A biometric identification method and system for pre-employment health detection of drivers

Abstract

This invention relates to the technical field of biometric identification, and provides a biometric identification method and system for pre-employment health detection of drivers. The method includes: acquiring the driver's physiological state characteristics during identity verification; acquiring the driver's behavioral state characteristics; acquiring environmental context information and the driver's individual baseline characteristic records; adjusting the individualized judgment range information corresponding to the physiological state characteristics based on the environmental context information and the individual baseline characteristic records, generating adjusted individualized judgment range information; performing a comprehensive judgment on the physiological state characteristics and behavioral state characteristics based on the adjusted individualized judgment range information, generating a comprehensive health judgment result; and outputting a warning message based on the comprehensive health judgment result. This invention improves the accuracy of pre-employment health detection for drivers.

Description

Technical Field

[0001] This invention relates to the technical field of biometric identification, specifically to a biometric identification method and system for pre-employment health checks of drivers. Background Technology

[0002] With increasing societal demands for safety management in public transportation and the transport industry, ensuring drivers possess good health and accurate identity information before starting work has become a crucial aspect of preventing traffic accidents and safeguarding public safety. Traditional pre-employment driver testing methods have numerous shortcomings in identity verification and health assessment, making it difficult to meet the industry's current needs for both efficient and accurate safeguards.

[0003] Specifically, current mainstream technologies have the following limitations: First, the devices used for identity verification and those used for health checks are often independent, resulting in a fragmented testing process, long processing times, and an inability to effectively prevent substitution. For example, during height or vision checks, cheating can occur through substitution. Second, existing health monitoring methods largely rely on contact sensors, such as heart rate sensors and breathalyzers. These devices are susceptible to environmental factors and pose a risk of cross-infection. Third, single-modal biometric recognition technologies, such as facial recognition alone, experience a significant drop in accuracy when drivers are wearing masks or other obstructed clothing at close range. Furthermore, they struggle to simultaneously analyze the driver's physiological indicators, such as fatigue or emotional abnormalities. In addition, existing systems lack the ability to fuse data from multiple dimensions, making it impossible to comprehensively consider various information to assess the driver's condition.

[0004] In the context of pre-employment health checks for drivers, traditional technologies suffer from several drawbacks. These include independent identity verification and health assessment equipment, fragmented processes, long processing times, inability to effectively prevent proxy testing, reliance on contact sensors for health monitoring which is susceptible to environmental influences and cross-infection, insufficient accuracy of single-modal biometric recognition under close-range obstruction, and difficulty in simultaneously and non-contactly assessing hidden health issues such as driver fatigue and emotional abnormalities, as well as achieving comprehensive multi-dimensional information judgment.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses a biometric identification method and system for pre-employment health testing of drivers, aiming to solve the technical problems of traditional pre-employment testing methods in terms of identity verification and health assessment, such as independent equipment, fragmented processes, long time consumption, inability to effectively prevent substitute testing, reliance on contact sensors for health monitoring which is susceptible to environmental influences and cross-infection, insufficient accuracy of single-modal biometric identification under close-range obstruction, and difficulty in simultaneously and non-contactly assessing hidden health problems such as driver fatigue and emotional abnormalities, and achieving comprehensive judgment of multi-dimensional information.

[0007] The technical solution of this application is as follows: In a first aspect, this application discloses a biometric identification method for pre-employment health detection of drivers, comprising: acquiring multimodal biometric information of drivers before they start work; performing an identity verification judgment on the driver in response to the multimodal biometric information; acquiring the driver's physiological state characteristic information during the identity verification judgment; acquiring the driver's behavioral state characteristic information during the identity verification judgment; acquiring environmental context information and the driver's individual baseline characteristic record information; adjusting the individualized judgment range information corresponding to the physiological state characteristic information based on the environmental context information and the individual baseline characteristic record information to generate adjusted individualized judgment range information; performing a comprehensive judgment on the physiological state characteristic information and the behavioral state characteristic information based on the adjusted individualized judgment range information to generate a comprehensive health judgment result information; outputting a warning message based on the comprehensive health judgment result information; and, during the detection process, performing liveness detection and consistency verification judgment based on the multimodal biometric information to generate a substitution risk judgment result information, and outputting a substitution warning message when the substitution risk judgment result information indicates the existence of a substitution risk.

[0008] This technical solution enables comprehensive, non-contact testing of drivers' pre-employment health status. It effectively integrates identity verification and health assessment processes, and significantly reduces the risk of substitution through liveness detection and consistency verification. This solves the problems of traditional testing methods, such as fragmented processes, low efficiency, susceptibility to substitution, and inability to comprehensively assess drivers' health status.

[0009] Secondly, this application also discloses a biometric identification system for pre-employment health checks of drivers, used to perform biometric identification for pre-employment health checks of drivers, including: a biometric acquisition module for acquiring multimodal biometric information of the driver before starting work; an identity verification execution module for performing identity verification of the driver in response to the multimodal biometric information; a physiological information acquisition module for acquiring physiological state characteristic information of the driver during the identity verification process; a behavioral information acquisition module for acquiring behavioral state characteristic information of the driver during the identity verification process; an individual information acquisition module for acquiring environmental context information and individual baseline characteristic record information of the driver; and a judgment range adjustment module. The system includes a health judgment generation module, which adjusts the individualized judgment range information corresponding to physiological state characteristics based on environmental context information and individual baseline characteristic records, generating adjusted individualized judgment range information; a health judgment generation module, which performs a comprehensive judgment on physiological state characteristics and behavioral state characteristics based on the adjusted individualized judgment range information, generating comprehensive health judgment result information; an early warning information output module, which outputs early warning information based on the comprehensive health judgment result information; and a substitution risk judgment module, which performs liveness detection and consistency verification judgment based on multimodal biometric information during the detection process to generate substitution risk judgment result information, and outputs substitution early warning information when the substitution risk judgment result information indicates the existence of substitution risk.

[0010] This application provides a system that integrates multimodal biometric acquisition, identity verification, physiological behavior monitoring, individualized judgment range adjustment, comprehensive health judgment, and substitute inspection risk assessment. It can achieve automated, intelligent, and highly safe pre-employment health testing for drivers, effectively solving the problems of traditional testing systems such as single function, fragmented process, and susceptibility to substitute inspection.

[0011] Beneficial Effects: This application discloses a biometric identification method for pre-employment health checks of drivers. It acquires multimodal biometric information of drivers before they start work and performs identity verification in response to this information. Simultaneously, it acquires the driver's physiological and behavioral state characteristics during the verification process. Based on this, and combining environmental context information and the driver's individual baseline characteristic records, it adjusts the individualized judgment range information corresponding to the physiological state characteristics, generating adjusted individualized judgment range information. Subsequently, based on the adjusted individualized judgment range information, it performs a comprehensive judgment on the physiological and behavioral state characteristics, generating a comprehensive health judgment result, and outputs a warning message based on this result. Furthermore, throughout the entire detection process, it performs liveness detection and consistency verification based on the multimodal biometric information to generate a substitution risk judgment result, and outputs a substitution warning message when a substitution risk exists.

[0012] Through the above technical solution, this application effectively solves the shortcomings of existing technologies, such as independent identity verification and health assessment equipment, fragmented processes, long processing times, inability to effectively prevent substitute testing, reliance on contact sensors for health monitoring which is susceptible to environmental influences and cross-infection, insufficient accuracy of single-modal biometric recognition under close-range obstruction, and difficulty in simultaneously and non-contactly assessing hidden health issues such as driver fatigue and emotional abnormalities, and achieving comprehensive multi-dimensional information judgment. This application achieves simultaneous identity verification and health assessment through the fusion of multi-modal biometric information, significantly shortening detection time and improving efficiency. Non-contact acquisition of physiological and behavioral state characteristics avoids the drawbacks of contact sensors and reduces the risk of cross-infection. More importantly, the introduction of liveness detection and consistency verification can effectively identify and warn of substitute testing risks, fundamentally eliminating cheating. Simultaneously, personalized adjustments to the judgment range based on environmental context and individual baseline information make health judgment results more accurate and reliable. In summary, this application provides an efficient, accurate, safe, and intelligent pre-employment health detection solution for drivers, significantly improving the safety management level of public transportation and the transportation industry. Attached Figure Description

[0013] Figure 1 This is a flowchart of a biometric identification method for pre-employment health testing of drivers, according to one embodiment of the present invention. Figure 2 This is a flowchart of a biometric identification method for pre-employment health testing of drivers, according to another embodiment of the present invention. Figure 3 This is a system block diagram of a biometric identification system for pre-employment health detection of drivers, according to another embodiment of the present invention. Explanation of reference numerals in the attached figures: 1. Biometric identification system for pre-employment health checks of drivers; 11. Biometric acquisition module; 12. Identity judgment execution module; 13. Physiological information acquisition module; 14. Behavioral information acquisition module; 15. Individual information acquisition module; 16. Judgment range adjustment module; 17. Health judgment generation module; 18. Early warning information output module; 19. Substitute inspection risk judgment module. Detailed Implementation

[0014] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0015] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0016] This application proposes a biometric identification method for pre-employment health testing of drivers, combining... Figure 1 As shown, it includes: S1, obtain multimodal biometric information of drivers before they start work; S2, in response to multimodal biometric information, performs a driver identity verification judgment; S3, during the identity verification process, obtain the driver's physiological state characteristics information; S4. During the identity verification process, obtain the driver's behavioral status characteristic information; S5, acquire environmental context information and driver's individual baseline characteristics record information; S6. Based on environmental context information and individual baseline characteristic record information, adjust the individualized judgment range information corresponding to physiological state characteristic information to generate adjusted individualized judgment range information. S7. Based on the adjusted individualized judgment range information, perform a comprehensive judgment on physiological state characteristic information and behavioral state characteristic information to generate comprehensive health judgment result information; S8, based on the comprehensive health assessment results, outputs a warning message; S9, during the detection process, performs liveness detection and consistency verification based on multimodal biometric information to generate substitution risk assessment results, and outputs substitution warning information when the substitution risk assessment results indicate the existence of substitution risk.

[0017] The "multimodal biometric information" mentioned in this application refers to a set of feature information acquired through multiple sensors or acquisition technologies, used to characterize the uniqueness of an individual's identity or physiological authenticity. Multimodal biometric information may include, but is not limited to, facial image information, fingerprint information, iris information, and voiceprint information. It can be used alone for identity verification, or it can be used through multimodal fusion to perform liveness detection and consistency verification, thereby improving the security and reliability of the identity recognition process.

[0018] "Physiological state characteristic information" refers to objective data reflecting the driver's internal physiological function, which may include heart rate, body temperature, blood pressure, blood oxygen saturation, respiratory rate, and skin conductance. This physiological state characteristic information is preferably acquired through non-contact sensors to avoid the risk of cross-infection associated with contact testing, while also minimizing interference with the driver's on-duty process, thereby improving the overall acceptability and efficiency of the testing.

[0019] "Behavioral state characteristic information" refers to a data set reflecting a driver's external behavior and psychological state, which may include facial expression information, eye movement information, posture information, and voice tone information. Facial expression information can be used to reflect fatigue, anxiety, or emotional fluctuations; eye movement information can be used to assess the level of attention; posture information can reflect physical tension or abnormal states; and voice tone information can be used to help determine emotional stability. Comprehensive analysis of these behavioral state characteristic information helps to provide a more holistic assessment of a driver's psychological load and fatigue risk.

[0020] "Environmental context information" refers to external condition information related to the driver's environment during the detection process. This can include ambient temperature, humidity, light intensity, noise level, and time. Since environmental conditions significantly affect a driver's physiological responses and behavioral performance, a comprehensive analysis combining environmental context information is necessary when judging physiological and behavioral characteristics to avoid misjudgments caused by environmental changes.

[0021] "Individual baseline characteristic record information" refers to physiological and behavioral characteristic data collected over a long period or multiple times when the driver is in a normal healthy state. This individual baseline characteristic record information is used to characterize the driver's personalized health characteristics, such as their average resting heart rate range, normal body temperature range, and typical facial expressions and behavioral patterns. By comparing the current test results with the individual baseline characteristic record information, the individual adaptation bias caused by the general threshold of the group can be effectively reduced, and the specificity of the judgment results can be improved.

[0022] "Individualized judgment range information" refers to a judgment threshold or range dynamically generated based on individual baseline characteristic records and current environmental context information. This threshold or range is used to determine whether physiological and behavioral state characteristics are within the normal range. By dynamically adjusting the judgment range, the system can maintain the rationality of its judgment scale under different environmental conditions and individual differences, thereby improving the accuracy and stability of health status assessment.

[0023] "Liveness detection and consistency verification" refers to the process of comprehensively judging whether the detected object is a real living person and whether its biometrics are consistent with the registration information based on multimodal biometric information. This judgment process can effectively prevent substitution testing through photos, videos, masks, or other means, thereby ensuring the authenticity and credibility of the detection process.

[0024] The implementation environment of this application can be an integrated detection terminal, which integrates multiple sensor components and a processing unit. The sensor components may include visible light cameras, near-infrared cameras, short-wave infrared cameras, thermal imaging sensors, and microphones, used to collect multimodal biometric information, physiological state information, and behavioral state information; the processing unit is used to perform data acquisition, feature extraction, analysis and judgment, and result output. This detection terminal can be deployed in areas that drivers must pass through before starting work, such as fleet entrances and exits, dispatch centers, or rest areas, to achieve centralized detection before starting work.

[0025] In one implementation, acquiring multimodal biometric information of a driver before they start work can be accomplished through the collaborative use of multiple sensors. For example, a visible light camera can be used to acquire facial images of the driver, a near-infrared camera can acquire iris images, a fingerprint recognition module can acquire fingerprint information, or a microphone can acquire voiceprint information. This multimodal biometric information can be collected individually or simultaneously within the same detection process to improve the accuracy and robustness of identity verification.

[0026] In response to the acquired multimodal biometric information, the system performs a driver identity verification process, which involves comparing the real-time collected multimodal biometric information with pre-stored driver registration information. When the comparison result meets a preset similarity threshold, the detected object is confirmed as the corresponding driver, thus completing the identity verification process.

[0027] During the identity verification process, the system simultaneously acquires the driver's physiological state characteristics. For example, it acquires body surface temperature information through a thermal imaging sensor, heart rate and blood oxygen saturation information through photoplethysmography (PPG) analysis, or respiratory rate information through a radar sensor. Because the acquisition of these physiological state characteristics is carried out simultaneously with the identity verification process, additional detection steps are avoided, improving the continuity of the detection process.

[0028] During the identity verification process, the system also simultaneously acquires information on the driver's behavioral characteristics. For example, it analyzes facial expressions, eye movements, and head posture using a visible light camera to determine if there are signs of fatigue, anxiety, or decreased attention; it also collects voice tone information through a microphone to assist in assessing emotional state.

[0029] In addition, the system acquires environmental context information and individual baseline characteristic records for the driver. Environmental context information is obtained through environmental sensors, while individual baseline characteristic records are read from the driver's personal health record database.

[0030] Based on environmental context information and individual baseline feature records, the individualized judgment range information corresponding to physiological state feature information is dynamically adjusted to generate adjusted individualized judgment range information. This method enables the judgment threshold to adapt to physiological changes under different environmental conditions, avoiding misjudging normal fluctuations caused by the environment as abnormal states.

[0031] Based on this, and according to the adjusted individualized judgment range information, a comprehensive judgment is performed on physiological and behavioral state characteristics to generate a comprehensive health judgment result. During the comprehensive judgment process, multiple physiological and behavioral indicators can be jointly analyzed. When multiple indicators are simultaneously abnormal or at a critical state, it is determined that the driver has a health risk.

[0032] Based on the comprehensive health assessment results, the system outputs a warning message to prompt the driver or manager to take appropriate measures.

[0033] Throughout the testing process, the system continuously performs liveness detection and consistency verification based on multimodal biometric information, generating a risk assessment result for substitution. When the risk assessment result indicates a risk of substitution, the system outputs a substitution warning and can interrupt the current testing process, thereby ensuring the authenticity and security of the test results.

[0034] Optional, combined Figure 2 As shown, the steps for adjusting the individualized judgment range information corresponding to the physiological state characteristic information based on environmental context information and individual baseline characteristic record information to generate the adjusted individualized judgment range information include: A1 performs real-time analysis of the driver's facial images to determine whether there are signs of physiological stress and extracts physiological stress feature information; A2. The preliminary estimated heart rate information and body surface temperature information are compared with the resting state normal range information recorded in the individual baseline characteristic record information to determine whether the heart rate information or body surface temperature information is higher than the upper limit threshold of resting state, and the comparison judgment result is obtained. A3 generates state classification results based on physiological stress characteristics and comparison judgment results, and divides the driver's state into three levels: immediate passage, observation pending, and immediate warning. A4. When the status classification result information indicates that the driver's status is pending observation, the observation window information is restored to extend the detection time. A5. During the window period corresponding to the recovery observation window information, physiological state characteristic information and behavioral state characteristic information are continuously acquired and updated. A6. Within the window period corresponding to the recovery observation window information, calculate the rate of change of physiological state characteristic information and behavioral state characteristic information. A7 updates the individualized judgment range information based on environmental context information, individual baseline feature record information, rate of change information, and restored observation window information to generate context-adjusted individualized judgment range information. A8, after the window period corresponding to the recovery observation window information ends, the corresponding comprehensive judgment is performed again based on the individualized judgment range information adjusted by the context: when both the physiological state feature information and the behavioral state feature information are restored to the individualized judgment range information adjusted by the context, a pass judgment result information is generated; otherwise, an immediate warning judgment result information is generated and a warning information is output.

[0035] Specifically, real-time analysis of driver facial images refers to the system continuously acquiring facial image information of the driver during the detection process and processing this information using computer vision technology to identify visual features related to the driver's physiological stress state. Specifically, the system can analyze the driver's facial flushing, signs of sweating, muscle tension, and eye focus through facial expression recognition, micro-expression analysis, and facial skin color change detection, thereby extracting physiological stress feature information to characterize the driver's physiological stress state. This physiological stress feature information serves as a crucial input for judging the driver's psychophysiological state and, together with subsequently acquired physiological and behavioral state feature information, participates in a comprehensive judgment.

[0036] The preliminary estimated heart rate and body surface temperature information can be obtained using either non-contact or contact sensors. Non-contact sensors may include infrared thermal imagers and PPG sensors based on photoplethysmography (PPG), used to acquire heart rate and body surface temperature information without interfering with driver behavior. Contact sensors may include wearable heart rate straps or thermometers, used to provide more accurate measurement data when needed. The system compares the acquired heart rate and body surface temperature information with the driver's pre-recorded normal heart rate and body temperature ranges at rest in the individual's baseline characteristic record information to determine whether the current physiological state deviates from the individual's normal levels. The upper limit threshold at rest is set based on a large amount of individual sample data and medical reference standards, used to distinguish between normal physiological fluctuations and potential physiological stress states, thereby avoiding misjudging short-term, slight fluctuations as abnormal.

[0037] Based on this, the system classifies the driver's current state according to physiological stress characteristics and the comparison results mentioned above, generating driver state classification results. The driver state classification results include at least three levels: immediate pass, pending observation, and immediate warning. Immediate pass indicates that both the driver's physiological and behavioral characteristics are within the normal range defined by the individualized judgment information; immediate warning indicates the detection of obvious physiological abnormalities or signs of physiological stress, posing a potential safety risk and requiring immediate intervention; pending observation indicates that the driver's state is between the two levels, exhibiting slight abnormalities or unstable fluctuations, requiring further observation to confirm its development trend.

[0038] When the driver's condition classification result is determined to be pending observation, the system initiates the recovery observation window. The recovery observation window defines a continuous observation period, the duration of which can be set from 5 to 15 minutes according to application requirements. Within the window period corresponding to the recovery observation window, the system continuously acquires and updates the driver's physiological and behavioral state characteristics at a preset sampling frequency. Physiological state characteristics may include heart rate, body surface temperature, and respiratory rate, while behavioral state characteristics may include eye movement features, posture features, and micro-movement features. By continuously collecting data within the window period corresponding to the recovery observation window, the system can obtain continuous changes in the driver's condition.

[0039] Within the observation window period corresponding to the recovery observation window information, the system further calculates the rate of change information of physiological state characteristics and behavioral state characteristics based on continuously collected data. The rate of change information is used to characterize the trend of relevant characteristics over time, such as the rate of increase or decrease of heart rate, the fluctuation range of body surface temperature, the stability of respiratory rate, and the frequency of changes in behavioral state characteristics. The rate of change information can reflect whether the driver's condition is tending to recover, remain stable, or continue to deteriorate, thus providing a dynamic trend basis for subsequent judgments.

[0040] The system dynamically adjusts the individualized judgment range information based on environmental context information, individual baseline feature records, rate of change information, and recovery observation window information, generating context-adjusted individualized judgment range information. By incorporating environmental context information, such as ambient temperature, humidity, and noise levels, the system can identify the impact of environmental factors on physiological state characteristics. By combining individual baseline feature records and rate of change information, the system can adaptively correct the judgment threshold according to the driver's individual differences and state change trends, thereby avoiding misjudgments caused by using fixed thresholds and enabling the judgment criteria to dynamically adjust with changes in context and individual state.

[0041] After the observation window period corresponding to the recovery period ends, the system, based on the individualized judgment range information adjusted for the context, performs a comprehensive judgment on the driver's physiological and behavioral characteristics. If the comprehensive judgment result indicates that all relevant indicators have recovered to the normal range defined by the individualized judgment range information adjusted for the context, the system generates a pass judgment result; if one or more indicators are still out of range, or the rate of change information does not show a recovery trend, the system generates an immediate warning judgment result and outputs a warning message.

[0042] In some preferred embodiments, assuming a driver, during a pre-employment health check, has a preliminary estimated heart rate slightly higher than the upper resting threshold recorded in the individual's baseline characteristics, and simultaneously identifies mild signs of physiological stress based on real-time facial image analysis, such as slight pupil dilation and facial muscle tension, the system, based on the physiological stress characteristics and comparison results, determines the driver's driver status classification as pending and initiates a 10-minute recovery observation window.

[0043] During the observation window corresponding to the recovery period, the system continuously acquires physiological state characteristics of the driver, such as heart rate, body surface temperature, and respiratory rate. Simultaneously, it continuously monitors behavioral characteristics, such as eye movement and posture, through a camera, and calculates the rate of change of these characteristics in real time. For example, when the system detects a gradual decrease in heart rate, a stabilizing respiratory rate, and a gradually relaxing facial expression, the rate of change indicates that the driver's physiological state is recovering towards normal. The system also incorporates current environmental context information, indicating a suitable ambient temperature and no significant noise interference. Based on the driver's individual baseline characteristics, it dynamically adjusts the individualized judgment range information to generate context-adjusted individualized judgment range information.

[0044] After the 10-minute recovery observation window ends, the system performs a comprehensive assessment of the driver's physiological and behavioral characteristics based on the context-adjusted individualized judgment range. If the assessment results show that heart rate, body temperature, respiratory rate, and behavioral characteristics have all returned to within the context-adjusted individualized judgment range, the system generates a pass assessment result and allows the driver to return to work. Conversely, if the relevant indicators have not yet returned to normal, the system generates an immediate warning assessment result and outputs a warning message, prompting the driver to postpone returning to work and undergo further inspection.

[0045] Optionally, the steps of performing liveness detection and consistency verification based on multimodal biometric information to generate the result information of the substitution risk assessment include: Acquire transmission and reflection images of the driver's face in the visible light, near-infrared and short-wave infrared bands to form multi-band facial image information; Extract multi-band facial image information in the visible light, near-infrared and short-wave infrared bands, including reflectance intensity or transmittance information. Based on the reflection intensity information or transmittance information, estimate the moisture content characteristics of the skin surface and subcutaneous tissue; Based on the reflection intensity information or transmittance information, estimate the relative concentration characteristics of deep skin hemoglobin; Based on the reflection intensity information or transmittance information, analyze the uniform distribution characteristics of melanin; Based on moisture content characteristics, relative concentration characteristics, and uniform distribution characteristics, biological composition index information is generated, and regional consistency difference information of the same biological composition index information between different facial regions is calculated. Monitor the volatility of biological composition index information across consecutive frames to generate time series stability information; By integrating biological composition index information, regional consistency difference information, and time series stability information, the risk assessment of substitution testing is performed, and the result information of the substitution testing risk assessment is output.

[0046] Specifically, multimodal biometric information can include facial image information, acquired in the visible, near-infrared, and short-wave infrared bands. The visible light band typically refers to light with wavelengths ranging from 400 to 700 nanometers, the near-infrared band typically refers to light with wavelengths ranging from 700 to 1000 nanometers, and the short-wave infrared band typically refers to light with wavelengths ranging from 1000 to 2500 nanometers. By acquiring transmission and reflection images of the face in these different bands, multi-band facial image information can be formed, thereby capturing the unique optical responses of the skin under different spectra.

[0047] Reflectance intensity or transmittance information refers to the intensity or proportion of light reflected or penetrated after hitting the skin surface. This information can be extracted using spectral imaging equipment. For example, by analyzing the reflection intensity or transmittance at different wavelengths, the moisture content characteristics of the skin surface and subcutaneous tissue can be estimated. Water absorbs light differently at different wavelengths, so the moisture content can be inferred from the absorption rate at a specific wavelength.

[0048] Furthermore, the relative concentration characteristics of deep skin hemoglobin can also be estimated by analyzing reflectance or transmittance information. Hemoglobin has specific absorption peaks in the visible and near-infrared bands, and changes in these absorption peaks can reflect the concentration of hemoglobin. In addition, the uniform distribution characteristics of melanin can be obtained by analyzing the reflectance in the visible light band, because melanin has a significant impact on the absorption of visible light, and its uniform distribution is an important indicator for judging the authenticity of the skin.

[0049] Therefore, based on moisture content, relative concentration, and uniform distribution characteristics, a biocomposition index can be generated. This index comprehensively reflects the intrinsic biological characteristics of the skin. Furthermore, by calculating the regional consistency differences of the same biocomposition index across different facial regions, the uniformity of biological characteristics across different facial regions can be assessed. For example, the biocomposition index of a real face typically exhibits high consistency across different regions, while a fake face may show significant differences.

[0050] Furthermore, by monitoring the fluctuations in biometric index information across consecutive frames, time-series stability information can be generated. The biometric index of a real human face typically exhibits some physiological fluctuations over a short period, but overall it has high stability, while a fake face may show abnormal static or unnatural fluctuations.

[0051] Optionally, the step of performing liveness detection and consistency verification based on multimodal biometric information to generate the result information of the substitution risk assessment also includes: Based on monitoring the fluctuation of biological composition index information between consecutive frames, the correlation information between different biological composition index information in the time series is analyzed. By integrating biological composition index information, regional consistency difference information, volatility information, and correlation information, the risk assessment of substitution testing is performed, and the result information of the substitution testing risk assessment is output.

[0052] Specifically, based on monitoring the fluctuations of biological composition index information across consecutive frames, the system further analyzes the correlation information between different biological composition index information over time. This biological composition index information may include, but is not limited to, water content characteristics, relative hemoglobin concentration characteristics, and melanin uniform distribution characteristics. The correlation information over time refers to how these different biological composition index information influence each other, change synchronously, or exhibit lag effects over a period of time. For example, in a real living organism, changes in skin surface water content may have a specific physiological association with subcutaneous blood perfusion (reflected in relative hemoglobin concentration), and this association exhibits a specific pattern over time. This correlation can be quantitatively analyzed using various statistical or signal processing methods, such as calculating Pearson correlation coefficients, Granger causality, cross-correlation functions, or employing dynamic time warping (DTW). The aim is to reveal the intrinsic connections within the organism's physiological activities, thereby distinguishing real physiological responses from fabricated, uncoordinated simulated signals.

[0053] Furthermore, when assessing the risk of substitution testing, the system will comprehensively consider information on biocompatibility indices, regional consistency differences, volatility, and newly introduced correlation information. This means that, in addition to evaluating the absolute values ​​of each biocompatibility index, the differences between different regions, and the dynamic changes of a single index, the system will also assess the degree of interrelationship between these indices over time. For example, a multi-dimensional feature vector containing all the above information can be constructed, and then machine learning models (such as support vector machines, neural networks, or decision trees) can be used for classification to output the result of the substitution testing risk assessment.

[0054] Optionally, the step of performing liveness detection and consistency verification based on multimodal biometric information to generate the result information of the substitution risk assessment also includes: Acquire image information of the driver's face in the visible light, near-infrared and short-wave infrared bands, and apply spectral stimulation information to the driver's face; Extract the temporal spectral response curve information of facial image information as a function of spectral stimulation information, showing the change of reflectance intensity or transmittance over time. Based on time-domain spectral response curve information, we analyze whether there are physiological synchronous response characteristics of the face under the action of spectral stimulation information. Based on the time-domain spectral response curve information, we analyze whether there are characteristics of the physiological recovery process after the removal of spectral stimulus information. Calculate the regional consistency difference information of time-domain spectral response curves of different facial regions under the same stimulus wavelength; The time-domain spectral response curve information is matched with the preset physiological response pattern information to generate physiological response pattern matching degree information. By integrating physiological synchronous response characteristics, physiological recovery process characteristics, regional consistency differences, and physiological response pattern matching information, a deep spectral camouflage substitution risk assessment is performed, and the deep spectral camouflage substitution assessment result information is output.

[0055] Specifically, acquiring image information of the driver's face in the visible, near-infrared, and short-wave infrared bands refers to capturing the reflected or transmitted images of the driver's face within different wavelength ranges using multispectral imaging equipment to obtain rich spectral information. Based on this, applying spectral stimulation information to the driver's face can be understood as irradiating the face with a light source of specific wavelength, intensity, and duration, such as using pulsed light, scintillation light, or a narrowband light source modulated at a specific frequency. The aim is to induce observable physiological responses in real biological tissues.

[0056] Among these, extracting the time-domain spectral response curve information of facial image information under the action of spectral stimulation information, which involves continuously monitoring the changes in the intensity of reflection or transmission signals of the face in various spectral bands throughout the entire process of applying and removing spectral stimulation, and plotting these changes as a time series curve. This curve can reflect the dynamic physiological response process of biological tissues to external stimuli.

[0057] In practical applications, analyzing whether facial physiological synchronous response characteristics exist under spectral stimulation based on time-domain spectral response curve information refers to detecting whether changes in facial reflectance or transmission signals are consistent with changes in the rhythm, frequency, or intensity of the spectral stimulation. For example, under specific spectral stimulation, real skin tissue experiences instantaneous changes in physiological parameters such as blood perfusion and oxygenation levels, resulting in synchronous fluctuations in reflectance or transmission spectral characteristics.

[0058] Furthermore, based on the time-domain spectral response curve information, analyzing whether there are physiological recovery process characteristics after the removal of spectral stimulus information refers to observing whether the facial spectral response signal gradually recovers to its initial state according to the inherent physiological mechanisms of the organism after the stimulus is stopped. For example, changes in blood flow caused by vasodilation or vasoconstriction will have a specific recovery time constant and recovery trajectory after the stimulus is removed.

[0059] Furthermore, calculating the regional consistency difference information of time-domain spectral response curves under the same stimulus wavelength among different facial regions refers to comparing the similarity or difference in shape, amplitude, and delay of the response curves of different facial parts (such as the forehead, cheeks, and tip of the nose) to the same spectral stimulus. The physiological responses of a real human face typically exhibit a certain degree of physiological consistency across different regions, while camouflage may display unnatural regional differences.

[0060] Matching time-domain spectral response curve information with preset physiological response pattern information to generate physiological response pattern matching degree information involves comparing the actually measured time-domain spectral response curve with typical physiological response patterns of a large number of real human faces under the same stimulus conditions to assess their degree of agreement. The preset physiological response pattern information can be derived from large-scale biological experimental data, covering typical response characteristics of different individuals and under different physiological states.

[0061] Therefore, by integrating physiological synchronous response characteristics, physiological recovery process characteristics, regional consistency differences, and physiological response pattern matching information, a deep spectral camouflage substitution risk assessment is performed, outputting the deep spectral camouflage substitution assessment result. This comprehensive assessment, through multi-dimensional analysis, fully evaluates the dynamic physiological response of the face to spectral stimuli, thereby effectively distinguishing real biological tissue from highly realistic camouflage.

[0062] Optionally, the steps for performing depth spectral camouflage substitution risk assessment include: By analyzing the rising edge shape, falling edge shape, slope, and feature point occurrence time information of the time-domain spectral response curve, multi-dimensional analysis results are obtained. The recovery speed and trajectory of the reflection intensity or transmittance information to the initial state after the stimulus is removed are analyzed to obtain the recovery analysis results; The recovery analysis results and multidimensional analysis results are compared with the preset physiological response pattern information to generate comparison result information. Based on the comparison results, output the depth spectral camouflage and substitution detection results.

[0063] Specifically, the aforementioned time-domain spectral response curve information refers to the dynamic curve showing the change in facial reflectance intensity or transmittance over time under the influence of spectral stimulation. The rising and falling edge shape information describes the specific form of the curve's change from baseline to peak and from peak to baseline, such as its smoothness, steepness, or the presence of a plateau phase. The slope information quantifies the speed of these changes, reflecting the instantaneous intensity of the physiological response. The timing information of feature points refers to the location of key inflection points, turning points, or extreme points on the curve along the time axis; these points are typically associated with specific physiological events (such as vasodilation or changes in blood oxygen saturation). By analyzing these multi-dimensional features, the dynamic response details of biological tissues to spectral stimulation can be captured more precisely.

[0064] The recovery speed and trajectory of reflectance or transmittance information to its initial state after stimulus removal refer to the process by which facial optical properties gradually return to their pre-stimulation baseline state after the spectral stimulus ceases to act. Recovery speed quantifies the rate of this process, while the recovery trajectory describes the shape of the recovery path, such as linear recovery, exponential decay recovery, or oscillatory recovery. These recovery characteristics are closely related to deep physiological mechanisms such as the elasticity, metabolic rate, and blood circulation of biological tissues, and are important indicators for distinguishing living organisms from non-living organisms or camouflage.

[0065] In practical applications, the preset physiological response pattern information is established based on a large amount of real human data. It includes typical response curve characteristics, recovery properties, and threshold ranges or pattern templates for multi-dimensional analysis results of biological tissues under different physiological states in response to specific spectral stimuli. By comparing the recovery analysis results and multi-dimensional analysis results obtained from actual testing with these preset patterns, the authenticity and physiological consistency of the current response can be assessed. The comparison result information can be a matching score, a classification label, or an indicator of the degree of difference.

[0066] Optionally, the steps to analyze the rising edge shape, falling edge shape, slope, and feature point occurrence time information of the time-domain spectral response curve to obtain multi-dimensional analysis results include: Calculate the instantaneous slope information of the rising and falling edges of the time-domain spectral response curve; Identify inflection points and extreme points on the time-domain spectral response curve; Record the position and time information of inflection points and extreme points on the time axis; Analyze the differences in the slope change patterns of the rising and falling edges of the same stimulation wavelength among different facial regions; Analyze the time-series differences of feature points under the same stimulation wavelength in different facial regions; Analyze the curvature changes at the rising and falling edges, and calculate the degree of curvature difference between different regions; By integrating instantaneous slope information, location and time information, slope change pattern difference information, feature point time series difference information, curvature change information, and difference degree information, the depth spectral camouflage and substitution detection result information is output.

[0067] Specifically, calculating the instantaneous slope information of the rising and falling edges of the time-domain spectral response curve refers to obtaining the rate of change of reflectance or transmittance information over time at a specific point in time by differentiating the curve or using finite difference methods. This instantaneous slope information can reflect the immediate response speed and intensity of biological tissue to spectral stimuli. Identifying inflection points and extreme points on the time-domain spectral response curve can be understood as detecting the zeros of the second derivative (inflection points) and the first derivative (extreme points) of the curve using mathematical methods. These points typically correspond to key turning points or maximum / minimum response states in the physiological response process. In practical applications, recording the position and time information of inflection points and extreme points on the time axis, such as precisely recording the timestamps of these key points, aims to capture the temporal characteristics of the physiological response, providing a temporal basis for subsequent pattern matching and anomaly detection.

[0068] Furthermore, analyzing the differences in the slope change patterns of the rising and falling edges of the same stimulus wavelength across different facial regions involves comparing the similarity or difference in the rates of rise and fall of reflectance or transmittance information in different facial regions (e.g., forehead, cheeks, nose) under the same spectral stimulation. This difference can reveal the uniformity of camouflage materials in different regions or local anomalies in biological tissue responses. Additionally, analyzing the time series differences of feature points under the same stimulus wavelength across different facial regions involves comparing the consistency or deviation of inflection point and extreme point information on the time axis across different facial regions. For example, when a real face is subjected to uniform stimulation, the physiological response time sequence of different regions should have high consistency, while camouflage may lead to temporal mismatches. Simultaneously, analyzing the curvature change information of the rising and falling edges and calculating the degree of curvature difference between different regions involves evaluating the degree of curvature and its trend by calculating the second derivative of the curve and comparing the differences in curvature across different facial regions. Curvature variation information can more precisely characterize the dynamic processes of physiological responses, such as subtle changes in blood perfusion or tissue hydration, while curvature differences between regions help identify local camouflage. Finally, by integrating instantaneous slope information, location-time information, slope variation pattern difference information, feature point time series difference information, curvature variation information, and difference degree information, a depth spectral camouflage detection judgment is performed to output the depth spectral camouflage detection judgment result information.

[0069] Optionally, the steps to analyze the recovery speed and trajectory of the reflection intensity or transmittance information to its initial state after stimulus removal, and to obtain the recovery analysis results, include: Continuously monitor the changes in reflectance or transmittance of different facial regions across multiple spectral bands and extract the temporal curve information of these changes. Based on the stimulus intensity and duration of the spectral stimulus information, adjust the analysis window and analysis weight of the changing time-domain curve information; Calculate the instantaneous recovery rate information of the changing time-domain curve information in different time periods; Identify key inflection points on the changing time-domain curve information and record the position and time information of these key inflection points on the time axis; Analyze the consistency between instantaneous recovery rate information and position-time information under different spectral bands; The analysis examines the recovery delay between the recovery initiation point and the time of stimulus removal, as well as the recovery time required to reach a steady state, under different stimulus intensity and duration conditions. Analyze the curve shape characteristics of the changing time-domain curve; By integrating instantaneous recovery rate information, location and time information, consistency information, recovery delay, recovery time, and curve shape characteristics, the system outputs depth spectral camouflage and substitution detection result information.

[0070] Specifically, "continuously monitoring changes in reflectance or transmittance of different facial regions across multiple spectral bands and extracting temporal curve information" refers to the system continuously collecting reflectance or transmittance data of different areas of the driver's face across multiple preset spectral bands, including visible light, near-infrared, and short-wave infrared, after the spectral stimulus information has been removed. These data are recorded over time, forming a series of continuous temporal curves to capture the dynamic process of physiological recovery. The aim is to obtain comprehensive and detailed physiological recovery data, providing a foundation for subsequent in-depth analysis.

[0071] Specifically, "adjusting the analysis window and weights of the time-domain curve information based on the stimulus intensity and duration of the spectral stimulus information" refers to dynamically adjusting the time range and importance of data from each time period based on parameters of the spectral stimulus information applied to the driver's face (e.g., light intensity and duration at a specific wavelength). For example, for stronger stimuli, the recovery process may be longer, and the analysis window will be extended accordingly; for certain key recovery stages, their data will be given higher weight. The aim is to make recovery analysis more targeted and accurate, adapting to the physiological response characteristics under different stimulus conditions.

[0072] In practical applications, "calculating the instantaneous recovery rate information of the changing time-domain curve information within different time periods" refers to quantifying the rate of change of facial reflex intensity or transmissivity within minute time intervals after stimulus removal by performing differential or difference operations on the changing time-domain curve information. This can reveal the speed and dynamic trends of the physiological recovery process. Its purpose is to precisely capture the dynamic characteristics of physiological recovery, rather than simply focusing on the overall recovery time.

[0073] Furthermore, "identifying key inflection points on the changing time-domain curve information and recording the position and time information of these key inflection points on the time axis" refers to automatically detecting the moments when the curve shape undergoes significant changes through algorithmic analysis of the time-domain curve information, such as the recovery start point, recovery acceleration point, recovery deceleration point, and inflection point reaching a stable state. The time positions of these key inflection points are precisely recorded. The purpose is to locate important stages in the physiological recovery process and provide key time references for subsequent comparison of physiological response patterns.

[0074] The section on "analyzing the consistency between instantaneous recovery rate information and position-time information across different spectral bands" refers to comparing whether the instantaneous recovery rate information of different areas of the driver's face and the position-time information of key turning points exhibit synchronicity or conformity to expected physiological correlation across different spectral bands. For example, the recovery rates of certain physiological indicators should have inherent coordination across different bands. The aim is to verify the authenticity and internal logic of the physiological recovery process, as spoofing may struggle to maintain high consistency across all bands and time points.

[0075] Specifically, "analyzing the recovery delay between the recovery initiation point and the time of stimulus removal, and the recovery time required to reach a steady state under different stimulus intensities and durations" refers to measuring the time interval from the complete removal of spectral stimulus information to the onset of facial physiological responses (recovery delay), and the time required from stimulus removal to the complete recovery of physiological indicators to their initial or steady state (recovery time). These time parameters are recorded and analyzed in detail under different stimulus conditions. The aim is to quantify the overall temporal characteristics of physiological recovery and assess its dependence on stimulus parameters to distinguish between genuine physiological responses and faking behavior.

[0076] Furthermore, "analyzing the curve shape characteristics of the changing time-domain curves" refers to an in-depth analysis of the morphological features of the entire recovery time-domain curve information, including but not limited to the curve's smoothness, concavity / convexity, oscillation patterns, peak or trough characteristics, and similarity to standard physiological recovery curves. Its purpose is to comprehensively characterize the physiological recovery pattern from both macroscopic and microscopic levels, providing richer diagnostic criteria.

[0077] Therefore, "outputting the judgment result of deep spectral camouflage substitution based on comprehensive instantaneous recovery rate information, location and time information, consistency information, recovery delay, recovery time, and curve shape feature information" refers to the multi-dimensional fusion of all the extracted and analyzed recovery feature parameters, and the comprehensive evaluation through machine learning models or expert systems to ultimately generate a judgment result regarding the existence of deep spectral camouflage substitution risk. Its purpose is to utilize multi-source information for cross-validation and comprehensive decision-making, thereby improving the accuracy and robustness of substitution judgment.

[0078] Optionally, the step of analyzing the recovery rate and recovery trajectory of the reflection intensity information or transmittance information to the initial state after stimulus removal also includes: A secondary spectral perturbation was applied after the stimulus was removed, and the transient response characteristics of the face under the secondary spectral perturbation were monitored. Analyze the difference in secondary perturbation response between instantaneous response characteristic information and the pre-defined physiological response pattern information of real biological tissues under the same perturbation; The recovery delay and recovery time are compared with preset physiological threshold information to generate threshold comparison result information; By integrating instantaneous recovery rate information, location and time information, consistency information, recovery delay, recovery time, curve shape feature information, secondary disturbance response difference information, and threshold comparison result information, the depth spectral camouflage and substitution detection result information is output.

[0079] Specifically, after the initial spectral stimulus is removed, a secondary spectral perturbation can be applied to more deeply probe the physiological authenticity of the tested object. This secondary spectral perturbation can be a spectral signal with a different wavelength, intensity, or duration than the initial stimulus, aiming to induce a new transient physiological response in the tested face. By monitoring the transient response characteristics of the face under the secondary spectral perturbation, such as rapid changes in reflectance or transmittance, and absorption or scattering characteristics in specific bands, richer physiological dynamic information can be obtained. The difference in the secondary perturbation response between the transient response characteristics and the pre-established physiological response pattern of real biological tissue under the same perturbation refers to comparing the monitored transient response characteristics under the secondary perturbation with a pre-established standard physiological response pattern of real biological tissue under the same secondary perturbation. Real biological tissue typically has specific physiological mechanisms and patterns in response to secondary perturbations, which imposters may find difficult to accurately simulate. By analyzing the differences between the two, imposter behavior can be effectively identified. In practical applications, comparing the recovery delay and recovery time with preset physiological thresholds to generate threshold comparison results involves comparing the recovery delay and recovery time after the removal of the first stimulus with preset physiological thresholds. These thresholds are normal physiological ranges derived from statistical analysis of a large amount of real biological tissue data. If the recovery delay or recovery time of the tested object exceeds these preset thresholds, it indicates that its physiological response may be abnormal, thus generating corresponding threshold comparison results as a basis for judging the risk of substitution. Ultimately, the deep spectral camouflage substitution judgment result information is generated by comprehensively considering instantaneous recovery rate information, location and time information, consistency information, recovery delay, recovery time, curve shape characteristics, secondary perturbation response differences, and threshold comparison results. These multi-dimensional and multi-stage feature information together constitute a comprehensive assessment of the physiological authenticity of the tested object, thereby improving the accuracy and robustness of deep spectral camouflage substitution judgment.

[0080] This application also discloses a biometric identification system for pre-employment health checks of drivers, used to perform biometric identification for pre-employment health checks of drivers, combined with... Figure 3 As shown, the biometric identification system 1 for pre-employment health checks of drivers includes: Biometric acquisition module 11 is used to acquire multimodal biometric information of drivers before they take up their posts; The identity verification execution module 12 is used to perform identity verification of the driver in response to multimodal biometric information; The physiological information acquisition module 13 is used to acquire the driver's physiological state characteristic information during the identity verification process; The behavior information acquisition module 14 is used to acquire the driver's behavior status feature information during the identity verification process; The individual information acquisition module 15 is used to acquire environmental context information and driver's individual baseline characteristic record information; The judgment range adjustment module 16 is used to adjust the individualized judgment range information corresponding to the physiological state feature information based on environmental context information and individual baseline feature record information, and generate the adjusted individualized judgment range information. The health assessment generation module 17 is used to perform a comprehensive assessment of physiological state characteristic information and behavioral state characteristic information based on the adjusted individualized assessment range information, and generate comprehensive health assessment result information. The early warning information output module 18 is used to output early warning information based on the comprehensive health assessment results. The substitution risk assessment module 19 is used to perform liveness detection and consistency verification based on multimodal biometric information during the detection process, so as to generate substitution risk assessment result information, and output substitution warning information when the substitution risk assessment result information indicates that there is substitution risk.

[0081] Specifically, the biometric acquisition module can include, but is not limited to, visible light cameras, near-infrared cameras, short-wave infrared cameras, fingerprint recognition sensors, iris scanners, or microphones, used to collect multimodal biometric information such as the driver's facial image information, fingerprint information, iris information, or voiceprint information. These sensors can be configured independently according to actual deployment needs, or integrated into the same multifunctional acquisition unit to work uniformly, simultaneously acquiring multiple biometric information in a single detection process. This provides a multi-source data foundation for subsequent identity verification, liveness detection, and consistency verification.

[0082] The identity verification execution module can be an embedded processor located inside the detection terminal, or a server-side computing unit connected via a communication interface. Internally, it runs algorithms for biometric comparison. The module receives multimodal biometric information collected by the biometric acquisition module and compares it with pre-stored driver registration information. It performs identity verification by calculating similarity or matching degree. When the comparison result meets a preset consistency condition, the module confirms that the current detection object is the corresponding driver, thus allowing subsequent health checks to continue.

[0083] The physiological information acquisition module can be configured with non-contact sensors, such as thermal imaging sensors to acquire body surface temperature information, and photoplethysmography (PPG) sensors or radar sensors to acquire heart rate and respiratory rate information. The physiological information acquisition module is configured to operate synchronously with the driver's identity verification process, eliminating the need for additional detection steps to collect physiological state characteristic information. This improves overall detection efficiency and avoids discomfort or resistance from multiple tests for the driver.

[0084] The behavior information acquisition module may include a vision processing unit equipped with image processing and analysis algorithms. Based on facial image information, eye region information, and head posture information acquired by a visible light camera, it analyzes the driver's facial expression features, eye movement features, and posture changes to generate behavioral state feature information. Furthermore, the behavior information acquisition module may also include a voice analysis unit to analyze voice tone information collected by a microphone to help determine the driver's emotional state and level of concentration. By comprehensively acquiring multiple behavioral features, a more complete reflection of the driver's external state changes can be obtained.

[0085] The individual information acquisition module can include environmental sensors, such as temperature, humidity, light, and noise sensors, to collect environmental context information and obtain objective conditions of the detection environment in real time. Simultaneously, the individual information acquisition module can also communicate with the driver's personal health record database via a network interface to obtain the driver's individual baseline characteristic records. By combining environmental context information with individual baseline characteristic records, the system can provide a reference basis for subsequent judgments that matches individual differences and environmental changes.

[0086] The judgment range adjustment module can be a processor unit running a specific judgment range adjustment algorithm. Based on environmental context information and individual baseline feature records provided by the individual information acquisition module, it dynamically adjusts the judgment thresholds or ranges corresponding to physiological and behavioral state feature information. For example, when the ambient temperature is high, the judgment range adjustment module can correspondingly widen the normal judgment range for body temperature information; when the driver's individual baseline feature records show a persistently high heart rate, the judgment range adjustment module can individually correct the normal fluctuation range of heart rate information. This dynamic adjustment mechanism avoids misjudgments caused by using uniform fixed thresholds, improving the accuracy and adaptability of the judgment results.

[0087] The health assessment generation module can be a central processing unit running a comprehensive assessment algorithm. This algorithm fuses and analyzes adjusted individualized assessment range information, physiological state characteristics, and behavioral state characteristics. The comprehensive assessment algorithm can be implemented based on a multi-parameter fusion model or a machine learning model. By comprehensively evaluating multiple physiological and behavioral indicators, it generates a comprehensive health assessment result reflecting the driver's overall health status. This comprehensive health assessment result can simultaneously consider abnormalities in individual indicators as well as the coordinated changes in multiple indicators, thus avoiding biased judgments based on a single feature.

[0088] The warning information output module may include a display screen, a speaker, or a network communication interface. When the comprehensive health assessment result information output by the health assessment generation module indicates that there is a health risk, it outputs warning information in the form of visual prompts, sound prompts, or remote notifications, thereby promptly reminding the driver or management personnel to take appropriate measures.

[0089] The proxy testing risk assessment module can be a standalone processing unit, internally running liveness detection and consistency verification algorithms. This module continuously analyzes the multimodal biometric information collected by the biometric acquisition module. By analyzing the temporal changes, detailed features, and multimodal consistency of biometrics, it determines whether the detected object is a genuine living person and verifies whether its biometric information matches the driver's registration information. When the proxy testing risk assessment result indicates a proxy testing risk, the module triggers the warning information output module to output a proxy testing warning and can interrupt the current detection process to ensure the authenticity and reliability of the detection results.

[0090] Traditional pre-employment driver screening methods generally suffer from problems such as independent equipment, fragmented processes, high risk of substitute testing, reliance on contact sensors for health monitoring, and insufficient accuracy of single-modal recognition in terms of identity verification and health assessment. The biometric identification system for pre-employment driver health screening proposed in this application effectively overcomes these technical shortcomings through modular design and multi-module collaborative operation.

[0091] Specifically, this system achieves simultaneous identity verification and health status detection through the coordinated operation of the biometric acquisition module, identity judgment execution module, physiological information acquisition module, and behavioral information acquisition module. This significantly shortens the overall detection time while ensuring detection accuracy, and avoids the operational complexity and inefficiency problems caused by traditional separate detection processes.

[0092] Furthermore, this system employs a non-contact method to acquire physiological and behavioral characteristic information, effectively reducing the risk of cross-infection and improving the driver's testing experience. Through the coordinated efforts of the individual information acquisition module, the judgment range adjustment module, and the health judgment generation module, this system can dynamically adjust the judgment range by combining environmental context information and individual baseline characteristic records, achieving more personalized and refined health assessments and avoiding misjudgments caused by a "one-size-fits-all" judgment standard.

[0093] Meanwhile, the proxy testing risk assessment module continuously performs liveness detection and consistency verification during the testing process, preventing proxy testing from occurring and ensuring the consistency between the tested object and the registered identity, thus giving the final test results higher credibility and security. The synergistic effect of these technical features gives the biometric identification system of this application significant technical advantages and practical application value in the field of pre-employment health testing for drivers.

[0094] 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 biometric method for pre-employment health detection of a driver, characterized by, include: Obtain multimodal biometric information of drivers before they start working; In response to the multimodal biometric information, an identity verification judgment of the driver is performed; During the identity verification process, the driver's physiological state characteristics information is obtained; During the identity verification process, obtain the driver's behavioral status characteristics information; Acquire environmental context information and driver's individual baseline characteristics; Based on the environmental context information and the individual baseline feature record information, the individualized judgment range information corresponding to the physiological state feature information is adjusted to generate the adjusted individualized judgment range information; Based on the adjusted individualized judgment range information, a comprehensive judgment is performed on the physiological state feature information and the behavioral state feature information to generate comprehensive health judgment result information; Based on the comprehensive health assessment results, an early warning message is output; During the detection process, liveness detection and consistency verification are performed based on the multimodal biometric information to generate substitution risk assessment results. When the substitution risk assessment results indicate the existence of substitution risk, substitution warning information is output.

2. The biometric identification method for pre-employment health testing of drivers according to claim 1, characterized in that, The step of adjusting the individualized judgment range information corresponding to the physiological state feature information based on the environmental context information and the individual baseline feature record information to generate the adjusted individualized judgment range information includes: Real-time analysis of the driver's facial images is performed to determine whether there are signs of physiological stress and to extract physiological stress feature information; The preliminary estimated heart rate information and body surface temperature information are compared with the resting state normal range information of the individual recorded in the individual baseline feature record information to determine whether the heart rate information or the body surface temperature information is higher than the upper limit threshold of resting state, and the comparison judgment result is obtained. Based on the physiological stress characteristic information and the comparison judgment results, state classification result information is generated, and the driver's state is divided into three levels: immediate passage, observation pending, and immediate warning. When the status classification result information indicates that the driver's status is pending observation, the observation window information is restored to extend the detection time; During the window period corresponding to the recovery observation window information, the physiological state feature information and the behavioral state feature information are continuously acquired and updated; Within the window period corresponding to the recovery observation window information, the rate of change information of the physiological state feature information and the behavioral state feature information is calculated; Based on the environmental context information, the individual baseline feature record information, the rate of change information, and the recovery observation window information, the individualized judgment range information is updated to generate context-adjusted individualized judgment range information; After the window period corresponding to the restored observation window information ends, the corresponding comprehensive judgment is performed again based on the context-adjusted individualized judgment range information: when both the physiological state feature information and the behavioral state feature information are restored to the context-adjusted individualized judgment range information, a pass judgment result information is generated; otherwise, an immediate warning judgment result information is generated and a warning information is output.

3. The biometric identification method for pre-employment health testing of drivers according to claim 1, characterized in that, The step of performing liveness detection and consistency verification based on the multimodal biometric information to generate the result information of the substitution risk assessment includes: Acquire transmission and reflection images of the driver's face in the visible light, near-infrared and short-wave infrared bands to form multi-band facial image information; Extract the reflection intensity or transmittance information of the multi-band facial image information in the visible light, near-infrared and short-wave infrared bands; Based on the reflection intensity information or transmittance information, estimate the moisture content characteristics of the skin surface and subcutaneous tissue; Based on the reflected intensity information or transmittance information, the relative concentration characteristics of deep skin hemoglobin are estimated; Based on the reflected intensity information or transmittance information, analyze the uniform distribution characteristics of melanin; Based on the moisture content feature information, the relative concentration feature information, and the uniform distribution feature information, biological composition index information is generated, and the regional consistency difference information of the same biological composition index information between different facial regions is calculated. The fluctuation of the biological composition index information between consecutive frames is monitored to generate time series stability information; By combining the biological composition index information, the regional consistency difference information, and the time series stability information, a substitution risk assessment is performed, and the substitution risk assessment result information is output.

4. The biometric identification method for pre-employment health testing of drivers according to claim 3, characterized in that, The step of performing liveness detection and consistency verification based on the multimodal biometric information to generate the result information of the substitution risk assessment further includes: Based on monitoring the fluctuation of the biological composition index information between consecutive frames, the correlation information between different biological composition index information in the time series is analyzed. By combining the biological composition index information, the regional consistency difference information, the volatility information, and the correlation information, a substitution risk assessment is performed, and the substitution risk assessment result information is output.

5. A biometric identification method for pre-employment health testing of drivers according to claim 1, characterized in that, The step of performing liveness detection and consistency verification based on the multimodal biometric information to generate the result information of the substitution risk assessment further includes: Acquire image information of the driver's face in the visible light, near-infrared and short-wave infrared bands, and apply spectral stimulation information to the driver's face; Extract the temporal spectral response curve information of facial image information as a function of spectral stimulation information, showing the change of reflectance intensity or transmittance over time. Based on the time-domain spectral response curve information, analyze whether there are physiological synchronous response characteristics on the face under the action of the spectral stimulus information; Based on the time-domain spectral response curve information, analyze whether there is physiological recovery process characteristic information after the removal of the spectral stimulus information; Calculate the regional consistency difference information of the time-domain spectral response curves of different facial regions under the same stimulation wavelength; The time-domain spectral response curve information is matched with preset physiological response pattern information to generate physiological response pattern matching degree information. By combining the physiological synchronous response feature information, the physiological recovery process feature information, the regional consistency difference information, and the physiological response pattern matching information, a deep spectral camouflage substitution risk assessment is performed to output the deep spectral camouflage substitution assessment result information.

6. A biometric identification method for pre-employment health testing of drivers according to claim 5, characterized in that, The steps for performing the risk assessment of deep spectral camouflage substitution detection include: By analyzing the rising edge shape information, falling edge shape information, slope information, and feature point occurrence time information of the time-domain spectral response curve information, multi-dimensional analysis results are obtained. The recovery speed and trajectory of the reflection intensity or transmittance information to the initial state after the stimulus is removed are analyzed to obtain the recovery analysis results; The recovery analysis results and multi-dimensional analysis results are compared with preset physiological response pattern information to generate comparison result information. Based on the comparison results, the depth spectral camouflage and substitution detection results are output.

7. A biometric identification method for pre-employment health testing of drivers according to claim 6, characterized in that, The steps for analyzing the rising edge shape information, falling edge shape information, slope information, and feature point occurrence time information of the time-domain spectral response curve to obtain multi-dimensional analysis results include: Calculate the instantaneous slope information of the rising and falling edges of the time-domain spectral response curve; Identify the inflection point and extreme point information on the time-domain spectral response curve; Record the position and time information of the inflection point and extreme point information on the time axis; Analyze the differences in the slope change patterns of the rising and falling edges of the same stimulation wavelength among different facial regions; Analyze the time-series differences of feature points under the same stimulation wavelength in different facial regions; Analyze the curvature changes at the rising and falling edges, and calculate the degree of curvature difference between different regions; By combining the instantaneous slope information, the location and time information, the slope change pattern difference information, the feature point time series difference information, the curvature change information, and the difference degree information, the depth spectral camouflage and substitution detection result information is output.

8. A biometric identification method for pre-employment health testing of drivers according to claim 6, characterized in that, The steps for analyzing the recovery speed and trajectory of the reflection intensity or transmittance information after the stimulus is removed to obtain the recovery analysis results include: Continuously monitor the changes in reflectance or transmittance of different facial regions across multiple spectral bands and extract the temporal curve information of these changes. Based on the stimulation intensity and duration of the spectral stimulation information, the analysis window and analysis weight of the changing time-domain curve information are adjusted; Calculate the instantaneous recovery rate information of the changing time-domain curve information in different time periods; Identify key inflection points on the changing time-domain curve information and record the position and time information of the key inflection points on the time axis; Analyze the consistency between the instantaneous recovery rate information and the position and time information under different spectral bands; The analysis examines the recovery delay between the recovery initiation point and the time of stimulus removal, as well as the recovery time required to reach a steady state, under different stimulus intensity and duration conditions. Analyze the curve shape characteristics of the changing time-domain curve information; By combining the instantaneous recovery rate information, the location and time information, the consistency information, the recovery delay, the recovery time, and the curve shape feature information, the depth spectral camouflage and substitution detection result information is output.

9. A biometric identification method for pre-employment health testing of drivers according to claim 8, characterized in that, The step of analyzing the recovery speed and trajectory of the reflection intensity information or transmittance information to the initial state after stimulus removal also includes: A secondary spectral perturbation was applied after the stimulus was removed, and the transient response characteristics of the face under the secondary spectral perturbation were monitored. Analyze the difference in secondary perturbation response between the instantaneous response feature information and the preset physiological response pattern information of real biological tissue under the same perturbation; The recovery delay and recovery time are compared with preset physiological threshold information to generate threshold comparison result information; By combining the instantaneous recovery rate information, the location and time information, the consistency information, the recovery delay, the recovery time, the curve shape feature information, the secondary perturbation response difference information, and the threshold comparison result information, the depth spectral camouflage and substitution detection judgment result information is output.

10. A biometric identification system for pre-employment health checks of drivers, used to perform biometric identification for pre-employment health checks of drivers, characterized in that, include: The biometric acquisition module is used to acquire multimodal biometric information of drivers before they start working. The identity verification execution module is used to perform an identity verification judgment on the driver in response to the multimodal biometric information; The physiological information acquisition module is used to acquire the driver's physiological state characteristics during the identity verification process. The behavior information acquisition module is used to acquire the driver's behavior status feature information during the identity verification process; The individual information acquisition module is used to acquire environmental context information and driver's individual baseline characteristic records. The judgment range adjustment module is used to adjust the individualized judgment range information corresponding to the physiological state feature information based on the environmental context information and the individual baseline feature record information, and generate the adjusted individualized judgment range information. The health assessment generation module is used to perform a comprehensive assessment of the physiological state feature information and the behavioral state feature information based on the adjusted individualized assessment range information, and generate comprehensive health assessment result information. The early warning information output module is used to output early warning information based on the comprehensive health assessment results. The substitution risk assessment module is used to perform liveness detection and consistency verification based on the multimodal biometric information during the detection process, so as to generate substitution risk assessment result information, and output substitution warning information when the substitution risk assessment result information indicates that there is a substitution risk.

Images