Construction site access control face recognition management system based on dynamic identity verification

Abstract

The invention relates to the technical field of biological recognition, and discloses a construction site access control face recognition management system based on dynamic identity verification. Comprising a multi-mode biological recognition module, an intelligent switching mechanism, a self-adaptive image acquisition device, a feature dynamic updating module and an integrated control system. According to the system, face recognition and iris recognition are combined, a recognition mode is dynamically adjusted in combination with environment monitoring data (such as illumination intensity and dust concentration), the problem that traditional face recognition is low in recognition accuracy in a complex construction site environment is solved, and the recognition precision is ensured through an intelligent switching mechanism under the conditions of low illumination and high dust concentration. Through feature dynamic updating of the deep learning algorithm, the system can optimize the recognition model according to new features collected in real time, the long-term stable recognition performance is ensured, and the problem of precision reduction caused by a static model is avoided. And the exposure time, the gain and the camera cleaning mechanism are automatically adjusted, so that the image acquisition always keeps high quality in various construction site environments.

Description

Technical Field

[0001] The invention relates to a construction site access control face recognition management system and system based on dynamic identity authentication, belonging to the technical field of biometrics. Background Art

[0002] As construction sites become increasingly complex, the construction industry has placed higher demands on access control management systems, especially at construction sites, where the identity verification of construction workers must not only be efficient and accurate, but also be able to accurately identify personnel in harsh environments such as insufficient light and high dust concentration. Traditional construction site access control systems mostly use a single face recognition technology, usually by installing high-definition cameras to collect and identify workers' facial images. These systems were originally designed to improve the safety and management efficiency of construction sites, but they face many challenges in actual applications. For example: 1. In the construction site environment, due to the large changes in natural light during construction or the dust obstruction caused by the flying dust on the construction site, ordinary face recognition systems often cannot guarantee stable operation in low light or extreme environments. Even high-quality cameras are difficult to provide clear images in dark or bright light conditions, thus affecting the accuracy of recognition.

[0003] 2. The wearing of protective equipment such as helmets and masks by workers on construction sites also greatly increases the difficulty of face recognition. These equipment not only block important facial features, but also prevent some facial features from being correctly collected, thus affecting the effect of the recognition algorithm. Although some existing technologies attempt to solve this problem by supplementing lighting or image clarity enhancement algorithms, it is still difficult to completely eliminate the negative impact of these factors, resulting in a high error recognition rate of the system. Especially in large-scale construction sites with frequent worker turnover, the stability and robustness of traditional face recognition systems are poor.

[0004] To solve these problems, some traditional systems have introduced other biometric recognition methods such as iris recognition, and have attempted to make up for the shortcomings of a single mode by combining multiple recognition methods (such as face recognition and iris recognition). However, single iris recognition technology also has the problem of poor adaptability, especially in construction sites, where iris images may be damaged by strong light and dust, which in turn affects recognition accuracy. When solving problems in this way, the system usually needs to do a lot of additional optimization and supplementation in hardware and algorithms, which not only increases equipment and operation and maintenance costs, but also makes it difficult to achieve system flexibility and efficiency. Summary of the invention

[0005] The present invention provides a construction site access control face recognition management system based on dynamic identity authentication, and its main purpose is to solve the problem. To achieve the above purpose, the present invention provides a construction site access control face recognition management system based on dynamic identity authentication, comprising: A multimodal biometric module, used to selectively perform identity authentication according to environmental conditions, the multimodal biometric module comprising: a face recognition unit, used to collect and recognize the face image of the construction worker; an iris recognition unit, used to collect and recognize the iris image of the construction worker; Among them, the face recognition unit and the iris recognition unit automatically switch working modes according to changes in environmental parameters to ensure high-precision identity recognition under different environmental conditions; An intelligent switching mechanism is used to automatically select an appropriate biometric recognition mode based on preset environmental parameter thresholds, including but not limited to light intensity, dust concentration, and whether the person is wearing a helmet. The intelligent switching mechanism includes: The environmental monitoring unit is used to monitor the light intensity and dust concentration in real time. The environmental data collected by the environmental monitoring unit is transmitted to the control system. When the monitored light intensity is lower than the preset threshold or the dust concentration exceeds the preset threshold, the control system triggers the iris recognition mode through an intelligent switching mechanism. When the lighting conditions improve or the dust concentration drops to a normal range, the system automatically switches back to the face recognition mode. The adaptive image acquisition device comprises: a camera with adjustable parameters, which is used to adjust the exposure time, aperture size and gain value of the camera according to the real-time light intensity and dust concentration to optimize the image acquisition quality; an automatic cleaning device, which is used to regularly clean the camera lens to reduce the influence of dust on image acquisition, and the automatic cleaning device cleans the lens surface by periodically spraying cleaning liquid and a scraper; a camera parameter adjustment mechanism adjusts the exposure time by a calculated weight coefficient, and the weight coefficient is obtained by fitting experimental data to ensure the optimization of image clarity; The feature dynamic update module is used to update the facial feature model of the construction workers in real time. The feature dynamic update module includes: a feature extraction unit based on deep learning, which uses a convolutional neural network (such as the ResNet-50 architecture) to extract features from newly collected images; the newly collected features are integrated with the original feature model, and the updated feature model is used for subsequent identity recognition; The fusion weight coefficient is determined by calculating the similarity between the new and old feature models. The weight coefficient value is between 0.5 and 0.8. The specific value is determined through experimental verification to ensure the balanced contribution of the new and old features.

[0006] Integrated control system to coordinate the operation of multiple modules, including: A central processor, for receiving and processing input data from the multimodal biometric module and the adaptive image acquisition device; A storage module for storing the biometric data of construction personnel, environmental parameter thresholds, camera parameter adjustment formulas, and feature fusion formulas; The central processor performs processing according to preset logical rules, and when abnormal light or dust concentration is detected, automatically switches the recognition mode, triggers the automatic cleaning device, or updates the recognition features to ensure stable and efficient operation of the system; The system optimization mechanism is used to optimize system performance based on real-time monitoring data, including: an environmental data analysis unit that analyzes long-term collected environmental data, optimizes switching thresholds and camera parameter adjustment formulas, and ensures the stability and high recognition rate of the system in various construction site environments; a user feedback module that adjusts system parameters through feedback information from construction personnel to further optimize the recognition effects of face recognition and iris recognition.

[0007] Preferably, the environmental parameter monitoring unit of the intelligent switching mechanism further comprises: Light sensor, used to detect the light intensity in the access control area of ​​the construction site in real time and compare the detection value with the preset light threshold; Dust concentration sensor, used to detect the concentration of particles in the air and decide whether to switch to iris recognition mode based on the detection results; The helmet detection module is used to determine whether the construction workers are wearing helmets. If it is detected that they are wearing helmets, the priority of iris recognition is increased to reduce the probability of face recognition failure due to the helmet covering the forehead.

[0008] Preferably, the intelligent switching mechanism adopts the following decision logic when adjusting the recognition mode according to changes in different construction site environments: when the light value detected by the light sensor is lower than 200 lux and the PM2.5 concentration detected by the dust concentration sensor exceeds 150 µg / m³, the system automatically switches to the iris recognition mode; when the light intensity is greater than 300 lux and the dust concentration is lower than 100 µg / m³, the system prioritizes the face recognition mode; when it is detected that the construction workers are wearing safety helmets or masks, the system adopts a mode based on the joint verification of local facial feature matching and iris recognition to improve the recognition accuracy.

[0009] Preferably, the automatic cleaning device of the adaptive image acquisition device includes: a rotating scraper structure, which is used to regularly clean the camera lens to reduce the impact of dust or water mist; a spray system, which is used to spray a small amount of cleaning liquid when the dust concentration is too high to remove dust from the lens; a fan dust removal system, which is used to blow away particulate matter attached to the camera through high-pressure airflow to ensure the clarity of image acquisition.

[0010] Preferably, the camera parameter adjustment method of the adaptive image acquisition device includes: automatic exposure adjustment, by adjusting the exposure time T to adapt to different lighting conditions, the exposure time adjustment formula is: in, is the standard exposure time, is the reference light intensity, is the currently detected light intensity; Automatic gain adjustment is used to improve image brightness in low light environments. The gain adjustment formula is: in, is the gain value, and the gain adjustment range is set between 1.0 and 3.0 to ensure stable image quality.

[0011] Preferably, the feature fusion method of the feature dynamic update module includes: Based on the feature fusion algorithm of weight adjustment, the weight coefficient of the new feature is set With the original feature weight The relationship between them satisfies: in, is the adjustment factor, the range is set from 0.5 to 0.8, is the newly extracted feature vector; The data samples are dynamically expanded, and the recognition accuracy is improved by storing the effective features identified each time in the database and regularly updating the model parameters.

[0012] Preferably, the feature dynamic update module adopts an adaptation algorithm based on a neural network, specifically including: ResNet-50 convolutional neural network is used for feature extraction, and feature mean calibration algorithm is used to reduce feature drift caused by environmental changes; A real-time feedback mechanism is adopted. When the system recognition failure rate is higher than 5%, the feature templates in the database are automatically updated to adapt to changes in the construction environment.

[0013] Preferably, the integrated control system further comprises: Log storage module, used to record all recognition results, including successful and failed recognition data, for subsequent analysis and optimization; The remote management terminal is used to connect to the management platform via wireless communication (Wi-Fi or 5G) to achieve remote monitoring and parameter adjustment of the system.

[0014] Preferably, the remote management terminal provides the following functions: System operation status monitoring, including camera working status, biometric module operation and automatic cleaning device start and stop status; Remote data synchronization is used to synchronize the identification logs of the construction site access control system to the cloud server for big data analysis and remote management.

[0015] Preferably, the system optimization mechanism further includes: optimizing environmental adaptability based on historical data, regularly counting changes in the construction site environment, and dynamically adjusting the illumination threshold and recognition strategy; The user feedback mechanism allows construction workers to submit recognition issues on the remote management terminal. The system optimizes the fusion strategy of face recognition and iris recognition through feedback data to improve adaptability.

[0016] Compared with the problems described in the background technology, the beneficial effects of the present invention are: in view of the large changes in the construction site environment, the complex clothing of the personnel, and the sweat and dust that are easy to mix and dirty on the workers' faces, through the organic combination of face recognition and iris recognition, combined with the intelligent switching mechanism, the recognition mode is adjusted in real time according to the actual environmental conditions (such as light intensity, dust concentration, etc.). Compared with the traditional single recognition mode, the system can maintain a stable high recognition rate in a complex environment, especially in a construction site environment with large changes in dust or light, relying on the dynamic selection of the most suitable recognition mode, significantly improving the accuracy of identity authentication and the adaptability of the system. This dynamic switching based on environmental changes not only effectively copes with the limitations of traditional face recognition modes in harsh environments, but also ensures extremely strong stability and reliability in recognition accuracy, thereby enhancing the flexibility and intelligence of the overall system; and combined with the feature dynamic update module of the deep learning algorithm, the solution can continuously optimize the facial recognition model through new features collected in real time, and maintain high recognition accuracy in long-term use. In a changing construction site environment, due to changes in facial features of personnel and dust occlusion, the recognition rate of conventional static models will gradually decrease. By continuously updating the feature model and adjusting its weight coefficient, the system can adapt to the changing environment and reduce misidentification caused by individual differences in personnel or working environment. This function based on real-time deep learning updates enables the system to still accurately identify in a dynamically changing environment, thereby greatly improving the robustness and long-term stability of the system; by automatically adjusting the exposure time and gain, combined with the fan dust removal system and automatic cleaning mechanism, it can be adjusted in real time for different construction site environments, thereby ensuring that the camera can always provide high-quality images in different environments. This design solves problems such as light changes and dust pollution, improves the accuracy of image acquisition, and greatly reduces equipment failure rate and maintenance costs. Through these dynamic adaptation mechanisms, the system can still maintain excellent recognition capabilities under more complex environmental conditions, further improving the adaptability and stability of the entire access control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the construction site access control face recognition management system based on dynamic identity authentication of the present invention.

[0018] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0020] The embodiment of the present application provides a construction site access control face recognition management system based on dynamic identity authentication. It includes: a multimodal biometric recognition module, which is used to selectively perform identity authentication according to environmental conditions, and the multimodal biometric recognition module includes: a face recognition unit, which is used to collect and identify the face image of the construction personnel; an iris recognition unit, which is used to collect and identify the iris image of the construction personnel; Among them, the face recognition unit and the iris recognition unit automatically switch working modes according to changes in environmental parameters to ensure high-precision identity recognition under different environmental conditions; An intelligent switching mechanism is used to automatically select an appropriate biometric recognition mode based on preset environmental parameter thresholds, including but not limited to light intensity, dust concentration, and whether the person is wearing a helmet. The intelligent switching mechanism includes: The environmental monitoring unit is used to monitor the light intensity and dust concentration in real time. The environmental data collected by the environmental monitoring unit is transmitted to the control system. When the monitored light intensity is lower than the preset threshold or the dust concentration exceeds the preset threshold, the control system triggers the iris recognition mode through an intelligent switching mechanism. When the lighting conditions improve or the dust concentration drops to a normal range, the system automatically switches back to the face recognition mode. The adaptive image acquisition device comprises: a camera with adjustable parameters, which is used to adjust the exposure time, aperture size and gain value of the camera according to the real-time light intensity and dust concentration to optimize the image acquisition quality; an automatic cleaning device, which is used to regularly clean the camera lens to reduce the influence of dust on image acquisition, and the automatic cleaning device cleans the lens surface by periodically spraying cleaning liquid and a scraper; a camera parameter adjustment mechanism adjusts the exposure time by a calculated weight coefficient, and the weight coefficient is obtained by fitting experimental data to ensure the optimization of image clarity; The feature dynamic update module is used to update the facial feature model of the construction workers in real time. The feature dynamic update module includes: a feature extraction unit based on deep learning, which uses a convolutional neural network (such as the ResNet-50 architecture) to extract features from newly collected images; the newly collected features are integrated with the original feature model, and the updated feature model is used for subsequent identity recognition; The fusion weight coefficient is determined by calculating the similarity between the new and old feature models. The weight coefficient value is between 0.5 and 0.8. The specific value is determined through experimental verification to ensure the balanced contribution of the new and old features.

[0021] Integrated control system to coordinate the operation of multiple modules, including: A central processor, for receiving and processing input data from the multimodal biometric module and the adaptive image acquisition device; A storage module for storing the biometric data of construction personnel, environmental parameter thresholds, camera parameter adjustment formulas, and feature fusion formulas; The central processor performs processing according to preset logical rules, and when abnormal light or dust concentration is detected, automatically switches the recognition mode, triggers the automatic cleaning device, or updates the recognition features to ensure stable and efficient operation of the system; The system optimization mechanism is used to optimize system performance based on real-time monitoring data, including: an environmental data analysis unit that analyzes long-term collected environmental data, optimizes switching thresholds and camera parameter adjustment formulas, and ensures the stability and high recognition rate of the system in various construction site environments; a user feedback module that adjusts system parameters through feedback information from construction personnel to further optimize the recognition effects of face recognition and iris recognition.

[0022] Preferably, the environmental parameter monitoring unit of the intelligent switching mechanism further comprises: Light sensor, used to detect the light intensity in the access control area of ​​the construction site in real time and compare the detection value with the preset light threshold; Dust concentration sensor, used to detect the concentration of particles in the air and decide whether to switch to iris recognition mode based on the detection results; The helmet detection module is used to determine whether the construction workers are wearing helmets. If it is detected that they are wearing helmets, the priority of iris recognition is increased to reduce the probability of face recognition failure due to the helmet covering the forehead.

[0023] Preferably, the intelligent switching mechanism adopts the following decision logic when adjusting the recognition mode according to changes in different construction site environments: when the light value detected by the light sensor is lower than 200 lux and the PM2.5 concentration detected by the dust concentration sensor exceeds 150 µg / m³, the system automatically switches to the iris recognition mode; when the light intensity is greater than 300 lux and the dust concentration is lower than 100 µg / m³, the system prioritizes the face recognition mode; when it is detected that the construction workers are wearing safety helmets or masks, the system adopts a mode based on the joint verification of local facial feature matching and iris recognition to improve the recognition accuracy.

[0024] Preferably, the automatic cleaning device of the adaptive image acquisition device includes: a rotating scraper structure, which is used to regularly clean the camera lens to reduce the impact of dust or water mist; a spray system, which is used to spray a small amount of cleaning liquid when the dust concentration is too high to remove dust from the lens; a fan dust removal system, which is used to blow away particulate matter attached to the camera through high-pressure airflow to ensure the clarity of image acquisition.

[0025] Preferably, the camera parameter adjustment method of the adaptive image acquisition device includes: automatic exposure adjustment, by adjusting the exposure time T to adapt to different lighting conditions, the exposure time adjustment formula is: in, is the standard exposure time, is the reference light intensity, is the currently detected light intensity; Automatic gain adjustment is used to improve image brightness in low light environments. The gain adjustment formula is: in, is the gain value, and the gain adjustment range is set between 1.0 and 3.0 to ensure stable image quality.

[0026] Preferably, the feature fusion method of the feature dynamic update module includes: setting the weight coefficient of the new feature based on the feature fusion algorithm adjusted by weight With the original feature weight The relationship between them satisfies: in, is the adjustment factor, the range is set from 0.5 to 0.8, is the newly extracted feature vector; The data samples are dynamically expanded, and the recognition accuracy is improved by storing the effective features identified each time in the database and regularly updating the model parameters.

[0027] Preferably, the feature dynamic update module adopts an adaptation algorithm based on a neural network, specifically including: ResNet-50 convolutional neural network is used for feature extraction, and feature mean calibration algorithm is used to reduce feature drift caused by environmental changes; A real-time feedback mechanism is adopted. When the system recognition failure rate is higher than 5%, the feature templates in the database are automatically updated to adapt to changes in the construction environment.

[0028] Preferably, the integrated control system further comprises: Log storage module, used to record all recognition results, including successful and failed recognition data, for subsequent analysis and optimization; The remote management terminal is used to connect to the management platform via wireless communication (Wi-Fi or 5G) to achieve remote monitoring and parameter adjustment of the system.

[0029] Preferably, the remote management terminal provides the following functions: system operation status monitoring, including the camera working status, the operation status of the biometric module and the start and stop status of the automatic cleaning device; remote data synchronization, used to synchronize the identification log of the construction site access control system to the cloud server for big data analysis and remote management.

[0030] Preferably, the system optimization mechanism further includes: environmental adaptability optimization based on historical data, regular statistics of site environment changes, and dynamic adjustment of lighting thresholds and recognition strategies; user feedback mechanism, allowing construction personnel to submit recognition problems on the remote management terminal, and the system optimizes the fusion strategy of face recognition and iris recognition through feedback data to improve adaptability.

[0031] Example 1: This example demonstrates how to dynamically switch identity authentication modes according to complex environmental changes to ensure efficient and accurate identity authentication, especially in construction environments with poor lighting and high dust concentrations. This example applies the system to a typical construction site, where environmental conditions such as dust, lighting, and construction workers wearing helmets dynamically affect the selection of identity authentication modes, thereby solving the recognition problem of traditional systems in complex environments.

[0032] In actual applications, construction workers frequently come and go on construction sites. Environmental factors such as strong sunlight, dust concentration, and workers wearing protective equipment such as helmets and masks make it impossible for traditional face recognition technology to guarantee accurate recognition. At this time, the intelligent switching mechanism in this system automatically switches to iris recognition mode. For example, when the light intensity in the environment is less than 200 lux and the dust concentration exceeds 150 µg / m³, the system will automatically select iris recognition to replace face recognition, ensuring that the identity of workers can still be accurately identified in low light and high dust conditions.

[0033] On the other hand, when environmental conditions improve, such as when the light intensity exceeds 300 lux and the dust concentration drops to 100 µg / m³, the system automatically resumes face recognition mode. In addition, if the construction worker wears a helmet or mask, the system will combine local facial feature matching with iris recognition for joint verification to further improve the accuracy and reliability of recognition.

[0034] To support all this, the system also includes an adaptive image acquisition device that can adjust the camera's exposure time, aperture size and gain according to the real-time monitoring of light intensity and dust concentration to ensure that high-quality images can always be acquired in complex environments. The automatic cleaning device regularly cleans the camera lens to reduce the impact of dust on image acquisition, while the fan dust removal system and spray cleaning mechanism can effectively prevent the impact of dust and dirt on recognition accuracy.

[0035] The dynamic feature update module uses a deep learning algorithm to extract the newly acquired image features in real time through a convolutional neural network (such as the ResNet-50 architecture) and fuse them with the original features. The updated feature model will be continuously optimized to adapt to changes in the construction site environment and personnel. This not only improves recognition accuracy, but also avoids the performance degradation of static models in long-term use.

[0036] For example, if a construction worker works at a specific construction site for a long time, the old static feature model may no longer be accurate due to changes in the environment (such as dust accumulation on the face). At this time, the deep learning algorithm automatically adjusts the updated feature model, allowing the system to adapt to the changing construction site environment in real time during the actual construction process, ensuring the recognition accuracy and stability of the system. Through these intelligent adjustment mechanisms, the system can not only dynamically adjust the recognition mode under different environmental conditions, but also maintain a high recognition rate and accuracy, greatly improving the stability and availability of the system, especially at construction sites in complex environments.

[0037] Example 2: This embodiment selected a traditional single face recognition system for comparison with the multimodal recognition system of the present invention in a comparative experiment. The experiment was conducted in multiple construction site environments, including harsh environments with light intensity less than 200 lux and dust concentration exceeding 150 µg / m³, and ideal environments with light intensity exceeding 300 lux and low dust concentration. The experimental results show that when the light intensity is too low and the dust concentration is high, the recognition success rate of the traditional single face recognition system is only 60%, while after the system of the present invention adopts iris recognition, the recognition success rate is increased to 92%. Under normal conditions, the recognition accuracy of the traditional face recognition system is 90%, while the recognition success rate of the present system combined with the intelligent switching mechanism is 98% in the face recognition mode. By comparing the data, it can be clearly seen that the system of the present invention significantly improves the accuracy of identity authentication in complex construction site environments. This advantage is mainly due to the intelligent switching mechanism of the present system. When the light intensity is too low or the dust concentration is too high, the system can automatically switch to the iris recognition mode in time, avoiding the limitations of traditional face recognition technology in these extreme environments.

[0038] In practical applications, the impact of construction workers wearing protective equipment such as helmets and masks on face recognition technology has also been effectively addressed. In this system, when it is detected that a worker is wearing a helmet, the system will give priority to the iris recognition mode to avoid recognition failures caused by the helmet blocking facial features. In addition, by combining the mode of local facial feature matching and iris recognition joint verification, the recognition success rate is increased by more than 10% compared with the single mode when wearing protective equipment. The automatic adjustment and cleaning functions of the image acquisition device significantly improve the recognition effect of the system in environments with high dust concentrations. In the experiment, we observed that when the dust concentration was higher than 150 µg / m³, the system equipped with an automatic cleaning device was able to maintain a high recognition accuracy, while the system without such a device had a significantly lower recognition rate due to the degradation of image quality caused by dust accumulation on the lens.

[0039] Furthermore, through the feature dynamic update module, the system can continuously optimize the facial recognition model through new features collected in real time. Experimental results show that after the system is updated after long-term use, the accuracy of the facial recognition model is improved by 7%, which is more adaptable than the traditional static model. The present invention not only improves the recognition accuracy, but also significantly improves the stability and reliability in complex construction site environments. Through the dynamic adjustment of the intelligent switching mechanism and the dynamic update of features supported by the deep learning algorithm, the system can adapt to changes in the construction site environment, effectively reduce the false recognition rate and missed recognition rate, and ensure the efficient operation of the access control management system in harsh environments.

[0040] Example 3: This example provides an optimization method for a construction site access control face recognition management system based on dynamic identity authentication, particularly improving the system performance, environmental adaptability optimization, and recognition accuracy mentioned in the previous implementation plan.

[0041] The module includes a face recognition unit and an iris recognition unit. According to environmental conditions, the system monitors the environment in real time through the environmental monitoring unit (light intensity and dust concentration sensor) and automatically switches the recognition mode. For example, when the light intensity is less than 200 lux, it automatically switches to the iris recognition mode; when the light intensity is higher than 300 lux, the face recognition mode is preferred. The system can work stably in various complex environments to ensure high-precision identity authentication. The intelligent switching mechanism adjusts the recognition mode in real time by monitoring changes in the construction site environment. Environmental data (light, dust concentration, helmet wearing) are transmitted to the integrated control system in real time, and the most appropriate recognition method is automatically selected through decision logic. If the light intensity is detected to be less than 200 lux and the dust concentration is more than 150 µg / m³, the system will automatically start the iris recognition mode to ensure high recognition accuracy even in dusty environments.

[0042] In order to obtain high-quality images under various environmental conditions, this system introduces an adaptive image acquisition device. The device includes a camera with adjustable exposure time, aperture size and gain, which can adjust shooting parameters according to real-time lighting and dust concentration to ensure image acquisition quality. In addition, the automatic cleaning device regularly cleans the camera lens to reduce the impact of dust accumulation on the recognition effect. The cleaning device periodically sprays cleaning fluid and scrapers to clean the lens surface, effectively improving image quality. The feature dynamic update module extracts newly acquired image features in real time through a feature extraction unit based on deep learning, and fuses them with the existing feature model. The feature update uses a convolutional neural network (such as the ResNet-50 architecture) and uses similarity calculation to determine the fusion weight coefficient to ensure the balance of new and old features. This module can update the feature model in real time, improve recognition accuracy, and avoid the problem of accuracy degradation caused by static models.

[0043] In this embodiment, through the precise environmental adaptability optimization algorithm, the system can adjust the recognition mode and camera parameters according to the environmental data collected over a long period of time. For example, the system adjusts the switching threshold by analyzing the light intensity and dust concentration data in real time. Through the feedback optimization of historical data, the system can continuously improve the adaptability and accuracy of recognition, especially in the changing construction site environment, the system can flexibly respond to various environmental conditions.

[0044] The system uses a specific algorithm to adjust the dynamic recognition mode. When the light value is lower than the preset 200 lux and the dust concentration exceeds 150 µg / m³, the system will automatically switch to iris recognition mode. When the light is strong or the dust concentration is reduced, the system will switch back to face recognition mode. In addition, when construction workers wear helmets or masks, the system will use a mode of joint verification of local facial features and iris recognition to improve recognition accuracy. In terms of image acquisition, the system automatically adjusts the exposure time, gain and aperture to ensure the clarity of the image under different lighting conditions. The adjustment formula for exposure time T is as follows:

[0045] in, is the standard exposure time, is the reference light intensity, is the light intensity of the current environment. The adjustment formula of the gain value G is:

[0046] The adjustment range of the gain value G is set between 1.0 and 3.0 to ensure stable image brightness. When the system detects that the environmental parameters (light intensity, dust concentration, etc.) reach the preset threshold, the most suitable biometric recognition mode is automatically selected. In the specific implementation process, the environmental monitoring unit obtains data in real time through the light sensor and the dust concentration sensor and transmits it to the control system. The control system automatically determines whether it is necessary to switch the recognition mode according to the algorithm, and controls the camera parameters to adjust. In an environment with high dust concentration, the automatic cleaning device is activated to ensure the cleanliness of the camera lens, thereby maintaining the clarity of image acquisition. The cleaning device combines a rotating scraper and a spray cleaning system to reduce the possibility of dust and stains affecting the recognition accuracy. The system updates the feature model in real time through a deep learning algorithm, uses a convolutional neural network to extract features from new images, and fuses them with the existing model. The updated model will be stored in the database and used for subsequent identity recognition. The feature fusion process is optimized through a weight adjustment algorithm to ensure the balance between new and old features and improve the recognition stability in long-term use.

[0047] In summary, this embodiment has made key technical improvements in system performance, environmental adaptability, and image acquisition quality, ensuring that the system can efficiently and accurately complete identity authentication tasks in complex environments such as construction sites, thereby improving recognition rate and stability.

[0048] Embodiment 4: This embodiment forms a complete identity authentication process through the collaborative work of a multimodal biometric module, an intelligent switching mechanism, an adaptive image acquisition device, a feature dynamic update module and an integrated control system. The functions and effects of each module are as follows: Face recognition unit: uses a high-definition camera to collect facial images of construction workers, and extracts and matches facial features through a pre-trained convolutional neural network. Iris recognition unit: When the lighting conditions are poor or the dust concentration is high, the system automatically switches to iris recognition mode to ensure high-precision identity authentication.

[0049] Intelligent switching mechanism: Real-time monitoring of the construction site environment through the environmental monitoring unit (light intensity sensor, dust concentration sensor). When the light intensity is lower than 200 lux or the dust concentration exceeds 150 µg / m³, the system automatically switches to iris recognition mode; when the environmental conditions return to the normal range, the system automatically switches back to face recognition mode. In the adaptive image acquisition device, exposure time and gain adjustment: The automatic exposure and gain adjustment of the camera ensures that the image quality is not affected by light changes and dust. The specific adjustment formula is as follows: Exposure time formula: in, is the standard exposure time, is the reference light intensity, is the currently detected light intensity.

[0050] Gain adjustment formula: in, is the gain value, is the reference light intensity, The light intensity of the current environment. The gain value is set between 1.0 and 3.0 to ensure stable image quality.

[0051] Feature dynamic update module: A feature extraction unit based on deep learning uses the ResNet-50 convolutional neural network to extract features from newly acquired images and fuse them with the original feature model. The fusion weight coefficient is calculated using the following formula: in, is the weight coefficient of the new feature, is the weight coefficient of the original feature, is the newly extracted feature vector, The adjustment factor has a value range of 0.5 to 0.8. The specific value is determined through experimental verification. For example, the weight coefficient α in feature fusion can be determined in the following way: Use the system to collect multiple sets of recognition data in different construction site environments. These data include different lighting conditions, dust concentrations, and workers wearing helmets and masks. The collected face images and iris images will be used for feature extraction and identity authentication through the existing recognition model.

[0052] Experimental design: By setting different environmental parameters (for example, low light, high dust, wearing a helmet, etc.), multiple rounds of experiments are conducted to test the recognition accuracy. Compare the recognition effects in different environments and observe the impact of the weight coefficient (α) on the recognition accuracy when fusing new and old features. Optimization of weight coefficient: In the experiment, by adjusting the value of α, observe the recognition success rate of the system in different environments. For example, adjust the value of α from 0.5 to 0.8, and record the recognition accuracy in each experimental environment. Through statistical analysis (such as indicators such as mean square error, accuracy, and recall), select the α value that maximizes the recognition accuracy. The optimal value obtained will be recorded and used as a standard for subsequent model training and practical applications.

[0053] Integrated control system: The central processor is responsible for receiving and processing the input data of the multimodal biometric module and the adaptive image acquisition device to ensure the stable and efficient operation of the system. The storage module is used to store the biometric data of construction personnel, environmental parameter thresholds, camera parameter adjustment formulas, and feature fusion formulas. And the system optimization mechanism: optimize the system performance according to the real-time monitoring data, including adjusting the switching threshold and camera parameter adjustment formula to ensure the stability and high recognition rate of the system in various construction site environments.

[0054] Example 5: The core function of this system is to provide high-precision identity authentication. In actual construction site applications, environmental conditions such as light intensity, dust concentration, and personnel clothing will significantly affect the recognition effect. Therefore, this system uses multimodal biometric technology (including face recognition and iris recognition) to cope with different environmental changes and ensure that efficient and accurate identity authentication is always provided.

[0055] During the workflow, the intelligent switching mechanism automatically selects the most suitable recognition mode by monitoring the changes in the site environment in real time. When the light intensity or dust concentration reaches the preset threshold, the system automatically switches to iris recognition mode, and automatically returns to face recognition mode when the conditions return to normal range. To this end, we have added an environmental monitoring unit, including a light sensor and a dust concentration sensor. These sensors collect data in real time and transmit it to the integrated control system to ensure that environmental changes are responded to immediately.

[0056] To further improve the quality of image acquisition, the system is equipped with an adaptive image acquisition device, which includes a camera with adjustable parameters and an automatic cleaning device. The camera's exposure time, gain, and aperture size are dynamically adjusted according to the real-time light intensity and dust concentration. For example, in low light conditions, the camera will automatically increase the exposure time and gain value to improve image brightness and clarity. The automatic cleaning device regularly cleans the camera lens to prevent dust from affecting image quality.

[0057] In this embodiment, we extract features from newly acquired images through deep learning algorithms and convolutional neural networks (such as the ResNet-50 architecture) and update the facial feature model in real time. This model can be continuously optimized as the construction site environment changes and the facial features of construction workers change, thereby improving recognition accuracy in long-term use. To achieve this goal, we introduced a feature fusion algorithm to ensure that newly acquired features can be effectively integrated into the existing model by adjusting the weight coefficients of the new and old feature models.

[0058] As for the exposure time adjustment of the camera, the following logic can be used: the standard value of the exposure time is set according to the reference light intensity obtained during the initial debugging of the system. If the current light intensity is lower than the standard value, the system will compensate for the lack of light by increasing the exposure time. The specific steps for adjusting the exposure time are: calculate the ratio of the current light intensity to the reference light intensity, and adjust the exposure time according to this ratio.

[0059] Similarly, the core logic of gain adjustment is based on the comparison between the current light intensity and the reference light intensity. If the ambient light is insufficient, the system will increase the gain value to increase the image brightness. The gain adjustment range is set between 1.0 and 3.0 to ensure that the image brightness adapts to different lighting conditions. At the same time, to ensure that the scheme is highly operational, this embodiment describes in detail the working steps and implementation methods of each module, especially in terms of dynamically adjusting exposure and gain, switching recognition modes, and feature updates. Specific steps include: environmental monitoring: through real-time monitoring of light sensors and dust concentration sensors, accurate collection of environmental data is ensured; switching mechanism: the intelligent control system makes judgments based on the collected data and automatically switches to the most suitable recognition mode; image acquisition and cleaning: dynamically adjust the various parameters of the camera, and use the automatic cleaning system to regularly clean the camera lens to reduce the impact of dust; feature update: use deep learning algorithms to extract new features in real time, and integrate and update them with existing feature models.

[0060] The innovation of this embodiment is that it proposes a solution to automatically switch the recognition mode based on the changes in the construction site environment. By combining multimodal biometrics and intelligent switching mechanisms, it ensures that high-precision identity authentication is always maintained in different environments. Through the feature dynamic update module, the system can continuously adapt to long-term environmental changes and avoid the problem of accuracy loss caused by static models. In addition, the automatic adjustment of exposure and gain, as well as the cleaning mechanism of the camera lens, effectively improve the image acquisition quality, further ensuring the system stability and recognition accuracy.

[0061] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A construction site access control face recognition management system based on dynamic identity authentication, characterized in that: include: A multimodal biometric module, used to selectively perform identity authentication according to environmental conditions, the multimodal biometric module comprising: a face recognition unit, used to collect and recognize the face image of the construction worker; an iris recognition unit, used to collect and recognize the iris image of the construction worker; Among them, the face recognition unit and the iris recognition unit automatically switch working modes according to changes in environmental parameters; An intelligent switching mechanism is used to automatically select an appropriate biometric recognition mode based on preset environmental parameter thresholds, including but not limited to light intensity, dust concentration, and whether the person is wearing a helmet. The intelligent switching mechanism includes: The environmental monitoring unit is used to monitor the light intensity and dust concentration in real time. The environmental data collected by the environmental monitoring unit is transmitted to the control system. When the monitored light intensity is lower than the preset threshold or the dust concentration exceeds the preset threshold, the control system triggers the iris recognition mode through an intelligent switching mechanism. When the lighting conditions improve or the dust concentration drops to a normal range, the system automatically switches back to the face recognition mode. The adaptive image acquisition device comprises: a camera with adjustable parameters, used to adjust the exposure time, aperture size and gain value of the camera according to the real-time light intensity and dust concentration to optimize the image acquisition quality; an automatic cleaning device, used to regularly clean the camera lens to reduce the influence of dust on image acquisition, and the automatic cleaning device cleans the lens surface by periodically spraying cleaning fluid and a scraper; The feature dynamic update module is used to update the facial feature model of the construction workers in real time. The feature dynamic update module includes: a feature extraction unit based on deep learning, which uses a convolutional neural network (such as the ResNet-50 architecture) to extract features from newly collected images; the newly collected features are integrated with the original feature model, and the updated feature model is used for subsequent identity recognition; Integrated control system to coordinate the operation of multiple modules, including: A central processor, for receiving and processing input data from the multimodal biometric module and the adaptive image acquisition device; A storage module for storing the biometric data of construction personnel, environmental parameter thresholds, camera parameter adjustment formulas, and feature fusion formulas; The central processor performs processing according to preset logical rules, and when abnormal light or dust concentration is detected, automatically switches the recognition mode, triggers the automatic cleaning device, or updates the recognition features to ensure stable and efficient operation of the system; The system optimization mechanism is used to optimize system performance based on real-time monitoring data, including: an environmental data analysis unit that analyzes long-term collected environmental data, optimizes switching thresholds and camera parameter adjustment formulas, and ensures the stability and high recognition rate of the system in various construction site environments; a user feedback module that adjusts system parameters through feedback information from construction personnel to further optimize the recognition effects of face recognition and iris recognition.

2. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 1, characterized in that: The environmental parameter monitoring unit of the intelligent switching mechanism further includes: a light sensor, which is used to detect the light intensity in the access control area of ​​the construction site in real time and compare the detection value with the preset light threshold; a dust concentration sensor, which is used to detect the concentration of particulate matter in the air and decide whether to switch to iris recognition mode based on the detection results; a safety helmet detection module, which is used to determine whether the construction workers are wearing safety helmets. If it is detected that they are wearing safety helmets, the priority of iris recognition is increased.

3. The construction site access control face recognition management system based on dynamic identity authentication as described in claim 2, characterized in that: The intelligent switching mechanism uses the following decision logic to adjust the recognition mode according to changes in different construction site environments: when the light value detected by the light sensor is lower than 200 lux and the PM2.5 concentration detected by the dust concentration sensor exceeds 150 µg / m³, the system automatically switches to iris recognition mode; when the light intensity is greater than 300 lux and the dust concentration is lower than 100 µg / m³, the system prioritizes face recognition mode; when it is detected that construction workers are wearing helmets or masks, the system adopts a mode based on joint verification of local facial feature matching and iris recognition.

4. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 1, characterized in that: The automatic cleaning device of the adaptive image acquisition device includes: a rotating scraper structure, which is used to regularly clean the camera lens to reduce the impact of dust or water mist; a spray system, which is used to spray a small amount of cleaning liquid when the dust concentration is too high to remove dust on the lens; a fan dust removal system, which is used to blow away particulate matter attached to the camera through high-pressure airflow.

5. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 4, characterized in that: The camera parameter adjustment method of the adaptive image acquisition device includes: automatic exposure adjustment, by adjusting the exposure time T to adapt to different lighting conditions, the exposure time adjustment formula is: in, is the standard exposure time, is the reference light intensity, is the currently detected light intensity; Automatic gain adjustment is used to improve image brightness in low light environments. The gain adjustment formula is: in, is the gain value, and the gain adjustment range is between 1.0 and 3.

0.

6. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 1, characterized in that: The feature fusion methods of the feature dynamic update module include: feature fusion algorithm based on weight adjustment, setting the weight coefficient of the new feature With the original feature weight The relationship between them satisfies: in, is the adjustment factor, the range is set from 0.5 to 0.8, is the newly extracted feature vector; the data samples are dynamically expanded by storing the effective features identified each time into the database and updating the model parameters regularly.

7. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 6, characterized in that: The feature dynamic update module adopts a neural network-based adaptation algorithm, which includes: using the ResNet-50 convolutional neural network for feature extraction, and reducing feature drift caused by environmental changes through a feature mean calibration algorithm; using a real-time feedback mechanism to automatically update the feature template in the database when the system recognition failure rate is higher than 5%.

8. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 1, characterized in that: The integrated control system further includes: a log storage module for recording all recognition results, including successful and failed recognition data, for subsequent analysis and optimization; a remote management terminal for connecting to a management platform via wireless communication to achieve remote monitoring and parameter adjustment of the system.

9. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 8, characterized in that: The remote management terminal provides the following functions: system operation status monitoring, including the working status of the camera, the operation of the biometric module, and the start and stop status of the automatic cleaning device; remote data synchronization, used to synchronize the identification log of the construction site access control system to the cloud server.

10. The construction site access control face recognition management system based on dynamic identity authentication as claimed in claim 1, characterized in that: The system optimization mechanism further includes: environmental adaptability optimization based on historical data, regular statistics of site environment changes, and dynamic adjustment of lighting thresholds and recognition strategies; user feedback mechanism, allowing construction personnel to submit recognition issues on the remote management terminal, and the system optimizes the fusion strategy of face recognition and iris recognition through feedback data.

Images