A high-precision face recognition system for face recognition access control

Abstract

The present invention discloses a high-precision face recognition system for face recognition access control, and relates to the technical field of face recognition access control. The high-precision face recognition system for face recognition access control includes a recognition access control self-check module, a matching information acquisition module, an identity recognition module, a face recognition module and a statistical module. Through double verification (time period + face recognition), it is ensured that only the right people visit a specific area at the right time, which greatly improves the security. The self-check module ensures the stable operation of the system. Once a problem is found, it can respond quickly to avoid security loopholes. The statistical module provides detailed access records, which is convenient for post-audit and behavior analysis, and helps to improve the overall supervision ability.

Description

Technical Field

[0001] The present invention relates to the technical field of face recognition access control, and in particular to a high-precision face recognition system for face recognition access control. Background Art

[0002] The entrance and exit door security management system is a new modern security management system, which integrates microcomputer automatic identification technology and modern security management measures. It involves many new technologies such as electronics, machinery, optics, computer technology, communication technology, biotechnology, etc. It is an effective measure to solve the security management of important entrances and exits. It is applicable to buildings, schools, banks, hotels, parking lot management, computer rooms, offices, intelligent communities, factories, etc. The access control system has long surpassed the simple doorway and key management, and it has gradually developed into a complete access management system. It plays a major role in administrative management work such as living environment safety and personnel attendance management.

[0003] The Chinese invention patent with announcement number CN109801420B discloses a multi-concurrent face recognition access control system based on a classification algorithm and its recognition method. The method includes: all user data containing facial feature information and identity information are stored in multiple distributed computing servers in units of users, and all distributed computing servers are uniformly scheduled by a scheduling platform server to recognize the facial feature information sent by multiple gates; when the scheduling platform server receives the facial feature information sent by the gate, the facial feature information is preferentially allocated to the distributed computing server that has the most successful matches with the facial feature information sent by the gate for face recognition; the present invention realizes rapid face recognition of a large-capacity user group through a classification algorithm and multi-concurrent recognition task processing, and has the advantages of user data expansion without server shutdown and continuous self-optimization of face recognition algorithm.

[0004] However, existing technologies are inconvenient to perform multiple authentications to ensure that only the right people access specific areas at the right time, resulting in insufficient security. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a high-precision face recognition system for face recognition access control, which solves the problem of the inconvenience of multiple authentication in the prior art, ensuring that only the right people access specific areas at the right time and the lack of security.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a high-precision face recognition system for face recognition access control, including a recognition access control self-check module, a matching information acquisition module, an identity recognition module, a face recognition module and a statistical module, wherein: the recognition access control self-check module is used to perform self-check on the face recognition access control to determine whether the face recognition access control can perform face recognition normally in the current time period. If normal face recognition cannot be performed, a prompt is sent to the control center for manual recognition. If normal face recognition can be performed, self-checking continues; the matching information acquisition module is used to obtain visitor information from a database when the face recognition access control can perform normal face recognition, and the visitor information includes a visitor facial feature data set, an identity information set and an allowed access time period corresponding to each identity information in the identity information set; the identity recognition module is used to obtain the visitor's identity to be verified. The system is used to identify the visitor based on the visitor's identity information to be verified, the actual visit time, the identity information and the allowed access time period, and determine whether the current visitor can perform facial recognition. If facial recognition is not possible, a prompt is sent to the control center. If facial recognition is possible, a facial recognition module and a statistical module are started; the facial recognition module is used to obtain the actual facial feature data of the visitor who can perform facial recognition, perform facial recognition based on the actual facial feature data and the visitor's facial feature data set, and determine whether the facial recognition is successful. If facial recognition is successful, the current visitor is allowed to pass through the access control. If facial recognition is unsuccessful, a prompt is sent to the control center for manual identification; the statistical module is used to record the number of visitors passing through the access control and the identity information to be verified, the actual visit time and the facial feature data during facial recognition of each visitor who is allowed to enter.

[0007] Furthermore, the determination of whether the face recognition access control can perform face recognition normally in the current time period includes the following steps: obtaining the current temperature of the recognition access control and the brightness of the environment in which the recognition access control is located; determining whether the brightness of the environment in which the recognition access control is located falls within the allowable brightness range stored in the database: if the brightness of the environment in which the recognition access control is located does not fall within the allowable brightness range stored in the database, normal face recognition cannot be performed; if the brightness of the environment in which the recognition access control is located falls within the allowable brightness range stored in the database, then based on the brightness of the environment in which the recognition access control is located, obtaining the recognition access control clarity parameter value corresponding to the brightness of the environment in which the recognition access control is located from the database; obtaining the access control temperature parameter value stored in the database; obtaining the current clarity of the recognition access control; obtaining a compliance value based on the current temperature of the recognition access control, the access control temperature parameter value, the current clarity of the recognition access control and the recognition access control clarity parameter value; determining whether the compliance value is greater than the compliance threshold value stored in the database: if the compliance value is greater than the compliance threshold value stored in the database, face recognition can be performed normally; if the compliance value is not greater than the compliance threshold value stored in the database, face recognition cannot be performed normally.

[0008] Furthermore, the matching value is obtained as follows:

[0009] ;

[0010] In the formula, To match the value, To identify the current temperature of the access control, is the access control temperature parameter value, To identify the current clarity of the access control, To identify the access control clarity parameter value.

[0011] Furthermore, the determination of whether the current visitor can perform facial recognition includes the following steps: determining whether the identity information to be verified belongs to an identity information set, wherein the identity information set includes an established visitor identity information set and a temporary visitor identity information set; if the identity information to be verified does not belong to the identity information set, facial recognition cannot be performed; if the identity information to be verified belongs to the identity information set, determining whether the actual access time falls within the allowed access time period corresponding to the identity information to be verified: if the actual access time falls within the allowed access time period corresponding to the identity information to be verified, facial recognition can be performed; if the actual access time does not fall within the allowed access time period corresponding to the identity information to be verified, determining Determine whether the identity information to be verified belongs to the established visitor identity information set: if the identity information to be verified belongs to the established visitor identity information set, facial recognition can be performed; if the identity information to be verified does not belong to the established visitor identity information set, determine whether the identity information to be verified belongs to the temporary visitor identity information set: if the identity information to be verified belongs to the temporary visitor identity information set, send a prompt to the control center for manual identification; if the result of manual identification is that facial recognition is allowed, facial recognition can be performed; if the result of manual identification is that facial recognition is not allowed, facial recognition cannot be performed; if the identity information to be verified does not belong to the temporary visitor identity information set, facial recognition cannot be performed.

[0012] Furthermore, the determination of whether facial recognition is successful includes the following steps: obtaining actual facial feature data of a visitor who can perform facial recognition, the actual facial feature data including actual geometric features, actual texture features and actual biometric features; obtaining a visitor facial feature data set, the visitor facial feature data set including a plurality of visitor facial feature data, each of which includes a parameterized geometric feature, a parameterized texture feature and a parameterized biometric feature; obtaining feature allowable deviation data during facial recognition, the feature allowable deviation data including a geometric feature allowable deviation value, a texture feature allowable deviation value and a biometric feature allowable deviation value; performing similarity processing on the actual facial feature data and each visitor facial feature data in the visitor facial feature data set, and obtaining a facial recognition matching value in combination with the feature allowable deviation data; obtaining a minimum facial recognition matching value; and determining whether the minimum facial recognition matching value is less than a facial recognition matching threshold stored in a database: if the minimum facial recognition matching value is less than the facial recognition matching threshold stored in the database, then facial recognition is successful; if the minimum facial recognition matching value is not less than the facial recognition matching threshold stored in the database, then facial recognition is unsuccessful.

[0013] Furthermore, the acquisition of feature allowable deviation data during facial recognition includes the following steps: acquiring the identity information to be verified of a visitor who can perform facial recognition; acquiring a feature allowable deviation data set corresponding to the identity information to be verified from a database based on the identity information to be verified, wherein the feature allowable deviation data set includes a number of feature allowable deviation data; acquiring the brightness of the environment in which the access control is located and the current clarity of the access control; acquiring the matching brightness of the environment in which the access control is located and the current matching clarity of the access control stored in the database corresponding to each feature allowable deviation data; comparing the brightness of the environment in which the access control is located and the current clarity of the access control with the matching brightness of the environment in which the access control is located and the current matching clarity of the access control to obtain a comparison coefficient; determining the matching brightness of the environment in which the access control is located and the current matching clarity of the access control corresponding to the minimum comparison coefficient; and obtaining the feature allowable deviation data based on the determined matching brightness of the environment in which the access control is located and the current matching clarity of the access control.

[0014] Furthermore, before obtaining the actual facial feature data of the visitor who can perform facial recognition, it also includes identifying the objects worn by the visitor who can perform facial recognition as determined by the identity recognition module, and judging whether the visitor who can perform facial recognition as determined by the identity recognition module is wearing the required objects. If the required objects are not worn, it is re-determined that facial recognition cannot be performed. If the required objects are worn, the actual facial feature data of the visitor who can perform facial recognition is obtained.

[0015] Further, the determination of whether the visitor who can perform facial recognition determined by the identity recognition module wears the required object includes the following steps: obtaining feature data of the object required to be worn by the visitor who can perform facial recognition, the feature data of the object required to be worn includes the shape of the object, the position of the center of the object relative to the eyes, the color of the object and the texture information of the object; obtaining feature parameter data of the object required to be worn stored in the database, the feature parameter data of the object required to be worn includes the parameterized shape of the object, the parameterized position of the center of the object relative to the eyes, the parameterized color of the object and the parameterized texture information of the object; obtaining the feature parameter data of the object required to be worn stored in the database, the feature parameter data of the object required to be worn includes the parameterized shape of the object, the parameterized position of the center of the object relative to the eyes, the parameterized color of the object and the parameterized texture information of the object; obtaining the feature data of the object required to be worn Feature allowable deviation data, the feature allowable deviation data of the object required to be worn includes the object shape allowable deviation value, the object center relative to the eye position allowable deviation value, the object color allowable deviation value and the object texture information allowable deviation value; based on the feature data of the object required to be worn, the feature parameter data of the object required to be worn and the feature allowable deviation data of the object required to be worn, obtain the object matching coefficient; if the object matching coefficient is less than the object matching threshold coefficient stored in the database, the required object is worn; if the object matching coefficient is not less than the object matching threshold coefficient stored in the database, the required object is not worn.

[0016] Furthermore, the method for obtaining the object matching coefficient is as follows:

[0017] ;

[0018] In the formula, is the object matching coefficient, is the shape of the object, is the position of the object center relative to the eye, is the object color, is the object texture information, Give the object a shape. is the position of the object center relative to the eye. Assign color to the object. Determine the texture information for the object. is the allowable deviation value of the object shape, is the allowable deviation value of the object center relative to the eye position, is the object color tolerance value. The allowed deviation value for the object's texture information.

[0019] Furthermore, the obtaining of the allowable deviation data of the characteristics of the required wearable object includes the following steps: obtaining a deviation matching data set, the deviation matching data set including a plurality of deviation matching data, the deviation matching data including a brightness matching value and a clarity matching value; comparing the brightness of the environment in which the access control is located and the current clarity of the access control with the brightness matching value and the clarity matching value to obtain a matching coefficient; determining that the deviation matching data corresponding to the minimum matching coefficient corresponds to the allowable deviation data of the characteristics of the required wearable object stored in the database.

[0020] The present invention has the following beneficial effects:

[0021] This high-precision face recognition system used for face recognition access control ensures that only the right people access specific areas at the right time through double verification (time period + face recognition), greatly improving security. The self-check module ensures the stable operation of the system. Once a problem is found, it can respond quickly to avoid security loopholes. The statistical module provides detailed access records for post-audit and behavior analysis, which helps to improve overall supervision capabilities.

[0022] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flowchart of the high-precision face recognition system used for face recognition access control in the present invention. DETAILED DESCRIPTION

[0024] The embodiment of the present application uses a high-precision face recognition system for face recognition access control, which ensures that only the right people access specific areas at the right time through double verification (time period + face recognition), greatly improving security.

[0025] See also Figure 1 , an embodiment of the present invention provides a technical solution: a high-precision face recognition system for face recognition access control, including a recognition access control self-check module, a matching information acquisition module, an identity recognition module, a facial recognition module and a statistical module.

[0026] The system first performs a self-check to ensure that all hardware and software components are working properly. If a fault is detected, the system notifies the control center for manual intervention to avoid erroneous authorization or denial of access when the equipment fails. After the system passes the self-check, the pre-stored visitor information is extracted from the database (visitor information includes established visitor information and temporary visitor information, where established visitor information is the employee information of the registered unit or factory, and temporary visitor information is the information of outsiders registered by appointment), including facial features, identity information and corresponding access time period.

[0027] The system determines whether the visitor is within the permitted access time based on the current time and the identity information provided by the visitor. If so, facial recognition will be performed; otherwise, access will be directly denied or the control center will be prompted for review. After confirming that the visitor is within the permitted access time, the system uses facial recognition technology to compare the visitor's actual facial features with the feature data set in the database to verify the authenticity of the identity. The system records the relevant information of each visitor who passes through the access control, including identity verification information, access time and facial recognition data, for subsequent security audits and data analysis.

[0028] By combining face recognition technology and identity verification, the system can automatically detect and verify the identity of visitors, ensuring that only authorized individuals can access restricted areas within a specified time. In addition, the system also improves the reliability and stability of the equipment through the self-check module, which can detect and handle faults in a timely manner. At the same time, the system records detailed access data through the statistics module, which is helpful for subsequent security supervision and data analysis. It can effectively solve the automation and security issues in security access control management.

[0029] The identification access control self-check module is used to perform self-check on the face recognition access control to determine whether the face recognition access control can perform face recognition normally in the current time period. If normal face recognition cannot be performed, a prompt is sent to the control center for manual recognition. If normal face recognition can be performed, self-check continues.

[0030] Regular self-checks and environmental adaptability tests help to promptly identify potential system problems and maintenance needs, ensuring long-term and stable operation of the system. Determine whether the face recognition access control can perform face recognition normally in the current time period, including the following steps: obtain the current temperature of the recognition access control and the brightness of the environment where the recognition access control is located; determine whether the brightness of the environment where the recognition access control is located falls within the allowable brightness range stored in the database: if the brightness of the environment where the recognition access control is located does not fall within the allowable brightness range stored in the database, normal face recognition cannot be performed; if the brightness of the environment where the recognition access control is located falls within the allowable brightness range stored in the database, then based on the brightness of the environment where the recognition access control is located, obtain the recognition access control clarity parameter value corresponding to the brightness of the environment where the recognition access control is located from the database; obtain the access control temperature parameter value stored in the database; obtain the current clarity of the recognition access control; based on the current temperature of the recognition access control, the access control temperature parameter value, the current clarity of the recognition access control and the recognition access control clarity parameter value, obtain a compliance value; determine whether the compliance value is greater than the compliance threshold stored in the database: if the compliance value is greater than the compliance threshold stored in the database, face recognition can be performed normally; if the compliance value is not greater than the compliance threshold stored in the database, face recognition cannot be performed normally.

[0031] The system first obtains the current ambient temperature and brightness of the access control system, which can be obtained through relevant sensors. The system checks whether the current brightness falls within the allowable range defined in the database. If it is not within the allowable range, the system will determine that the current environment is not suitable for face recognition; if it is within the range, it will proceed to the next step. When the brightness is appropriate, the system obtains the clarity parameter value corresponding to the brightness from the database, and dynamically adjusts the recognition parameters according to the real-time environmental data, which enhances the system's flexibility and ability to cope with various environmental conditions.

[0032] The system obtains the access control temperature parameter value and the current clarity stored in the database. The compliance value is calculated based on the current temperature, temperature parameter value, current clarity (which can be obtained by, but not limited to, image clarity evaluation algorithm, image quality index, where the image clarity evaluation algorithm can be obtained by edge detection or contrast measurement, and the image quality index can be obtained by Laplacian method or blur detection algorithm) and clarity parameter value. The calculated compliance value is compared with the compliance threshold in the database. If the compliance value is greater than the threshold, the system considers that the environmental conditions are suitable for face recognition; if not, the system determines that the environmental conditions are not suitable.

[0033] Through detailed environmental condition detection (such as brightness and temperature), the system can perform face recognition under optimal conditions, thereby improving recognition accuracy and reliability. When the environmental conditions are not good (such as too dark or too hot), the system will automatically stop face recognition to prevent misidentification caused by environmental factors and avoid security risks.

[0034] The method for obtaining the matching value is as follows:

[0035] ;

[0036] In the formula, To match the value, To identify the current temperature of the access control, is the access control temperature parameter value, To identify the current clarity of the access control, To identify the access control clarity parameter value.

[0037] In this formula, By comparison and as well as and The formula uses the method of summing the reciprocals of the absolute differences and then taking the natural logarithm. This calculation method takes into account the degree of deviation of the two factors, temperature and clarity, and combines the two deviations by weighting. Adding 1 to avoid division by zero can quantify the deviation between the two parameters and their parameter values.

[0038] The sum of the two inverses means that the impact of larger differences (more deviations) is reduced, and the impact of smaller differences is increased, which helps to balance the weights of the two deviations. Taking the natural logarithm is to further modulate the impact of the two deviation values. The slow growth of the logarithmic function can prevent extreme values ​​from having an excessive impact on the results.

[0039] The performance of electronic devices, especially those containing sensing components and cameras, may be affected by temperature changes. Extreme temperature conditions may cause hardware failure or performance degradation, such as impaired camera focusing ability and delayed sensor response time. Clarity is directly related to the core requirement of face recognition technology, namely image quality. If the image captured by the camera is not clear enough, it will directly affect the accuracy and reliability of the recognition algorithm. Although temperature and clarity are not directly related in physical properties, they jointly affect the overall effect of the access control system at the operating environment and device performance level. For example, in high or low temperature environments, the imaging elements of the camera may be affected, resulting in a decrease in image quality, thereby affecting clarity. In addition, the ambient temperature also has an important impact on the efficiency of the cooling system (such as fans) of electronic equipment, which indirectly affects the stable operation and performance of the equipment.

[0040] The matching information acquisition module is used to obtain visitor information from the database when the face recognition access control can perform normal face recognition. The visitor information includes a visitor facial feature data set, an identity information set, and an allowed access time period corresponding to each identity information in the identity information set. The visitor information includes established visitor information and temporary visitor information.

[0041] The identity recognition module is used to obtain the visitor's identity information to be verified and the actual visit time, identify the visitor based on the visitor's identity information to be verified and the actual visit time and the identity information and the allowed access time period, and determine whether the current visitor can perform facial recognition. If facial recognition is not possible, a prompt is sent to the control center. If facial recognition is possible, the facial recognition module and the statistical module are started.

[0042] The design of the identity recognition module follows a strict identity authentication process to ensure that only authorized individuals can enter restricted areas. Determining whether the current visitor can perform facial recognition includes the following steps: determining whether the identity information to be verified belongs to the identity information set, which includes the established visitor identity information set and the temporary visitor identity information set; if the identity information to be verified does not belong to the identity information set, facial recognition cannot be performed; first check whether the identity information to be verified provided by the visitor is in the identity information set pre-stored in the system (including established visitors and temporary visitors). If it is not in any set, the system will refuse to perform facial recognition because the visitor's authorization status cannot be confirmed.

[0043] If the identity information to be verified belongs to the identity information set, then determine whether the actual access time falls within the allowed access time period corresponding to the identity information to be verified: If the actual access time falls within the allowed access time period corresponding to the identity information to be verified, facial recognition can be performed. This step ensures that the visitor accesses within the authorized time.

[0044] If the actual access time does not fall within the allowed access time period corresponding to the identity information to be verified, it is determined whether the identity information to be verified belongs to the established visitor identity information set: if the identity information to be verified belongs to the established visitor identity information set, facial recognition can be performed; if the identity information to be verified does not belong to the established visitor identity information set, it is determined whether the identity information to be verified belongs to the temporary visitor identity information set: if the identity information to be verified belongs to the temporary visitor identity information set, a prompt is sent to the control center for manual identification. If the result of manual identification is that facial recognition is allowed, facial recognition can be performed. For temporary visitors' access attempts at unauthorized times, manual identification is used to determine whether entry is allowed. This can handle accidents or emergencies while maintaining security control. If the result of manual identification is that facial recognition is not allowed, facial recognition cannot be performed; if the identity information to be verified does not belong to the temporary visitor identity information set, facial recognition cannot be performed.

[0045] Through multi-layer verification (identity and time verification), the system ensures that only the right people have access at the right time, greatly increasing security. By distinguishing between established visitors and temporary visitors, the system is able to apply different security measures according to different types of visitors, providing flexibility while maintaining necessary control. The double check of identity and time reduces the risk of unauthorized access and avoids potential security loopholes. In the case of non-routine access requests, by introducing a manual identification step, the system can effectively handle emergency or special situations, ensuring a fast and safe response to these situations.

[0046] The reason for this design is to provide a comprehensive and hierarchical security verification framework that can flexibly respond to various normal and abnormal situations to ensure the efficiency and security of the access control system. While protecting sensitive areas, it can also handle various unexpected access requests, such as temporary visitors or emergency situations, making the system both strict and humane.

[0047] The facial recognition module is used to obtain the actual facial feature data of the visitor who can perform facial recognition, perform facial recognition based on the actual facial feature data and the visitor facial feature data set, and determine whether the facial recognition is successful. If the facial recognition is successful, the current visitor is allowed to pass through the access control. If the facial recognition is unsuccessful, a prompt is sent to the control center for manual recognition.

[0048] Determining whether facial recognition is successful includes the following steps: obtaining actual facial feature data of visitors who can be facially recognized, wherein the actual facial feature data includes actual geometric features (such as nose, eyes, facial proportions, position and size of mouth, etc.), actual texture features (such as skin features such as wrinkles and moles) and actual biometric features (such as iris and pupil, ear shape, etc.); wherein for actual geometric features, images can be captured using a high-quality camera, and then the facial area is located using a facial detection algorithm. Next, facial key points are identified and marked using facial tagging technology (such as a deep learning model), thereby calculating facial geometric features, such as the distance and angle between key points. The acquisition of actual texture features is accomplished by analyzing the detailed texture of the facial image, including the texture pattern of the skin, moles, wrinkles, and other epidermal features. These features are usually identified through images captured by a high-resolution camera, and then image processing techniques such as gray level co-occurrence matrix (GLCM) or local binary pattern (LBP) are applied for detailed analysis. The acquisition of actual biometric features involves identifying and analyzing individual unique physiological or behavioral features, such as skin texture, vascular pattern, pupil response, etc. Such features are typically captured through specialized biometric techniques, such as using infrared imaging to detect blood vessel patterns or advanced camera technology to observe pupil changes.

[0049] Acquire a visitor facial feature data set, the visitor facial feature data set includes a plurality of visitor facial feature data, each of the visitor facial feature data includes a parameterized geometric feature, a parameterized texture feature and a parameterized biometric feature; acquire feature allowable deviation data during facial recognition, the feature allowable deviation data includes a geometric feature allowable deviation value, a texture feature allowable deviation value and a biometric feature allowable deviation value; perform similarity processing on the actual facial feature data and each visitor facial feature data in the visitor facial feature data set, and obtain a facial recognition matching value in combination with the feature allowable deviation data.

[0050] The system compares the visitor's actual features with the feature data set stored in the database, which contains the baseline facial data of each visitor, covering geometry, texture and biometric features. In order to cope with natural changes (such as lighting changes, picture clarity), specific allowable deviation values ​​are set, which define the acceptable range of feature changes. The actual features are compared with the pre-stored features, and the similarity is calculated in combination with the allowed deviations to obtain the facial recognition match value for each visitor.

[0051] Obtaining a minimum facial recognition match value; determining whether the minimum facial recognition match value is less than a facial recognition match threshold stored in a database: if the minimum facial recognition match value is less than the facial recognition match threshold stored in the database, facial recognition is successful; if the minimum facial recognition match value is not less than the facial recognition match threshold stored in the database, facial recognition is unsuccessful.

[0052] By integrating multiple facial features (geometry, texture, biology), the system can more accurately identify individuals and reduce the rate of false positives. By setting feature tolerances, the system can adapt to natural changes in the face over time, improve the user experience, and reduce rejection rates due to slight changes in appearance. Accurate facial matching ensures that only individuals with exact matches can pass verification, greatly improving the security of the system. Through similarity processing and strict threshold comparison, the system ensures stable operation and high reliability in various environments.

[0053] The reason for this design is that a recognition system is needed that can both accurately match and adapt to daily changes. As a biometric technology, facial recognition must be able to handle external influences such as lighting, while ensuring that there are no false or missed recognitions due to external factors. Setting the allowable deviation and matching threshold is the key to ensuring that the system is neither overly sensitive nor overly lenient, so that the system can provide a smooth and user-friendly access experience while ensuring security.

[0054] The method for obtaining the facial recognition matching value is:

[0055] ;

[0056] In the formula, is the facial recognition matching value, is the actual geometric feature, is the actual texture feature, For actual biological characteristics, To determine the geometric features, To determine the texture features, To determine the biometric characteristics, is the allowable deviation value of the geometric feature, is the allowable deviation value of texture features, is the allowable deviation value of the biometric feature, for and The similarity function of for and The similarity function of for and The similarity function above can use Euclidean distance, Manhattan distance or Bhattacharyya distance.

[0057] The reason why actual geometric features, actual texture features, and actual biometric features are used as the basis for facial recognition system analysis is that each of these three provides unique and complementary information, which together constitute a complete and multi-dimensional description of a person's face. This comprehensive analysis method significantly improves the accuracy and robustness of facial recognition systems. Geometric features provide basic structural information for facial recognition and are a key factor for preliminary positioning and rough matching during the recognition process. Texture features increase the ability to recognize an individual's unique appearance, especially when geometric features are interfered with by similar facial structures. Biometric features provide an identification basis that is difficult to forge or change, greatly enhancing the security and anti-fraud capabilities of facial recognition systems.

[0058] These three types of features are interrelated and together form a complex recognition network, which enables facial recognition technology to remain efficient and accurate under various conditions. For example, if an environmental factor (such as light changes) affects the visibility of texture features, geometric and biometric features can compensate for the information loss in the recognition process. In addition, the combination of different features can be optimized in facial recognition algorithms through data fusion technology, and each feature fights against a certain type of variation or fraud attempt, increasing the overall robustness of the system. By evaluating the similarity between these features and the standard features stored in the database, combined with the allowed deviation value, a comprehensive facial recognition match value can be accurately calculated. Using Euclidean distance, Manhattan distance or Bhattacharyya distance as the similarity function, the similarity between features can be evaluated from different mathematical perspectives, which helps to evaluate the match in a balanced manner in various situations and ensure the accurate execution of facial recognition. This multi-feature fusion method improves the effect of facial recognition technology in practical applications, making it more reliable and difficult to be deceived.

[0059] Acquiring feature tolerance data during facial recognition includes the following steps: obtaining identity information of a visitor who can perform facial recognition, which is the key to retrieving relevant data from a database. Based on the identity information to be verified, a feature tolerance data set corresponding to the identity information to be verified is obtained from the database, and different feature deviation tolerances are provided according to different identity information.

[0060] The feature allowable deviation data set includes several feature allowable deviation data; the brightness of the environment in which the access control is located and the current clarity of the access control are obtained; the matching brightness of the environment in which the access control is located and the current matching clarity of the access control are obtained and stored in the database corresponding to each feature allowable deviation data; the brightness of the environment in which the access control is located and the current clarity of the access control are compared with the matching brightness of the environment in which the access control is located and the current matching clarity of the access control to obtain a comparison coefficient.

[0061] Determine the matching brightness of the environment where the access control is located and the current matching clarity of the access control corresponding to the minimum comparison coefficient; and obtain feature allowable deviation data based on the determined matching brightness of the environment where the access control is located and the current matching clarity of the access control.

[0062] Fine-tune the deviation value to match the current environment, reduce misidentification caused by environmental changes, and improve the overall accuracy of the system. Ensure the stability and reliability of the facial recognition system under various operating conditions by adjusting and optimizing the deviation setting in real time. Allow personalized deviation settings based on different individual identity information and environmental requirements, increasing the flexibility of the system and user satisfaction.

[0063] This design allows the facial recognition system to better adapt to different usage environments and user needs. By dynamically adjusting to environmental factors such as brightness and clarity, the system is able to provide optimized recognition performance under various conditions. This approach helps to handle changes in natural light and other environmental factors, ensuring the accuracy and efficiency of the recognition process, which is extremely important in public safety and high-security locations.

[0064] The method for obtaining the comparison coefficient is:

[0065] ;

[0066] In the formula, is the comparison coefficient, To identify the brightness of the environment where the access control is located, To identify the current clarity of the access control, To identify the brightness of the environment where the access control is located, To identify the current matching clarity of the access control, is a natural constant.

[0067] Brightness directly affects the lighting conditions of the image. An environment that is too dark or too bright will affect the quality of the facial image captured by the camera. Under inappropriate lighting, the image may be overexposed or underexposed, with serious loss of details, which will affect the extraction of features. Clarity refers to the level of detail and edge sharpness of the image, which directly affects the accuracy of positioning and recognition of facial feature points. Insufficient clarity may be caused by a variety of reasons, including the focal length setting of the camera, lens cleanliness, and insufficient ambient brightness. Ambient brightness and clarity are physically interdependent. Appropriate brightness helps improve image clarity, making facial features in the image more clearly identifiable. Conversely, inappropriate brightness not only affects the intuitive texture of the image, but may also cause the autofocus technology to fail, further affecting the clarity of the image.

[0068] By analyzing the current ambient brightness and clarity, the system can dynamically adjust the feature deviation allowed in the facial recognition process. For example, in low-light conditions, it may be necessary to increase the feature deviation tolerance to adapt to the difficulty of feature recognition caused by insufficient lighting. This dynamic adjustment enables the facial recognition system to maintain a high recognition accuracy rate under various environmental conditions. The use of ambient brightness and current clarity as the basis for obtaining feature allowable deviation data is to make the facial recognition system environmentally adaptable and flexible. This design approach ensures that the system can achieve optimal operating results under various environmental conditions, especially in public safety or commercial safety applications, which can significantly improve the practicality and reliability of the system. This approach not only improves the overall performance of the system, but also optimizes the user experience and ensures the efficiency and accuracy of the system in practical applications.

[0069] Before obtaining the actual facial feature data of the visitor who can perform facial recognition, it also includes identifying the objects worn by the visitor who can perform facial recognition as determined by the identity recognition module, and judging whether the visitor who can perform facial recognition as determined by the identity recognition module is wearing the required objects. If the required objects are not worn, it is re-determined that facial recognition cannot be performed. If the required objects are worn, the actual facial feature data of the visitor who can perform facial recognition is obtained.

[0070] In some occasions, such as factories or specific laboratories, for safety and hygiene reasons, it is required to wear specific safety equipment, such as helmets, goggles, masks, etc. This step ensures that every individual entering these areas complies with the necessary safety regulations, thereby reducing work-related injuries and health risks. In areas with strict access control, such as clean rooms or safety-critical production areas, automatically detecting whether employees are wearing the necessary equipment as required can improve the automation and operational efficiency of the entire safety management system. This avoids the need for manual inspections, reducing human errors and management costs.

[0071] Determining whether a visitor who can perform facial recognition determined by an identity recognition module is wearing a required object includes the following steps: obtaining feature data of the object required to be worn by the visitor who can perform facial recognition, wherein the feature data of the object required to be worn includes object shape, position of the center of the object relative to the eyes, object color and object texture information, and these data are obtained through images or videos captured in real time. In addition, if these data cannot be obtained through images or videos captured in real time, then it can be directly determined that the visitor is not wearing an object.

[0072] Acquire characteristic parameter data of the object to be worn stored in a database, wherein the characteristic parameter data of the object to be worn includes the parameter shape of the object, the parameter position of the center of the object relative to the eyes, the parameter color of the object, and the parameter texture information of the object; acquire characteristic allowable deviation data of the object to be worn, wherein the characteristic allowable deviation data of the object to be worn includes the allowable deviation value of the object shape, the allowable deviation value of the position of the center of the object relative to the eyes, the allowable deviation value of the object color, and the allowable deviation value of the object texture information. These deviation data help the system flexibly identify objects that meet the standards but have slight differences.

[0073] Based on the characteristic data of the object required to be worn, the characteristic parameter data of the object required to be worn and the characteristic allowable deviation data of the object required to be worn, the object matching coefficient is obtained; if the object matching coefficient is less than the object matching threshold coefficient stored in the database, the required object is worn; if the object matching coefficient is not less than the object matching threshold coefficient stored in the database, the required object is not worn.

[0074] Ensure that every individual entering a specific area strictly complies with safety regulations, such as wearing a hard hat in an industrial environment, to reduce safety accidents. Automatically detect and verify whether the necessary safety equipment is worn to help companies comply with relevant occupational health and safety standards and regulations. By setting feature tolerances, the system can adapt to the natural wear or slight changes of objects in actual use, avoiding misjudgments due to subtle differences.

[0075] The method for obtaining the object matching coefficient is as follows:

[0076] ;

[0077] In the formula, is the object matching coefficient, is the shape of the object, is the position of the object center relative to the eye, The object color, expressed in RGB values. is the object texture information, Give the object a shape. is the position of the object center relative to the eye. Assign color to the object. Determine the texture information for the object. is the allowable deviation value of the object shape, is the allowable deviation value of the object center relative to the eye position, is the object color tolerance value. The above similarity function can use Euclidean distance, Manhattan distance, etc.

[0078] The shape of an object is a fundamental feature for identifying its type and purpose. For example, there is a significant difference in shape between a hard hat and a regular hat, and being able to distinguish based on shape is critical for the system to correctly identify the type of object. The position of an object relative to the eye can help the system confirm that the object is worn correctly. For example, safety glasses need to cover the eyes, and if the position is too far away, they may not provide the expected protection. Color not only helps to identify the type of object, but may also be part of the compliance requirements, especially in work environments that require high-visibility clothing. Texture information can help to further distinguish subtle differences, especially when the color and shape of the objects are similar, texture can provide an additional dimension of distinction.

[0079] By combining shape, position, color and texture for evaluation, the system is able to fully understand the characteristics of the object, thereby improving the accuracy of recognition. This comprehensive approach reduces errors caused by misjudgment of a single feature. Each feature has a set deviation value, which takes into account the slight changes and wear that may occur in actual use. This design improves the practicality and flexibility of the system and can adapt to changes in the real working environment. By calculating the similarity between each feature and its parameter value and taking into account the allowable deviation, the system is able to generate a comprehensive matching coefficient for the final judgment of whether the object meets the requirements.

[0080] The method of obtaining the allowable deviation data of the characteristics of the object to be worn includes the following steps: obtaining a deviation matching data set, the deviation matching data set includes a plurality of deviation matching data, the deviation matching data includes a brightness matching value and a clarity matching value, which are preset according to the optimal object recognition performance under different environmental conditions. The brightness of the environment where the access control is located and the current clarity of the access control are compared with the brightness matching value and the clarity matching value to obtain a matching coefficient. This can evaluate the consistency between the current environment and the ideal or preset environment.

[0081] Determine that the deviation matching data corresponding to the minimum matching coefficient corresponds to the allowable deviation data of the required worn object features stored in the database.

[0082] The matching coefficient is obtained as follows:

[0083] ;

[0084] In the formula, is the matching coefficient, To identify the brightness of the environment where the access control is located, To identify the current clarity of the access control, is the brightness matching value, is the clarity matching value, is a natural constant. The formula reflects the difference between the actual environmental parameters and the preset parameters. The exponential form in the formula means that any small deviation will lead to a large coefficient change, emphasizing the importance of precise matching.

[0085] Brightness directly affects the visibility of objects and the quality of images captured by the camera. If the environment is too bright or too dark, it may affect the true representation of the color and texture of the object, making it difficult to identify. Appropriate brightness is key to ensuring that the features of an object are correctly captured and recognized. Under different brightness conditions, the visual features of an object, such as color and texture, may appear different. By analyzing and adapting to the ambient brightness, the recognition algorithm can adjust its handling of these changes to maintain recognition accuracy. Details such as the shape and texture of an object are difficult to accurately recognize when the clarity is insufficient. For example, the details of a mandatory safety sign or work badge may not be recognizable in a blurred image. Recognition access control systems often include autofocus functions to ensure the clarity of the image. Recognition systems need to be able to operate in a variety of environmental conditions, including different lighting and viewing conditions. Using ambient brightness and clarity as the basis for analyzing deviation data can make the system more resilient to these changes in conditions and improve its adaptability and flexibility.

[0086] The use of the ambient brightness and current clarity of the environment in which the access control system is located as the basis for the analysis of the allowable deviation data required for the characteristics of the worn object is to ensure that the facial recognition system can operate effectively under various environmental conditions. Considering the ambient brightness and clarity ensures that the system can maintain efficient recognition capabilities under different lighting and visual conditions, thereby adapting to a wider range of application scenarios.

[0087] The statistical module is used to record the number of visitors passing through the access control system, as well as the identity information to be verified when each visitor is allowed to enter, the actual visit time, and the facial feature data during facial recognition.

[0088] The statistics module records the total number of visitors who pass through the access control system every day or every time period. This data can help managers monitor the peak visit volume and visit frequency, so as to make appropriate resource allocation and arrangements. The identity information to be verified for each visitor is recorded to ensure detailed visitor records. This not only helps with security audits, but also provides a basis for subsequent identity verification. The actual visit time of each visitor is recorded, which helps to analyze visit patterns and ensure that visitors enter within the allowed time.

[0089] An electronic device includes: a processor and a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the high-precision face recognition system for face recognition access control as described above.

[0090] A computer-readable storage medium is used to store a program, and when the program is executed by a processor, the high-precision face recognition system for face recognition access control as described above is implemented.

[0091] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] The present invention is described with reference to flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0095] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0096] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A face recognition system for face recognition access control, characterized in that: It includes access control self-check module, matching information acquisition module, identity recognition module, face recognition module and statistics module, among which: The identification access control self-check module is used to perform self-check on the face recognition access control to determine whether the face recognition access control can perform face recognition normally in the current time period. If normal face recognition cannot be performed, a prompt is sent to the control center for manual recognition. If normal face recognition can be performed, self-checking continues; The matching information acquisition module is used to obtain visitor information from a database when the face recognition access control can perform normal face recognition, and the visitor information includes a visitor facial feature data set, an identity information set, and an allowed access time period corresponding to each identity information in the identity information set; The identity recognition module is used to obtain the visitor's identity information to be verified and the actual visit time, identify the visitor based on the visitor's identity information to be verified and the actual visit time and the identity information and the allowed visit time period, and determine whether the current visitor can perform facial recognition. If facial recognition is not possible, a prompt is sent to the control center. If facial recognition is possible, the facial recognition module and the statistical module are started; The facial recognition module is used to obtain the actual facial feature data of the visitor who can perform facial recognition, perform facial recognition based on the actual facial feature data and the visitor facial feature data set, and determine whether the facial recognition is successful. If the facial recognition is successful, the current visitor is allowed to pass through the access control. If the facial recognition is unsuccessful, a prompt is sent to the control center for manual recognition; The step of determining whether facial recognition is successful comprises the following steps: Acquire actual facial feature data of a visitor capable of facial recognition, wherein the actual facial feature data includes actual geometric features, actual texture features, and actual biometric features; Acquire a visitor facial feature data set, wherein the visitor facial feature data set includes a plurality of visitor facial feature data, each of which includes a parameterized geometric feature, a parameterized texture feature, and a parameterized biological feature; Acquire feature allowable deviation data during face recognition, wherein the feature allowable deviation data includes a geometric feature allowable deviation value, a texture feature allowable deviation value, and a biometric feature allowable deviation value; Perform similarity processing on the actual facial feature data and the facial feature data of each visitor in the visitor facial feature data set, and obtain the facial recognition matching value by combining the feature allowable deviation data; Get the minimum facial recognition match value; Determine whether the minimum facial recognition match value is less than the facial recognition match threshold stored in the database: If the minimum facial recognition match value is less than the facial recognition match threshold stored in the database, the facial recognition is successful; If the minimum facial recognition match value is not less than the facial recognition match threshold stored in the database, facial recognition is unsuccessful; The method of obtaining feature allowable deviation data during facial recognition comprises the following steps: Obtaining identity information of visitors who can be verified through facial recognition; Based on the identity information to be verified, acquiring a feature allowable deviation data set corresponding to the identity information to be verified from a database, wherein the feature allowable deviation data set includes a plurality of feature allowable deviation data; Get the brightness of the environment where the access control is located and the current clarity of the access control; Obtain the allowable deviation data of each feature corresponding to the matching brightness of the environment where the access control is located and the current matching clarity of the access control stored in the database; Compare the brightness of the environment where the access control is located and the current clarity of the access control with the matching brightness of the environment where the access control is located and the current clarity of the access control to obtain a comparison coefficient; Determine the matching brightness of the environment where the access control is located and the current matching clarity of the access control corresponding to the minimum comparison coefficient; Based on the determined matching brightness of the environment where the access control is located and the current matching clarity of the access control, feature allowable deviation data is obtained; The statistical module is used to record the number of visitors passing through the access control system, as well as the identity information to be verified when each visitor is allowed to enter, the actual visit time, and the facial feature data during facial recognition.

2. A face recognition system for face recognition access control according to claim 1, characterized in that: The step of judging whether the face recognition access control system can perform face recognition normally in the current time period includes the following steps: Get the current temperature of the access control system and the brightness of the environment where the access control system is located; Determine whether the brightness of the environment where the access control is located falls within the allowable brightness range stored in the database: If the brightness of the environment where the access control is located does not fall within the allowed brightness range stored in the database, normal face recognition cannot be performed; If the brightness of the environment where the access control is located falls within the brightness allowable range stored in the database, then based on the brightness of the environment where the access control is located, the access control clarity parameter value corresponding to the brightness of the environment where the access control is located is obtained from the database; Get the access control temperature parameter value stored in the database; Get the current clarity of access control recognition; Based on identifying the current temperature of the access control, the access control temperature parameter value, identifying the current clarity of the access control, and identifying the access control clarity parameter value, obtaining a compliance value; Determine whether the matching value is greater than the matching threshold stored in the database: If the match value is greater than the match threshold stored in the database, face recognition can be performed normally; If the match value is not greater than the match threshold stored in the database, face recognition cannot be performed normally.

3. A face recognition system for face recognition access control according to claim 2, characterized in that: The method for obtaining the matching value is as follows: ； In the formula, To match the value, To identify the current temperature of the access control, is the access control temperature parameter value, To identify the current clarity of the access control, To identify the access control clarity parameter value.

4. The face recognition system for face recognition access control according to claim 1, characterized in that: The step of determining whether the current visitor can be subjected to facial recognition comprises the following steps: Determine whether the identity information to be verified belongs to an identity information set, wherein the identity information set includes an established visitor identity information set and a temporary visitor identity information set; If the identity information to be verified does not belong to the identity information set, facial recognition cannot be performed; If the identity information to be verified belongs to the identity information set, determine whether the actual access time falls within the allowed access time period corresponding to the identity information to be verified: If the actual access time falls within the allowed access time period corresponding to the identity information to be verified, facial recognition can be performed; If the actual access time does not fall within the allowed access time period corresponding to the identity information to be verified, then determine whether the identity information to be verified belongs to the established visitor identity information set: If the identity information to be verified belongs to the established visitor identity information set, facial recognition can be performed; If the identity information to be verified does not belong to the established visitor identity information set, determine whether the identity information to be verified belongs to the temporary visitor identity information set: If the identity information to be verified belongs to the temporary visitor identity information set, a prompt is sent to the control center for manual identification. If the result of manual identification is that facial recognition is allowed, facial recognition can be performed. If the result of manual identification is that facial recognition is not allowed, facial recognition cannot be performed. If the identity information to be verified does not belong to the temporary visitor identity information set, facial recognition cannot be performed.

5. The face recognition system for face recognition access control according to claim 1, characterized in that: Before obtaining the actual facial feature data of the visitor who can perform facial recognition, it also includes identifying the objects worn by the visitor who can perform facial recognition as determined by the identity recognition module, and judging whether the visitor who can perform facial recognition as determined by the identity recognition module is wearing the required objects. If the required objects are not worn, it is re-determined that facial recognition cannot be performed. If the required objects are worn, the actual facial feature data of the visitor who can perform facial recognition is obtained.

6. A face recognition system for face recognition access control according to claim 5, characterized in that: The step of determining whether the visitor who can be identified by the identity recognition module for facial recognition is wearing the required object comprises the following steps: Acquire feature data of an object required to be worn by a visitor who can perform facial recognition, wherein the feature data of the object required to be worn includes object shape, position of the center of the object relative to the eyes, object color, and object texture information; Acquire characteristic parameter data of the object to be worn stored in a database, wherein the characteristic parameter data of the object to be worn includes object parameter shape, object center relative eye parameter position, object parameter color and object parameter texture information; Acquire the allowed deviation data of the characteristics of the object to be worn, wherein the allowed deviation data of the characteristics of the object to be worn includes the allowed deviation value of the shape of the object, the allowed deviation value of the position of the center of the object relative to the eyes, the allowed deviation value of the color of the object, and the allowed deviation value of the texture information of the object; Obtaining an object matching coefficient based on the characteristic data of the object to be worn, the characteristic parameter data of the object to be worn, and the characteristic allowable deviation data of the object to be worn; If the object matching coefficient is less than the object matching threshold coefficient stored in the database, the required object is worn; If the object matching coefficient is not less than the object matching threshold coefficient stored in the database, the required object is not worn.

7. A face recognition system for face recognition access control according to claim 6, characterized in that: The method for obtaining the object matching coefficient is as follows: ； In the formula, is the object matching coefficient, is the shape of the object, is the position of the object center relative to the eye, is the object color, is the object texture information, Give the object a shape. is the position of the object center relative to the eye. Assign color to the object. Determine the texture information for the object. is the allowable deviation value of the object shape, is the allowable deviation value of the object center relative to the eye position, is the object color tolerance value. The allowed deviation value for the object's texture information.

8. The face recognition system for face recognition access control according to claim 6, characterized in that: The step of obtaining the allowable deviation data of the characteristics of the worn object includes the following steps: Acquire a deviation matching data set, wherein the deviation matching data set includes a plurality of deviation matching data, and the deviation matching data includes a brightness matching value and a clarity matching value; Compare the brightness of the environment where the access control is located and the current clarity of the access control with the brightness matching value and the clarity matching value to obtain a matching coefficient; Determine that the deviation matching data corresponding to the minimum matching coefficient corresponds to the allowable deviation data of the required worn object features stored in the database.

Images