operational behavior trajectory information, and take appropriate guidance or denial measures. For example, if the comparison results show that the operator's physiological state is within the normal or slightly tense range, the system will consider that they are not under external threat. At this time, the system will further analyze the operator's continuous operational behavior trajectory information before and after the abnormal operation. For example, if the operator immediately shows confusion or hesitation after attempting to open an unauthorized storage cell, and then tries to access their authorized storage cell again, this may be judged as unintentional. Alternatively, machine learning models can be used to learn the operator's historical operational behavior patterns to more accurately determine the intent of the current abnormal behavior. If it is judged to be unintentional, the system may provide guiding prompts on the operation interface, such as "You do not currently have permission to access this storage cell; please confirm your task permissions," to help the operator correct the error. If the analysis shows that the operator, after attempting an unauthorized operation, exhibits signs of repeated probing, rapidly switching targets, or trying to conceal their behavior, it may be judged as malicious intent. In this situation, the system will immediately deny access and trigger a higher-level alarm, or even lock the safe.

[0017] This application aims to address the security vulnerabilities and limitations of traditional safe access control systems by introducing continuous monitoring, abnormal operation identification, non-contact physical state assessment, and context-based intent judgment mechanisms. The core of this method lies in its focus not only on the operator's identity but also on their real-time behavior and physiological state during the access session. This allows for dynamic assessment of the operator's credibility and differentiated response measures based on different contexts.

[0018] The multifunctional safe identification and access control method of this application firstly initiates continuous monitoring of the operator after the operator completes identity verification and establishes an access session. This monitoring is not a one-time authentication but continues throughout the entire access session, capturing all interactive actions of the operator within the safe's operating area in real time. Secondly, during continuous monitoring, the system acquires information on the operator's interaction with the safe and, based on this information and preset task permissions, identifies any abnormal operation information that does not conform to the task permissions. For example, if the operator attempts to access a storage cell outside their permissions, the system immediately marks it as abnormal. Next, in response to the identification of abnormal operation information, the system non-contactly acquires the operator's physical state information, such as physiological indicators like heart rate, respiration, and skin temperature, and compares it with the operator's personal physiological response range under different situations. This step aims to determine whether the operator's abnormal behavior is caused by external threats (such as coercion). Subsequently, based on the comparison results, the system determines whether the operator is under external threat. If an external threat is detected, the system will silently send an alarm to the security center and perform deceptive actions to buy time and protect the operator. If the system determines that the operator is not under external threat, it will further analyze the operator's continuous operational behavior to distinguish between unintentional errors and malicious intent. Finally, based on the intent determination, the system will take appropriate guidance or denial measures, such as providing operation prompts or directly denying access and triggering an alarm. This entire process forms a closed-loop, dynamic security control mechanism that can effectively respond to various complex and unexpected situations.

[0019] In some embodiments, the steps of non-contactly acquiring the operator's physical state information in response to identifying abnormal operation information and comparing the physical state information with a pre-established range of the operator's personal physiological responses in different situations include: A group of miniature environmental sensors is deployed in the operating area of ​​the safe panel to measure the local temperature gradient around the operator's exposed skin and the airflow velocity and its minute fluctuations in the operating area in real time, so as to obtain local microenvironment data. Simultaneously activate the miniature environmental sensor array and the non-contact physiological sensor. The non-contact physiological sensor is used to collect the operator's heart rate, breathing rhythm and skin temperature in real time to obtain physiological data. Based on local microenvironment data and physiological data collection, a physiological and environmental data package is formed; The local processing unit contains an embedded physiological and environmental interaction logic module, which includes a set of functions based on human physiological principles to describe the effects of specific microenvironmental changes on physiological indicators such as heart rate, respiratory rhythm, and skin temperature. By utilizing the microenvironmental data in the physiological and environmental data package, and through the physiological and environmental interaction logic module, environmental effect compensation is performed on the physiological data to generate purified physiological data. The purified physiological data were compared with a pre-stored set of personalized physiological responses of operators in different situations to assess the degree of deviation.

[0020] The micro-environmental sensor array can be understood as a collection of micro-MEMS devices integrating temperature, humidity, and airflow sensors. Deployed in the safe's operating area, such as at the edge of the keyboard, fingerprint reader, or display screen, its purpose is to accurately capture changes in the local microenvironment around the operator's exposed skin on their hands or face, thereby quantifying the potential impact of environmental factors on physiological indicators. Non-contact physiological sensors can employ sensors based on radar, infrared thermal imaging, laser Doppler, or photoplethysmography (PPG) principles. For example, a small camera or dedicated sensor array integrated into or above the safe panel can accurately measure heart rate, respiratory rate, and skin temperature without contact with the operator. Physiological and environmental correlated data packets encapsulate local microenvironmental data and physiological data collected within the same time window to ensure data temporal synchronization and correlation, providing a basis for subsequent environmental effect compensation.

[0021] The Physiological-Environment Interaction Logic Module is a core component embedded in the local processing unit. This module contains a set of complex mathematical models or algorithms based on extensive physiological research and experimental data. These models quantify how different environmental parameters (such as temperature and airflow) affect physiological responses such as heart rate, respiration, and skin temperature. For example, when the ambient temperature rises, skin temperature naturally increases, and heart rate and respiratory rate may also change accordingly. This module can distinguish between these environmentally induced physiological changes and those caused by psychological stress or external threats. Environmental effect compensation refers to using the Physiological-Environment Interaction Logic Module to remove or correct the portion of physiological data caused by environmental factors, thereby obtaining purified physiological data that more closely approximates the operator's actual physiological state. For example, if an increase in ambient temperature causes a rise in skin temperature of X degrees, the compensation process subtracts X degrees from the measured skin temperature to eliminate the environmental impact. The personalized physiological response reference set refers to the baseline range of purified physiological data after environmental effect compensation, pre-established for each operator under different situations (such as normal operation, mild stress, fatigue, etc.). This set is dynamically updated and learned, and can reflect the individual differences and physiological fluctuations of operators.

[0022] This application deploys a miniature environmental sensor array in the operating area of ​​a safe panel to acquire real-time local microenvironmental data around the operator's exposed skin. Simultaneously, non-contact physiological sensors collect physiological data such as heart rate, respiratory rhythm, and skin temperature. This data is then integrated into a physiological-environment correlated data package. Crucially, a physiological-environment interaction logic module is embedded in the local processing unit. This module accurately describes the impact of specific microenvironmental changes on physiological indicators based on human physiological principles. Therefore, using the microenvironmental data in the correlated data package, this logic module compensates for environmental effects on the original physiological data, generating purified physiological data unaffected by environmental interference. Finally, this purified physiological data is compared with a pre-stored set of personalized physiological responses for operators in different situations. This allows for a more accurate assessment of deviations, significantly improving the accuracy of judging the operator's true physiological state and avoiding interference from environmental factors.

[0023] Suppose an operator is working on a safe. At this moment, a miniature environmental sensor array in the safe's operating area detects a sudden 2-degree Celsius increase in the local temperature around the operator's hands. Simultaneously, non-contact physiological sensors collect data showing increases in the operator's heart rate, respiratory rate, and skin temperature. Without environmental effect compensation, the system might misjudge the operator as being under stress or threat. However, in this application's solution, the physiological-environment interaction logic module calculates the normal physiological effects of a 2-degree Celsius increase in ambient temperature on heart rate, respiration, and skin temperature based on a preset set of functions. For example, this module might determine that a 2-degree Celsius increase in ambient temperature would lead to a natural 1.5-degree Celsius increase in skin temperature. Subsequently, the system subtracts this 1.5-degree Celsius environmental effect from the collected skin temperature data to obtain purified physiological data. This purified physiological data is then compared with the operator's personalized physiological response reference set under normal operating conditions. If the comparison result shows that the purified physiological data is still within the normal range, the system can accurately determine that the operator is not under external threat, thus avoiding misjudgments caused by environmental changes.

[0024] In some embodiments, if it is determined that the operator is not under external threat, the steps of distinguishing whether the abnormal operation information is unintentional or malicious based on the operator's continuous operation behavior trajectory information, and taking corresponding guidance or rejection measures, include: If it is determined that the operator is not under external threat, the interaction behavior pattern of the operator with the unauthorized storage cell within the preset time window is analyzed to distinguish whether the abnormal operation information is unintentional or malicious. If the abnormal operation information is determined to be unintentional, then a guiding prompt will be provided; If the abnormal operation information is determined to be a malicious attempt, access will be denied and an alarm will be triggered.

[0025] Specifically, when the system determines that the operator is not under external threat, it further analyzes the operator's interaction patterns with unauthorized storage cells within a preset time window. The preset time window is a configurable duration, such as 5 seconds, 10 seconds, or longer, during which the system continuously monitors and records the operator's interactions with the safe. By collecting continuous interaction data within this time window, the system can capture the sequence, frequency, and persistence of behaviors, rather than just isolated events. Unauthorized storage cells refer to any internal storage area or compartment of the safe that the operator is not allowed to access based on their current task permissions. Interaction patterns encompass all physical or proximity interactions between the operator and these unauthorized storage cells, such as the frequency, duration, intensity, and geographical and temporal proximity to authorized operations, including actions like hand touches, hovering, and attempts to open.

[0026] Based on in-depth analysis of these interaction patterns, the system can distinguish between unintentional errors and malicious attempts. Unintentional errors typically manifest as brief, accidental, and non-repetitive interactions, such as an operator accidentally touching an adjacent unauthorized storage cell while performing an authorization task, and then immediately correcting their behavior. Malicious attempts, on the other hand, often involve continuous, repetitive, and purposeful probing or attempts, such as an operator repeatedly touching or attempting to open unauthorized storage cells, or lingering in unauthorized areas for extended periods, with behavioral patterns clearly inconsistent with normal authorization procedures.

[0027] If the system determines that the abnormal operation was unintentional, it will provide a guiding prompt. This prompt is non-intrusive and aims to gently correct the operator's behavior. For example, it might display a brief text message on the interface (e.g., "Please note that this area is outside your current permissions"), emit a soft sound, or provide a visual indicator to guide the operator back to the correct procedure. This approach avoids unnecessary alerts while helping operators correct errors promptly, thus improving the user experience.

[0028] If the system determines that the abnormal operation information is a malicious attempt, it will immediately take more stringent measures, namely, denying access and triggering an alarm. Denying access means that the system will proactively prevent operators from further performing operations on the unauthorized storage cell, such as temporarily locking the access mechanism of the storage cell or disabling its user interface. Simultaneously, triggering an alarm means that the system will silently send an alert message to the security monitoring center or designated security personnel without any audible or visual cue on-site, thereby avoiding alerting potential malicious actors and buying time for security personnel to intervene.

[0029] This application addresses the ambiguity in judging abnormal operational intentions in basic solutions by introducing refined analysis of operator interaction patterns within a preset time window. By capturing the continuity and pattern characteristics of behavior, the system can more accurately identify whether an operator's actions are due to negligence or malicious intent. For example, a brief, single touch event might be categorized as unintentional, while a series of repetitive, continuous touches or attempts to open the device would be identified as malicious intent. This behavior pattern-based judgment mechanism allows the system to take differentiated response measures based on the actual level of risk. When an unintentional touch is identified, guiding prompts can promptly correct the operator's error, avoiding misjudgments and unnecessary security responses. When a malicious intent is identified, access denial and silent alarms can effectively prevent potential threats and provide valuable time for security personnel to intervene, thus effectively curbing malicious behavior without exposing the system's defense mechanisms.

[0030] In some embodiments, the steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the task permissions based on the interaction information and preset task permissions, include: Acquire video streams of the operator's hand movements; Image processing is performed on video streams of hand movements to identify the position and movement trajectory of the operator's hands, as well as their contact or approach to the various storage compartments of the safe; Miniature contact sensors or proximity sensors are deployed on the door handles or operating areas of each storage compartment of the safe to detect the physical interaction between the operator and the specific storage compartment in real time. The hand motion video stream data is synchronized and fused with touch or proximity sensor data in time to obtain comprehensive interactive information; Establish a behavior context analysis module, which contains logical rules based on preset operation processes and behavior characteristics; When an operator's hand is detected touching or approaching an unauthorized memory cell, the duration, force intensity, repetition frequency, geographical proximity to the currently authorized operation, and temporal proximity of the touching behavior are evaluated based on comprehensive interaction information. Based on the evaluation results, the behavioral context analysis module determines whether the touch behavior is an accidental behavior that occurs along with the authorized operation or a purposeful exploration or attempt. Intentional probing or exploratory behavior will be identified as abnormal operation information that is inconsistent with the task's permissions.

[0031] The acquisition of video streams depicting the operator's hand movements aims to capture these movements within the safe's operating area using visual sensors. This video stream can be captured by a camera deployed above the safe's operating area to ensure the clarity and completeness of the hand movements. Further, image processing is performed on the hand movement video stream to extract key hand behavior features from the raw video data. Specifically, image processing techniques may include, but are not limited to, algorithms such as object detection, motion tracking, and pose estimation to accurately identify the position of the operator's hands in three-dimensional space, their movement trajectory, and the precise contact or proximity of the hands with various storage compartments of the safe (e.g., drawers, shelves, or doors).

[0032] The deployment of miniature contact sensors or proximity sensors on the door handles or operating areas of each storage compartment in the safe aims to provide a complementary and highly accurate means of detecting physical interactions. Contact sensors detect actual physical contact, while proximity sensors emit signals when a hand approaches but does not actually make contact, thus enabling real-time and accurate detection of physical interactions between the operator and a specific storage compartment. Therefore, synchronizing and fusing hand movement video stream data with contact or proximity sensor data in time aims to integrate data from different types of sensors, forming a more comprehensive and reliable integrated interaction information. Time synchronization ensures the consistency of data from different sensors along the timeline, while data fusion can compensate for potential blind spots or errors in single sensors, improving the accuracy of interaction information recognition.

[0033] A behavioral context analysis module is established, containing logical rules based on preset operation procedures and behavioral characteristics. This module can be understood as an intelligent decision-making system, internally pre-setting legal operation procedures for different task permissions (e.g., authorized personnel can only open specific storage cells) and characteristic patterns of abnormal behaviors. These logical rules are established based on a large amount of historical operation data and expert experience, used for in-depth analysis of operator interaction behavior. Specifically, when it is detected that an operator's hand touches or approaches an unauthorized storage cell, the system will evaluate the touch behavior from multiple dimensions based on comprehensive interaction information. The evaluation parameters include the duration of the touch behavior (e.g., a brief touch or a long stay), the intensity of the touch (e.g., a slight touch or a forceful attempt to open), the repetition frequency (e.g., a single touch or repeated attempts), the geographical proximity to the current authorized operation (e.g., whether the touched storage cell is adjacent to the authorized operation's storage cell), and the temporal proximity (e.g., whether the touch behavior occurs within a close time window of the authorized operation). The combination of these parameters can provide rich clues to the behavioral intent. Based on the above assessment results, the behavioral context analysis module determines whether the touching behavior is an accidental action accompanying authorized operations or a purposeful probe or attempt. For example, if an operator accidentally touches an unauthorized storage cell while opening an authorized storage cell, and the touch is brief and light, it may be judged as an accidental action. Conversely, if the operator repeatedly touches an unauthorized storage cell for a long period of time with strong force, it may be judged as a purposeful probe or attempt. Ultimately, purposeful probe or attempt behaviors are identified as abnormal operation information inconsistent with task permissions. This identification result will serve as the basis for subsequent judgments on whether the operator is under external threat or to distinguish between unintentional mistakes and malicious attempts.

[0034] This application effectively improves the accuracy of abnormal operation information recognition by combining multi-source data acquisition and intelligent analysis. First, by acquiring video streams of the operator's hand movements, the overall hand activity can be captured macroscopically, and image processing technology can be used to identify the hand's position, trajectory, and contact or proximity to the safe's storage compartments, thus providing visual behavioral evidence. Second, deploying miniature touch sensors or proximity sensors in each storage compartment provides high-precision physical interaction detection, compensating for potential blind spots or insufficient accuracy in visual recognition. Synchronizing and fusing these two types of data in time forms a more comprehensive and accurate integrated interaction information, avoiding misjudgments that may occur with a single data source. Based on this, the behavior context analysis module uses preset logical rules to perform multi-dimensional evaluations of touching unauthorized storage compartments, such as duration, force intensity, and repetition frequency, and combines this with geographical and temporal proximity to the currently authorized operation, thereby distinguishing between accidental unintentional touches and purposeful probing or attempts. It is precisely because of this multi-dimensional, context-aware analysis mechanism that the system can more accurately determine the operator's true intentions, thereby identifying purposeful probing or exploratory behavior as abnormal operation information that is inconsistent with the task's permissions.

[0035] In some embodiments, the steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the task permissions based on the interaction information and preset task permissions, include: Multiple visual sensors are deployed in the safe's operating area to acquire the overall video stream of the operating area; Using the overall video stream, the identity information and spatial location of all personnel within the operation area are identified; Miniature contact sensors or proximity sensors are deployed on the door handles or operating areas of each storage compartment of the safe to detect physical interactions between people and specific storage compartments in real time. By synchronizing and fusing the identity information, spatial location, and physical interaction between the person and a specific storage cell in time, multi-person interaction information is obtained; Establish a personnel behavior attribution module, which includes logical rules based on personnel identity, role permissions, and spatial location; When it is detected that someone's hand is touching or approaching an unauthorized storage cell, the touch behavior is accurately attributed to a specific operator or non-operator based on the multi-person interaction information and the personnel behavior attribution module. If the touching behavior is attributed to a non-operator, it is judged as unintentional touching and not identified as abnormal operation; If the touch action belongs to the operator, then determine whether the touch action is inconsistent with the task permissions based on the operator's role permissions; Touch actions that are inconsistent with task permissions will be identified as abnormal operation information that is inconsistent with task permissions.

[0036] Specifically, multiple visual sensors, such as high-resolution cameras, are deployed in the safe's operating area to comprehensively cover the area and capture the activities of all personnel within it. By performing image processing and computer vision analysis on the overall video stream acquired by these visual sensors, the identity information of all personnel in the operating area can be identified, for example, through facial recognition or gait recognition technology, and their spatial location and movement trajectory can be tracked in real time.

[0037] Miniature contact sensors or proximity sensors are deployed on the door handles or operating areas of each storage compartment in the safe. These sensors can detect any physical interaction between a person and a specific storage compartment in real time, such as hand touch or approach behavior. Contact sensors can detect actual physical contact, while proximity sensors can detect the approach of an object without contact.

[0038] Personnel identification information and spatial location data acquired through visual sensors are synchronized and fused with physical interaction data detected through touch or proximity sensors. The purpose of this multi-source data fusion is to construct a comprehensive multi-person interaction information, ensuring that all relevant data are aligned on the timeline for subsequent accurate analysis.

[0039] Establish a module for attributing user behavior. This module contains embedded logical rules based on user identity, role permissions, and spatial location. These rules define the scope of operations that users with different identities and permissions can perform in a specific context, as well as the correlation between their spatial location and interactive behavior. For example, if an unauthorized user lingers near a storage cell for an extended period and engages in touch behavior, the module will make a judgment based on their identity and permissions.

[0040] When the system detects any person's hand touching or approaching an unauthorized storage cell, the personnel behavior attribution module uses the integrated multi-person interaction information to accurately attribute the touch to a specific operator or non-operator. For example, it compares which identified person's hand was in the touch area when the touch occurred, and combines this with their spatial location information to make a determination.

[0041] If the touch is attributed to a non-operator, the system will classify it as an unintentional touch and will not identify it as abnormal operation information. This effectively avoids false alarms caused by accidental actions of non-operators.

[0042] If the touch action is attributed to an operator, the system will determine whether the touch action is inconsistent with the current task permissions based on the operator's role and permissions. For example, if an operator is authorized to access storage cell A, but their hand touches or approaches an unauthorized storage cell B, this action will be considered inconsistent with task permissions.

[0043] The aforementioned touch behaviors that are inconsistent with the task permissions will be identified as abnormal operation information that is inconsistent with the task permissions, and will be used as the basis for subsequent judgment on whether the operator is under external threat.

[0044] This application effectively solves the problem of accurately identifying the attribution of abnormal operations in complex environments with multiple personnel by introducing multiple visual sensors and micro-touch or proximity sensors, combined with a personnel behavior attribution module. Specifically, multiple visual sensors provide a global view of the operating area, enabling the identification and spatial location of all personnel present, thus constructing a macro-context for personnel behavior. Simultaneously, micro-touch or proximity sensors deployed on the storage compartments provide micro-level, precise data on direct interaction with the safe. By synchronizing and fusing this macro- and micro-level data over time, the system obtains comprehensive multi-person interaction information. Based on this, the personnel behavior attribution module, using preset logical rules, can accurately attribute any touching or approaching of an unauthorized storage compartment to a specific individual. This attribution mechanism allows the system to distinguish between abnormal behavior by authorized personnel and unintentional touches by unauthorized personnel, thereby avoiding false alarms and ensuring accurate identification of genuine abnormal behavior.

[0045] Imagine an authorized operator performing routine maintenance on a safe in a bank vault, while an unauthorized security guard patrols the vicinity.

[0046] First, multiple visual sensors deployed in the safe's operating area continuously acquire the overall video stream, identifying authorized operators and security personnel and their respective spatial locations. Simultaneously, miniature contact sensors or proximity sensors deployed on the door handles of each safe compartment monitor any physical interaction between personnel and the compartments in real time.

[0047] At one point, a security guard accidentally touches a storage cell that an authorized operator is not authorized to access while on patrol. The system then uses visual sensor data to identify the guard's hand movement and spatial location, and touch sensors to detect the physical interaction with the storage cell. This information is synchronized and integrated into a multi-person interaction report.

[0048] Subsequently, the personnel behavior attribution module determines whether the touch action belongs to a non-operating personnel based on the security personnel's identity (non-operating personnel) and their role permissions. According to preset logic rules, the system will determine that the action is unintentional and therefore will not identify it as abnormal operation information, avoiding unnecessary alarms.

[0049] However, if an authorized operator intentionally or unintentionally touches or approaches a storage cell outside their authorized task permissions while performing their authorized task, the personnel behavior attribution module will identify that the touch behavior belongs to the operator. The system will then determine whether the touch behavior is inconsistent with the operator's current task permissions. If it is inconsistent, the behavior will be identified as an abnormal operation that is inconsistent with task permissions, triggering subsequent threat assessment procedures.

[0050] In this way, the solution proposed in this application can accurately distinguish the interactive behaviors of different personnel in complex environments where multiple people coexist, and accurately identify real abnormal operations, thereby improving the intelligence and security of safe access control.

[0051] In some embodiments, the steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the task permissions based on the interaction information and preset task permissions, include: The system acquires interaction information between the operator's hands and the various storage compartments of the safe. This interaction information includes a sequence of events where the hand touches or approaches each compartment, along with a timestamp for each event. Specifically, the interaction information refers to the physical contact or spatial proximity between the operator's hands and the various storage compartments of the safe during interaction. These actions are recorded as a series of discrete events, each accompanied by a precise timestamp, to construct a complete time sequence of operational behavior. For example, actions such as the operator's hand touching the doorknob of a storage compartment or approaching the panel of a storage compartment are captured and recorded.

[0052] An operation behavior sequence analysis unit is established, which contains a set of pre-defined behavior pattern rules related to task permissions. These rules define the legal sequences of authorized operations and the combinations of unauthorized operations. The operation behavior sequence analysis unit is a module specifically designed to process and analyze operation behavior sequences, its core being the pre-defined series of behavior pattern rules. These rules not only clarify which operation sequences are legal and permitted under specific task permissions (e.g., opening storage cell A -> retrieving an item -> closing storage cell A), but also define which operation sequences or combinations are inconsistent with authorized tasks (e.g., attempting to access storage cell C immediately after unauthorized access to storage cell B). These rules are established based on a deep understanding of legal operation flows and the prediction of potential abnormal behavior patterns.

[0053] The event sequence is input into the operation behavior sequence analysis unit. Specifically, the interaction information between the operator's hands and the various storage compartments of the safe, acquired through sensors or other data acquisition devices, i.e., the event sequence, is transmitted to the operation behavior sequence analysis unit for processing in real time or near real time.

[0054] The operation behavior sequence analysis unit performs temporal analysis on the event sequence based on behavior pattern rules to identify whether there are combined behavior patterns that are inconsistent with task permissions. Temporal analysis refers to an in-depth analysis of the time sequence, duration, intervals between events, and combinations of events in the event sequence. By comparing the actual event sequence with preset behavior pattern rules, the operation behavior sequence analysis unit can identify combined behavior patterns that do not conform to authorized operation procedures or demonstrate unauthorized intent. For example, if an operator, after being authorized to access storage cell A, repeatedly touches or lingers in the unauthorized storage cell B within a short period of time, this behavior sequence may be identified as a combined behavior pattern that is inconsistent with task permissions.

[0055] Combination behavior patterns that do not conform to task permissions are identified as abnormal operation information that does not conform to task permissions. Once the operation behavior sequence analysis unit identifies any combination behavior patterns that contradict the preset behavior pattern rules through time sequence analysis, these patterns are confirmed as abnormal operation information and used as the basis for subsequent judgment of the operator's status.

[0056] This application, by introducing an operation behavior sequence analysis unit and behavior pattern rules, enables a more refined and in-depth analysis of the interaction between the operator and the safe. While basic solutions may focus only on whether a single action exceeds authority, this solution focuses on the combination and temporal relationship of a series of actions. Specifically, by acquiring the event sequence and timestamps of the operator's hands interacting with each storage cell of the safe, the system can construct a complete behavioral trajectory of the operator. The operation behavior sequence analysis unit uses preset behavior pattern rules to perform temporal analysis on these event sequences. These rules not only define the standard procedures for legitimate operations but also cover various combinations of unauthorized operations. For example, a single touch of an unauthorized storage cell might be considered unintentional, but if this touch is followed by a series of rapid, repetitive probing actions, or operations logically unrelated to the current authorized task, temporal analysis can accurately identify these combined behavioral patterns as potential abnormal operations. This sequence- and pattern-based analysis method allows the system to capture more complex and subtle abnormal behaviors, not just simple unauthorized actions, thereby improving the accuracy and robustness of abnormal operation identification.

[0057] In some embodiments, the steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the task permissions based on the interaction information and preset task permissions, include: Acquire the interaction information between the operator's hands and each storage cell of the safe. The interaction information includes the event sequence of the hand touching or approaching each storage cell and the timestamp of each event. For each interactive action event in the event sequence, extract its behavioral features, including duration, intensity, repetition frequency, geographical proximity to the currently authorized operation, and temporal proximity. Behavioral features are combined to form a multi-dimensional behavioral feature vector; Establish a behavioral deviation assessment unit, which includes a set of preset behavioral deviation thresholds based on the operator's historical behavioral data in different situations; Input multi-dimensional behavioral feature vectors into the behavioral deviation assessment unit; The behavioral deviation assessment unit comprehensively evaluates multi-dimensional behavioral feature vectors based on a set of behavioral deviation thresholds. When a combination of multiple minor deviations is identified, and the combination of multiple minor deviations collectively indicates a potential risk; The combination of multiple minor deviations is identified as abnormal operation information that does not conform to the task permissions.

[0058] Specifically, interaction action information refers to the record of physical or close-range contact events between an operator's hand and the various storage compartments of the safe. These records are presented as event sequences, with each event accompanied by a precise timestamp for time-series analysis. For example, when an operator's hand touches or approaches a storage compartment, the type of event, the time of occurrence, and the identifier of the storage compartment involved are recorded.

[0059] Behavioral characteristics refer to the quantitative indicators extracted from each interactive action event, used to describe the characteristics of that event. Duration can be understood as the length of time a hand touches or approaches a storage cell; force intensity refers to the pressure or intensity of the touch, which can be estimated through tactile sensors or image analysis; repetition frequency refers to the number of times the same or related storage cells are touched or approached within a specific time period; geographical proximity refers to the physical distance between the storage cell of the current interactive action and the storage cells currently authorized for access by the operator; temporal proximity refers to the time interval between the current interactive action and the most recent authorized operation. The extraction of these behavioral characteristics aims to characterize the subtle behavioral patterns of operators from multiple dimensions.

[0060] A multi-dimensional behavioral feature vector integrates the extracted behavioral features into a unified mathematical representation to facilitate subsequent quantitative analysis and evaluation. For example, values ​​such as duration, intensity, repetition frequency, geographical proximity, and temporal proximity can be combined into a vector.

[0061] The behavioral deviation assessment unit can be understood as an intelligent analysis module, its core being a set of preset behavioral deviation thresholds. This threshold set is established based on learning and modeling historical behavioral data of specific operators under different situations. For example, by observing operators' operating habits over a long period under different situations such as normal, stressful, and fatigued conditions, normal fluctuation ranges and abnormal thresholds for various behavioral characteristics under these situations can be established. When multi-dimensional behavioral feature vectors are input into the behavioral deviation assessment unit, the unit uses its internal logical rules and algorithms to compare the input vectors with the preset threshold set. The purpose is to identify combinations of features that may not be significant individually, but where multiple features deviate from the normal range. Such combined deviations are considered indicators of potential risk; even if a single deviation is insufficient to trigger an alarm, its cumulative effect may reveal anomalies.

[0062] When the behavioral deviation assessment unit identifies a combination of multiple minor deviations that collectively indicate a potential risk, it identifies this as abnormal operational information inconsistent with task permissions. This identification method can capture more subtle and complex abnormal behavioral patterns. For example, an operator may lightly touch an unauthorized storage cell multiple times in a short period of time. Each touch is very short in duration and light in force, and may be ignored individually, but the combination of these behaviors may indicate that the operator is making a exploratory attempt.

[0063] This application addresses the limitations in identifying complex and concealed abnormal operations by introducing multi-dimensional behavioral feature vectors and a behavioral deviation assessment unit. Basic solutions may focus only on single, explicit violations, neglecting subtle, cumulative deviations within behavioral patterns. This solution first acquires information on the interaction between the operator's hands and the various storage compartments of the safe, extracting multi-dimensional behavioral features from the event sequence, including duration, intensity, repetition frequency, geographical proximity, and temporal proximity. These features are combined into a multi-dimensional behavioral feature vector, comprehensively characterizing the operator's subtle operational habits. Subsequently, the behavioral deviation assessment unit uses a set of behavioral deviation thresholds established based on historical behavioral data to comprehensively evaluate these multi-dimensional feature vectors. The core advantage is that even if the deviation of a single behavioral feature is insufficient to trigger an alarm, this unit can identify combinations of multiple subtle deviations. It is precisely this sensitivity to combined biases that enables this scheme to capture more subtle and complex abnormal behavior patterns that may indicate potential risks or malicious intent. For example, an operator may lightly touch an unauthorized storage cell multiple times in a short period of time. Each touch is very short and light, and may be ignored on its own, but the combination of these behaviors may indicate that the operator is making a exploratory attempt.

[0064] Suppose an operator is authorized to access storage cells A and B in a safe. During an access session, the system continuously monitors the operator's hand gestures.

[0065] First, the system acquires information about the interaction between the operator's hands and the various storage compartments of the safe. For example, it records that after accessing storage compartment A, the operator's hand lightly touched the door handle of the unauthorized storage compartment C three times in a short period of time (e.g., within 5 seconds). The duration of each touch was less than 0.5 seconds, and the intensity of the touch was lower than the average force applied when normally operating storage compartments A or B.

[0066] Next, the system extracts behavioral features from these three touch events. For each touch, it extracts the duration (e.g., 0.3 seconds, 0.4 seconds, 0.35 seconds), the intensity of the touch (e.g., slight), the frequency of repetition (3 times within 5 seconds), the geographical proximity to the current authorized operation (accessing cell A) (cell C is farther from cell A), and the temporal proximity (touch C occurs shortly after accessing A).

[0067] These behavioral features are then combined into a multi-dimensional behavioral feature vector. For example, the vector might contain [0.3s, slight, 3 times / 5s, far, short time interval].

[0068] The behavioral deviation assessment unit receives these vectors. The preset set of thresholds within this unit may indicate that a single slight touch of 0.3 seconds is not enough to constitute an anomaly, but when three slight touches to unauthorized cells occur in a very short period of time, and the geographical and temporal proximity to authorized operations does not match, this combination of patterns will significantly deviate from the operator's historical normal behavioral patterns.

[0069] Ultimately, the behavioral deviation assessment unit, based on its threshold set, comprehensively evaluates and concludes that the combination of these multiple minor deviations collectively indicates a potential risk, such as the operator possibly engaging in exploratory activities. Therefore, the system identifies this combined behavior as abnormal operational information inconsistent with task permissions and triggers subsequent processes for acquiring and comparing physical status.

[0070] In some embodiments, the step of deploying multiple visual sensors in the safe operation area to acquire the overall video stream of the operation area includes: At least three visual sensors are deployed at the top, sides, and diagonal of the main activity area of ​​the operator in the safe operation area. At least three vision sensors with high resolution and wide dynamic range are configured with wide-angle lenses; At least three visual sensors are equipped with automatic exposure and white balance functions to obtain a complete video stream of the operating area.

[0071] Specifically, visual sensors can be understood as devices capable of capturing image or video data, such as, but not limited to, high-definition cameras, infrared cameras, or sensors with depth sensing capabilities. At least three visual sensors are deployed at the top, sides, and diagonally across the operator's main activity area within the safe's operating area. This aims to eliminate blind spots through multi-angle and multi-directional coverage, ensuring that all critical activities within the operating area, particularly the operator's interactions with the safe, are captured completely and clearly. The diagonal positions within the operator's main activity area refer to areas that maximize coverage of the operator's hands, body posture, and interactions with the safe's storage compartments. Diagonal deployment effectively reduces obstruction and provides a more comprehensive field of view.

[0072] At least three vision sensors are required to have high resolution and wide dynamic range, and to be equipped with wide-angle lenses. High resolution ensures the clarity of image details, enabling detailed analysis of operator hand movements, facial expressions (if needed), and precise interactions with the storage cells. Wide dynamic range (WDR) allows the sensors to capture clear images even in environments with significant differences in lighting conditions (e.g., areas that are too bright or too dark), avoiding the loss of critical information due to lighting issues. Wide-angle lenses expand the field of view of a single sensor, reducing the number of sensors required while ensuring comprehensive coverage of the entire operating area.

[0073] At least three vision sensors also feature automatic exposure and white balance. The automatic exposure function adjusts exposure parameters automatically based on changes in ambient light, ensuring the video stream is appropriately bright and avoiding overexposure or underexposure. The white balance function corrects color deviations in the image, resulting in realistic and natural colors in the video stream, which helps in accurately identifying objects and environmental features. These functions work together to acquire a high-quality, distortion-free overall video stream that accurately reflects the operating area.

[0074] This application addresses the challenge of acquiring comprehensive, high-quality video streams in complex lighting and multi-angle interactive environments by optimizing the deployment strategy and technical characteristics of visual sensors. Specifically, at least three visual sensors are deployed at the top, sides, and diagonally across the operator's main activity area of ​​the safe's operating area. This multi-sensor collaborative approach forms a three-dimensional monitoring network, effectively eliminating blind spots and obstructions that may exist from a single viewpoint, ensuring that operator interactions from any position are captured. Furthermore, by requiring visual sensors with high resolution, wide dynamic range, and wide-angle lenses, the clarity, detail richness, and coverage of the acquired video stream are guaranteed, providing stable and reliable image data even in uneven lighting or fast-moving scenarios. Automatic exposure and white balance functions further enhance the image quality of the video stream, maintaining accurate color and appropriate brightness under various environmental conditions. This provides high-quality input data for subsequent image processing, behavior recognition, and abnormal operation assessment, thereby improving the accuracy and robustness of the entire access control system in identifying abnormal behavior.

[0075] like Figure 2 As shown in the embodiments, this application also discloses a multifunctional safe identification and access control system, including: Monitoring startup module 1 is used to start continuous monitoring of the operator after the operator completes identity verification and establishes an access session; The interaction action acquisition and recognition module 2 is used to acquire the interaction action information between the operator and the safe during continuous monitoring, and to identify abnormal operation information that does not conform to the task permissions based on the interaction action information and the preset task permissions. The body status acquisition and comparison module 3 is used to acquire the operator's body status information non-contactly in response to the identification of abnormal operation information, and compare the body status information with the pre-established range of the operator's personal physiological response in different situations; The external threat assessment and alarm execution module 4 is used to determine whether the operator is under external threat based on the comparison results; if the operator is determined to be under external threat, an alarm message is sent silently and a deceptive operation is performed. The Intent Differentiation and Measure Execution Module 5 is used to, if it is determined that the operator is not under external threat, distinguish whether the abnormal operation information is unintentional or malicious, based on the operator's continuous operation behavior trajectory information, and take corresponding guidance or rejection measures.

[0076] The system provided in this application can effectively address the shortcomings of traditional systems in dealing with external threats and internal malicious attempts, and realizes real-time and dynamic management of operator behavior and status, significantly improving the overall security and intelligence level of safe access control.

[0077] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A multifunctional safe identification and access control method, characterized in that, include: After the operator completes identity verification and establishes an access session, continuous monitoring of the operator is initiated. During continuous monitoring, information on the interaction between the operator and the safe is acquired, and abnormal operation information that does not conform to the preset task permissions is identified based on the interaction information and the preset task permissions. In response to the detection of abnormal operation information, the system obtains the operator's physical status information non-contactly and compares the physical status information with the pre-established range of the operator's personal physiological response in different situations; Based on the comparison results, determine whether the operator is under external threat; If it is determined that the operator is under external threat, an alarm message is sent silently and a deceptive operation is performed. If it is determined that the operator is not under external threat, then based on the operator's continuous operational behavior trajectory information, it is determined whether the abnormal operation information is unintentional or malicious, and corresponding guidance or rejection measures are taken.

2. The multifunctional safe identification and access control method according to claim 1, characterized in that, The steps of responding to the detection of abnormal operation information, non-contactly acquiring the operator's physical state information, and comparing the physical state information with a pre-established range of the operator's personal physiological responses in different situations include: A group of miniature environmental sensors is deployed in the operating area of ​​the safe panel to measure the local temperature gradient around the operator's exposed skin and the airflow velocity and its minute fluctuations in the operating area in real time, so as to obtain local microenvironment data. Simultaneously activate the miniature environmental sensor array and the non-contact physiological sensor. The non-contact physiological sensor is used to collect the operator's heart rate, breathing rhythm and skin temperature in real time to obtain physiological data. Based on local microenvironment data and physiological data collection, a physiological and environmental data package is formed; The local processing unit contains an embedded logic module for the interaction between physiology and the environment; By utilizing the microenvironmental data in the physiological and environmental data package, and through the physiological and environmental interaction logic module, environmental effect compensation is performed on the physiological data to generate purified physiological data. The purified physiological data were compared with a pre-stored set of personalized physiological responses of operators in different situations to assess the degree of deviation.

3. The multifunctional safe identification and access control method according to claim 1, characterized in that, If it is determined that the operator is not under external threat, the steps to distinguish whether the abnormal operation information is unintentional or malicious, based on the operator's continuous operation behavior trajectory information, and to take corresponding guidance or rejection measures include: If it is determined that the operator is not under external threat, the interaction behavior pattern of the operator with the unauthorized storage cell within the preset time window is analyzed to distinguish whether the abnormal operation information is unintentional or malicious. If the abnormal operation information is determined to be unintentional, then a guiding prompt will be provided; If the abnormal operation information is determined to be a malicious attempt, access will be denied and an alarm will be triggered.

4. The multifunctional safe identification and access control method according to claim 1, characterized in that, The steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the preset task permissions based on the interaction information and the preset task permissions, include: Acquire video streams of the operator's hand movements; Image processing is performed on video streams of hand movements to identify the position and movement trajectory of the operator's hands, as well as their contact or approach to the various storage compartments of the safe; Miniature contact sensors or proximity sensors are deployed on the door handles or operating areas of each storage compartment of the safe to detect the physical interaction between the operator and the specific storage compartment in real time. The hand motion video stream data is synchronized and fused with touch or proximity sensor data in time to obtain comprehensive interactive information; Establish a behavior context analysis module, which contains logical rules based on preset operation processes and behavior characteristics; When an operator's hand is detected touching or approaching an unauthorized memory cell, the duration, force intensity, repetition frequency, geographical proximity to the currently authorized operation, and temporal proximity of the touching behavior are evaluated based on comprehensive interaction information. Based on the evaluation results, the behavioral context analysis module determines whether the touch behavior is an accidental behavior that occurs along with the authorized operation or a purposeful probe or attempt. Intentional probing or exploratory behavior will be identified as abnormal operation information that is inconsistent with the task's permissions.

5. The multifunctional safe identification and access control method according to claim 1, characterized in that, The steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the preset task permissions based on the interaction information and the preset task permissions, include: Multiple visual sensors are deployed in the safe's operating area to acquire the overall video stream of the operating area; By utilizing the overall video stream, the identity information and spatial location of all personnel within the operating area can be identified; Miniature contact sensors or proximity sensors are deployed on the door handles or operating areas of each storage compartment of the safe to detect physical interactions between people and specific storage compartments in real time. By synchronizing and fusing the identity information, spatial location, and physical interaction between the person and a specific storage cell in time, multi-person interaction information is obtained; Establish a personnel behavior attribution module, which includes logical rules based on personnel identity, role permissions, and spatial location; When it is detected that someone's hand is touching or approaching an unauthorized storage cell, the touch behavior is accurately attributed to a specific operator or non-operator based on the multi-person interaction information and the personnel behavior attribution module. If the touching behavior is attributed to a non-operator, it is judged as unintentional touching and not identified as abnormal operation; If the touch action belongs to the operator, then determine whether the touch action is inconsistent with the task permissions based on the operator's role permissions; Touch actions that are inconsistent with task permissions will be identified as abnormal operation information that is inconsistent with task permissions.

6. The multifunctional safe identification and access control method according to claim 1, characterized in that, The steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the preset task permissions based on the interaction information and the preset task permissions, include: Acquire the interaction information between the operator's hands and each storage cell of the safe. The interaction information includes the event sequence of the hand touching or approaching each storage cell and the timestamp of each event. An operation behavior sequence analysis unit is established, which contains a set of preset behavior pattern rules related to task permissions. The behavior pattern rules define the legal sequence of authorized operations and the combination pattern of unauthorized operations. Input the event sequence into the operation behavior sequence analysis unit; The operation behavior sequence analysis unit performs time-series analysis on the event sequence based on behavior pattern rules to identify whether there are combined behavior patterns that are inconsistent with task permissions; Combined behavior patterns that are inconsistent with task permissions are identified as abnormal operation information that is inconsistent with task permissions.

7. The multifunctional safe identification and access control method according to claim 1, characterized in that, The steps of acquiring interaction information between the operator and the safe during continuous monitoring, and identifying abnormal operation information that does not conform to the preset task permissions based on the interaction information and the preset task permissions, include: Acquire the interaction information between the operator's hands and each storage cell of the safe. The interaction information includes the event sequence of the hand touching or approaching each storage cell and the timestamp of each event. For each interactive action event in the event sequence, extract its behavioral features, including duration, intensity, repetition frequency, geographical proximity to the currently authorized operation, and temporal proximity. Behavioral features are combined to form a multi-dimensional behavioral feature vector; Establish a behavioral deviation assessment unit, which includes a set of preset behavioral deviation thresholds based on the operator's historical behavioral data in different situations; Input multi-dimensional behavioral feature vectors into the behavioral deviation assessment unit; The behavioral deviation assessment unit comprehensively evaluates multi-dimensional behavioral feature vectors based on a set of behavioral deviation thresholds. When a combination of multiple minor deviations is identified, and the combination of multiple minor deviations collectively indicates a potential risk; The combination of multiple minor deviations is identified as abnormal operation information that does not conform to the task permissions.

8. The multifunctional safe identification and access control method according to claim 5, characterized in that, The step of deploying multiple visual sensors in the safe's operating area to acquire the overall video stream of the operating area includes: At least three visual sensors are deployed at the top, sides, and diagonal of the main activity area of ​​the operator in the safe operation area. At least three vision sensors with high resolution and wide dynamic range are configured with wide-angle lenses; At least three visual sensors are equipped with automatic exposure and white balance functions to obtain a complete video stream of the operating area.

9. The multifunctional safe identification and access control method according to claim 2, characterized in that, The physiological-environment interaction logic module contains a set of functions based on human physiological principles, used to describe the effects of specific microenvironmental changes on physiological indicators such as heart rate, respiratory rhythm, and skin temperature.

10. A multifunctional safe identification and access control system, characterized in that, include: The monitoring startup module is used to start continuous monitoring of operators after they have completed identity verification and established an access session. The interaction action acquisition and recognition module is used to acquire interaction action information between the operator and the safe during continuous monitoring, and to identify abnormal operation information that does not conform to the task permissions based on the interaction action information and the preset task permissions. The body status acquisition and comparison module is used to acquire the operator's body status information non-contactly in response to the identification of abnormal operation information, and compare the body status information with the pre-established range of the operator's personal physiological response in different situations; The external threat assessment and alarm execution module is used to determine whether the operator is under external threat based on the comparison results. If it is determined that the operator is under external threat, an alarm message is sent silently and a deceptive operation is performed. The intent differentiation and action execution module is used to determine whether the abnormal operation information is unintentional or malicious if it is determined that the operator is not under external threat, based on the operator's continuous operation behavior trajectory information, and to take corresponding guidance or rejection measures.

Images