Electronic class board attendance system for palm vein recognition

Abstract

The invention relates to the technical field of attendance management, and discloses an electronic class board attendance system for palm vein recognition, which comprises a palm vein data acquisition module for irradiating the palm of a user through a near-infrared light source and acquiring a palm vein image; the image optimization processing module is used for performing optical modeling optimization, noise removal and quality enhancement on the acquired palm vein image; the deep learning recognition module is used for extracting palm vein features and performing identity matching; the attendance record storage module is used for storing and managing user attendance data; and the control and communication module is used for managing cooperative work of all the modules and performing data interaction with an external attendance database. Through the technical scheme of combining near-infrared illumination and the light transmission equation, the palm vein feature image is accurately extracted, the effect of optimizing palm vein contrast and detail definition is achieved, the problems of palm vein feature edge blurring and detail loss are solved, and the image processing quality is improved.

Description

Technical Field

[0001] The invention relates to the technical field of attendance management, in particular to an electronic class card attendance system for palm vein recognition. Background Art

[0002] With the development of biometric technology, palm vein recognition has gradually become an important identity authentication method due to its high precision, strong security and difficulty in forging. Palm vein recognition technology mainly uses near-infrared light to illuminate the palm, collect reflected or transmitted infrared light signals, and form a palm vein image. In the existing technology, palm vein images are usually captured using optical imaging technology, and vein features are extracted through image processing algorithms for identity authentication. With the advancement of deep learning and artificial intelligence technology, some methods have improved the image feature extraction capability through technologies such as convolutional neural networks, which has improved the recognition accuracy and efficiency of the system.

[0003] However, existing palm vein recognition technology still faces some challenges in complex environments. First, traditional image acquisition methods are sensitive to factors such as ambient lighting and user palm posture, which may cause the captured palm vein images to lose details or be blurred. Even with the use of more advanced image processing technology, the quality of the image is still difficult to guarantee in low-light or changing environments, affecting the effect of subsequent recognition modules. Secondly, traditional deep learning models mainly focus on the extraction of local features and lack modeling of global image information. Therefore, they are still limited in the extraction and precise matching of complex palm vein features, resulting in the inability to further improve the recognition rate. Finally, existing methods have limited ability to repair and optimize low-quality images, especially when the image is disturbed or the light is insufficient, it is difficult to generate clear vein patterns, affecting the adaptability and recognition stability of the system. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an electronic class card attendance system for palm vein recognition, which solves the problems of poor palm vein image acquisition quality, detail loss and low recognition accuracy in complex environments.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an electronic class card attendance system for palm vein recognition, comprising: A palm vein data acquisition module, used to illuminate the user's palm through a near-infrared light source and obtain a palm vein image; Image optimization processing module, used for optical modeling optimization, noise removal and quality enhancement of the collected palm vein images; A deep learning recognition module for extracting palm vein features and performing identity matching; Attendance record storage module, used to store and manage user attendance data; Control and communication module, used to manage the collaborative work of each module and exchange data with the external attendance database; Among them, the image optimization processing module calculates the propagation path of light in skin tissue based on the light transmission equation and adjusts the image contrast to optimize the visibility of palm vein features.

[0006] Preferably, the palm vein data acquisition module uses a near-infrared optical camera to illuminate the user's palm with near-infrared light with a wavelength of 760nm to 850nm, and receives transmission or reflection signals through a photoelectric sensor to form a palm vein image.

[0007] Preferably, the image optimization processing module is based on the light transmission equation: ; in, Pixel The light intensity at is the incident light intensity, is the light absorption coefficient, is the propagation path length of light in skin tissue, is the base of natural logarithms; The image optimization processing module calculates the transmittance based on the equation and optimizes the contrast of the palm vein image by adjusting the illumination parameters and the image enhancement algorithm.

[0008] Preferably, the image optimization processing module performs denoising and edge preservation on the palm vein image based on a nonlinear variational method, and its optimization objective function is defined as: ; in, For the optimized image, is the original collected image, To control the parameters of edge preservation ability, is the regularization weight, is the image domain, Representing the image domain A small area on the surface is used to calculate the energy function , by solving the variational equation, the image quality is optimized and the edge structure of the palm vein is enhanced.

[0009] Preferably, the nonlinear variational method uses a gradient descent method for numerical solution, and performs geometric optimization on the palm vein image based on a Laplace-Beltrami operator.

[0010] Preferably, the deep learning recognition module adopts a neural network model including a convolutional neural network and a Transformer structure, the convolutional neural network is used to extract local features of palm veins, and the Transformer structure is used to model global patterns to complete palm vein identity matching.

[0011] Preferably, the Transformer structure of the deep learning recognition module calculates the palm vein feature map based on the self-attention mechanism, and the calculation method is: ; in, , , are the query matrix, key matrix and value matrix of the input features respectively, is the feature dimension, is the key matrix The transposition of is used to improve the accuracy of palm vein feature matching through the self-attention mechanism.

[0012] Preferably, the deep learning recognition module performs image optimization based on a generative adversarial network, the adversarial network generator is used to optimize the quality of the palm vein image, the discriminator is used to evaluate the consistency between the generated image and the real palm vein image, and the objective function of the generator is defined as: ; in, is the discriminator’s result, The image generated by the generator, is the real data distribution, is a random noise distribution, For the generator, is the discriminator, is a random variable representing a sample sampled from a certain distribution. For real data The expected value of It represents the expected value operation, which usually refers to solving the average value or weighted average value of a random variable. The generator is optimized through this objective function so that the generated image is closer to the real palm vein image.

[0013] Preferably, the attendance record storage module includes a database management system for storing user identity information, attendance time, attendance status, and providing a data synchronization interface to support remote access and management.

[0014] Preferably, the control and communication module uses an encrypted communication protocol to protect data transmission, and exchanges data with the attendance management platform via wireless communication.

[0015] The present invention provides an electronic attendance card system for palm vein recognition. It has the following beneficial effects: 1. The present invention uses a technical solution that combines near-infrared illumination and light transmission equations to accurately extract palm vein feature images, thereby achieving the effect of optimizing palm vein contrast and detail clarity. Compared with the prior art solution that only relies on simple optical imaging, the present invention solves the problems of blurred edges and detail loss of palm vein features, and improves the image processing quality.

[0016] 2. The present invention introduces Transformer and self-attention mechanism through deep learning model. The present invention realizes global modeling and fine matching of palm vein features, and achieves the effect of fast and accurate identity recognition. Compared with the traditional single convolutional network feature extraction method, the present invention solves the deficiency that complex palm print features cannot be fully captured, so that the matching accuracy is higher.

[0017] 3. The present invention uses an image optimization processing module to perform denoising and structural enhancement based on a nonlinear variational method. The optimization target is clear and the calculation is efficient. Compared with the commonly used filtering algorithm in the prior art, it can effectively retain the subtle structural features of the palm veins, overcome the defect that the traditional denoising algorithm is prone to lose edge details, and improve the practicality and stability of the palm vein image.

[0018] 4. The present invention optimizes the quality of palm vein images by generating adversarial networks. The present invention achieves the effect of enhancing low-quality images and improving the overall image clarity. Different from the traditional method of simply relying on post-algorithm repair, the present invention fundamentally solves the bottleneck of poor image quality under insufficient lighting or interference conditions through collaborative optimization of the generator and the discriminator, and improves the system's adaptability in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Please see attached Figure 1 The embodiment of the present invention provides an electronic class card attendance system for palm vein recognition, comprising: A palm vein data acquisition module, used to illuminate the user's palm through a near-infrared light source and obtain a palm vein image; Image optimization processing module, used for optical modeling optimization, noise removal and quality enhancement of the collected palm vein images; A deep learning recognition module for extracting palm vein features and performing identity matching; Attendance record storage module, used to store and manage user attendance data; Control and communication module, used to manage the collaborative work of each module and exchange data with the external attendance database; The image optimization processing module calculates the propagation path of light in skin tissue based on the light transmission equation and adjusts the image contrast to optimize the visibility of palm vein features. The palm vein data acquisition module uses a near-infrared optical camera to illuminate the user's palm with near-infrared light of 760nm to 850nm wavelength, and receives the transmission or reflection signal through the photoelectric sensor to form a palm vein image; The image optimization processing module is based on the light transmission equation: ; in, Pixel The light intensity at is the incident light intensity, is the light absorption coefficient, is the propagation path length of light in skin tissue, is the base of natural logarithms; The image optimization processing module calculates the transmittance based on the equation and optimizes the contrast of the palm vein image by adjusting the illumination parameters and the image enhancement algorithm; The image optimization processing module performs denoising and edge preservation on the palm vein image based on the nonlinear variational method, and its optimization objective function is defined as: ; in, For the optimized image, is the original collected image, To control the parameters of edge preservation ability, is the regularization weight, is the image domain, Representing the image domain A small area on the surface is used to calculate the energy function ,The image quality is optimized by solving the variational equation and the edge structure of the palm vein is enhanced; The nonlinear variational method uses the gradient descent method for numerical solution and performs geometric optimization of the palm vein image based on the Laplace-Beltrami operator; The deep learning recognition module uses a neural network model that includes a convolutional neural network and a Transformer structure. The convolutional neural network is used to extract local features of palm veins, and the Transformer structure is used to model global patterns to complete palm vein identity matching. The Transformer structure of the deep learning recognition module calculates the palm vein feature map based on the self-attention mechanism. The calculation method is: ; in, , , are the query matrix, key matrix and value matrix of the input features respectively, is the feature dimension, is the key matrix The transposition of , through the self-attention mechanism, improves the accuracy of palm vein feature matching; The deep learning recognition module performs image optimization based on the generative adversarial network. The adversarial network generator is used to optimize the quality of the palm vein image, and the discriminator is used to evaluate the consistency between the generated image and the real palm vein image. The objective function of the generator is defined as: ; in, is the discriminator’s result, The image generated by the generator, is the real data distribution, is a random noise distribution, For the generator, is the discriminator, is a random variable representing a sample sampled from a certain distribution. For real data The expected value of It represents the expected value operation, which usually refers to solving the average value or weighted average value of a random variable. The generator is optimized through this objective function so that the generated image is closer to the real palm vein image. The attendance record storage module includes a database management system for storing user identity information, attendance time, attendance status, and provides a data synchronization interface to support remote access and management; The control and communication module uses encrypted communication protocols to protect data transmission and interacts with the attendance management platform via wireless communication.

[0022] Specifically, in this embodiment, the palm vein data acquisition module illuminates the user's palm with a near-infrared light source, collects the reflected or transmitted near-infrared light signals, and thus generates clear palm vein images. These images play a core role in subsequent image optimization processing and deep learning recognition modules.

[0023] In some embodiments, this module ensures high-quality image acquisition through precise optical design, combined with quantum light transmission modeling and high-precision sensor configuration. The application of light transmission modeling technology effectively improves the accuracy and clarity of image acquisition, especially in high light, low light or complex environments, and can provide stable acquisition effects.

[0024] Generally, the palm vein data acquisition module in this embodiment uses a near-infrared light source with a wavelength range of 760nm to 850nm. Light in this wavelength band can effectively penetrate human skin and interact with blood in blood vessels, and is reflected or transmitted back to the sensor after being absorbed or scattered by the blood vessels. According to the absorption characteristics of the skin, the choice of light source and its light intensity will directly affect the clarity of the vein pattern.

[0025] In practical applications, the palm vein data acquisition module can automatically adjust the intensity of the light source according to changes in ambient light, ensuring that images of appropriate quality can be obtained under different lighting conditions. In low-light environments, the module can increase the intensity of the light source to improve image brightness; in strong light environments, the module automatically adjusts the light source to avoid overexposure and maintain image quality and details.

[0026] The present invention specifically introduces a light transmission modeling method to accurately simulate the propagation path of light in skin tissue, thereby affecting the optimization of image quality. Specifically, the propagation of light in the skin can be modeled using the light transmission equation in quantum optics. The light transmission equation is as follows: ; in: For the pixel The light intensity received at is the incident light intensity, which indicates the intensity of light emitted from the light source; is the light absorption coefficient, which indicates the absorption characteristics of light in the skin and is related to blood and tissue types; is the propagation path length of light in skin tissue, describing the light propagation path from the light source to the sensor.

[0027] According to this formula, the palm vein data acquisition module can accurately calculate the propagation characteristics of light in skin tissue, thereby optimizing the light intensity distribution during image acquisition. During image acquisition, this light transmission equation can help determine the true reflected or transmitted light intensity of each pixel, thereby improving the image contrast and ensuring that the details of the palm vein pattern are more clearly visible.

[0028] As an option, the resolution of the infrared image sensor used in the palm vein data acquisition module is usually not less than 1280x720 pixels to ensure high-quality image acquisition. The higher the resolution of the sensor, the richer the palm vein details that can be captured, especially in the display of small blood vessels, which is very important for subsequent image optimization and recognition modules.

[0029] Specifically, the sensitivity and response time of the sensor are also key factors affecting image quality. The sensitivity of the sensor determines its ability to work in low-light environments, while the response time determines its performance in dynamic environments. In order to ensure the efficiency and stability of the module, the sensor will be designed in combination with high-sensitivity infrared camera technology to ensure that clear palm vein images can be captured stably in any environment.

[0030] In order to ensure that the quality of palm vein images can meet the requirements of the subsequent deep learning recognition module, the connection between the palm vein data acquisition module and the image optimization module is crucial. The image optimization processing module uses a high-order nonlinear variational model to denoise, enhance and preserve the edges of the collected palm vein images.

[0031] In this process, the optimization objective function is: ; in: For the optimized image; is the original acquisition image; Parameters for controlling edge retention; It is the regularization weight to control the balance between denoising and image restoration; is the image domain, which represents the area of ​​all pixels in the image.

[0032] By solving the optimization equation, the noise of the palm vein image is removed and the image details are effectively preserved, especially the vascular structure and distribution of the palm vein. This enables subsequent image recognition to more accurately extract the features of the palm vein.

[0033] In this embodiment, the collected high-quality palm vein images will be initially processed in the data acquisition module and then transmitted to the subsequent image optimization processing module. The image data is transmitted via wired or wireless communication to ensure the rapid transmission and security of the data. The image optimization processing module performs denoising, enhancement, contrast adjustment and other processing on the image, and the optimized image is then transmitted to the deep learning recognition module to complete the final palm vein identity recognition.

[0034] In this embodiment, the main task of the image optimization processing module is to pre-process the original image transmitted by the palm vein data acquisition module to improve the quality of the image so that it can better adapt to the subsequent deep learning recognition module. By denoising, enhancing, edge thinning and texture optimization of the original image, the image optimization processing module ensures the improvement of image quality, thereby improving the accuracy of recognition.

[0035] Generally, after receiving the image from the data acquisition module, the image optimization processing module will first perform denoising. After denoising, the image may have insufficient contrast or residual noise. Next, the image enhancement algorithm is used to improve the visual effect of the image and ensure the clarity of the palm vein pattern. The optimized image will be passed as input to the deep learning recognition module.

[0036] In this embodiment, the image optimization processing module uses a denoising method based on a high-order nonlinear variational model. This method can not only effectively remove noise, but also ensure the clarity and detail retention of the vein pattern. The optimization objective function is as described above.

[0037] Specifically, The term plays a role in enhancing edges and details during the optimization process, avoiding the loss of important image structures during denoising. The term represents the difference between the denoised image and the original image, which is used to balance the contradiction between denoising and retaining the characteristics of the original image.

[0038] In some embodiments, the denoising process uses an advanced non-local mean filtering (NLM) method. The NLM method uses similar regions in the image to perform weighted average denoising, which can effectively handle non-stationary noise and performs well in preserving image details.

[0039] The image after denoising may have insufficient contrast. In order to improve the visual effect of the image, the image optimization processing module also combines histogram equalization and gamma correction technology to enhance the image contrast.

[0040] Histogram equalization redistributes the grayscale values ​​of the image so that the contrast of the image is evenly distributed across the entire brightness range. Its mathematical formula is: ; in: Represents grayscale value; is the probability distribution of pixel grayscale values.

[0041] This method can effectively improve the global contrast of the image, especially in low-contrast images.

[0042] Gamma correction is used to correct the brightness of an image so that it is more consistent with the characteristics of human eye perception. The formula is: ; in: is the pixel value of the input image; is the pixel value of the corrected image; It is the gamma coefficient, which is usually adjusted between 0.5 and 2 to achieve different degrees of brightness correction.

[0043] In some cases, when the brightness of an image is uneven, an adaptive gamma correction technique may be used to adjust the brightness of different areas according to the local characteristics of the image.

[0044] In this embodiment, the image optimization processing module also performs edge refinement processing. The purpose of edge refinement is to enhance the vascular structure in the palm vein image so that the vein texture is more clearly visible. In order to improve edge clarity, the image optimization processing module adopts an edge detection method based on gradient enhancement. Its core formula is as follows: ; in: Represents image points The gradient of and Respectively represent the images in and The local gradient of the direction.

[0045] This formula is used to calculate the gradient of each pixel. The areas with larger gradient values ​​usually correspond to the edges of the image. By enhancing these edges, the outlines of the veins can be clearly displayed. Specifically, the module enhances the high-gradient areas in the image to highlight the veins and textures.

[0046] In addition, the image optimization processing module also uses wavelet transform technology for texture enhancement. Wavelet transform enhances the high-frequency part of the image through multi-resolution analysis, making the details more prominent and enhancing the texture of the image. In wavelet transform, the image is decomposed into sub-images of different scales, and the high-frequency components are weighted and enhanced to finally synthesize the optimized image.

[0047] The optimized image will be stored in a standard image format (such as PNG or JPEG) and output to the subsequent deep learning recognition module. In some embodiments, the output image can be compressed and stored as needed, using lossless compression technology (such as PNG format) to save storage space and reduce the burden of data transmission.

[0048] Specifically, the image optimization processing module transmits the processed image to the subsequent module through a high-speed data interface (such as USB3.0, Gigabit Ethernet or other real-time transmission methods). The transmission of image data not only ensures the image quality, but also protects the user's privacy information through encryption technology.

[0049] The image optimization processing module can also achieve adaptive optimization, automatically adjusting the optimization strategy according to the quality of the collected image. For example, in low-light environments, the module will automatically increase the intensity of image enhancement; while in high-light conditions, it will focus on denoising and detail retention.

[0050] As an option, the image optimization processing module also supports multi-threading or parallel computing to increase processing speed when performing large-scale image processing, reduce delays in the image optimization process, and ensure the real-time performance of the system.

[0051] In this embodiment, the deep learning recognition module combines the convolutional neural network (CNN) and the Transformer structure to extract image features from local features and global information to achieve accurate identity matching. Through this combination, the deep learning module can effectively cope with the changing palm postures and background noise in complex environments, ensuring the efficiency and accuracy of identity recognition.

[0052] Generally, the deep learning recognition module processes high-quality images from the image optimization processing module, which are processed by denoising, enhancement and edge refinement to ensure the clarity of the vein pattern. The key to this module is to extract local features through convolutional neural networks (CNN) and use the Transformer network structure to further analyze global information, so as to comprehensively and accurately identify the characteristics of palm veins.

[0053] In this embodiment, a convolutional neural network (CNN) first extracts local features from an image. CNN is the most commonly used neural network architecture in the field of image processing. Through a series of convolutional layers, pooling layers, and fully connected layers, CNN can automatically extract spatially closely related features from the original image.

[0054] Convolution operation is the most basic operation in CNN. It performs convolution calculation on the image by sliding a convolution kernel (filter) to obtain a feature map. The mathematical representation of convolution operation is: ; in: is the output image after the convolution operation; For the input image at position The pixel value of It is the convolution kernel, which is used to extract local features; Represents the sliding position of the convolution kernel on the input image.

[0055] The result of the convolution operation Multiple feature maps will be generated, representing the features of the input image at different spatial scales. The function of the convolution layer is to extract the features of the local area, such as the vascular structure and texture of the palm vein. Specifically, the convolution kernel in this embodiment will be trained according to the characteristic structure of the vein so as to be able to extract the key details in the palm vein image.

[0056] As an option, CNN can also use multiple layers of convolutional layers to gradually extract more advanced features. The first convolutional layer is responsible for extracting basic features, such as edges and corners; while deeper convolutional layers can extract more complex features, such as texture and structural information. After each convolution layer, a pooling layer (usually maximum pooling or average pooling) is used to downsample the image to reduce the amount of data and retain the most important information.

[0057] In some embodiments, the deep learning recognition module also adopts a Transformer network structure. The biggest advantage of this structure is that it can process global information and capture long-range dependencies between different regions in the image through a self-attention mechanism.

[0058] The core of Transformer is the self-attention mechanism, which can assign weighted values ​​to each feature by calculating the relationship between input features and strengthen the feature information of key areas. The mathematical formula of the self-attention mechanism is: ; in: is the query matrix, which represents the features of the current input image; is the key matrix, representing the features of other regions in the image; is a value matrix containing feature information of all regions in the image; is the dimension of the key matrix, used to normalize the calculation results; is a normalization function that ensures that the sum of all weighting coefficients is 1.

[0059] The self-attention mechanism calculates and The dot product of is used to measure the relationship between different regions of the image. Then, The weighted sum of will generate a new feature representation. In this way, Transformer can capture the mutual influence between distant regions in the image and enhance the understanding of the global information in the palm vein structure.

[0060] Specifically, through the self-attention mechanism of Transformer, the model can identify the association between various regions in the palm vein image, for example, the relationship between the palm area and the edge area, and the relationship between the veins and the background. This feature makes Transformer very effective in identifying complex vein structures, especially when faced with multiple palm postures and complex backgrounds.

[0061] In the deep learning recognition module, the local features extracted by CNN and the global features extracted by Transformer will be merged in the subsequent fusion layer. Through the fully connected layer, these extracted features will be fused into a complete feature vector that represents the overall information of the image.

[0062] The fused feature vector is passed to the identity matching module and compared with the vein features in the database. For this purpose, cosine similarity or Euclidean distance is used as the matching metric.

[0063] By calculating the cosine similarity between the feature vectors, the similarity between the image and the vein features stored in the database can be determined. If the similarity exceeds a preset threshold, the image is considered to have been successfully matched to the user identity.

[0064] During the training process, the deep learning recognition module uses supervised learning methods and is trained through annotated palm vein datasets. The network updates weights and biases through the back propagation algorithm to optimize the performance of the model. The cross entropy loss function is used as the objective function, and the formula is: ; in: is the true label, indicating the category to which the sample belongs; is the class probability predicted by the model.

[0065] By minimizing the loss function, the network gradually optimizes its parameters, thereby improving the accuracy of image recognition.

[0066] The recognition results output by the deep learning recognition module will be transmitted to the attendance record storage module through the control and communication module for recording and management. The recognition results include identity information, attendance status, recognition confidence, etc., ensuring the rapid confirmation of user identity and the security of data storage.

[0067] In this embodiment, the attendance record storage module ensures that each attendance record can be stored in the database in real time and accurately, and has powerful data query and report generation functions. This module also takes into account the security and privacy protection of data, uses encryption technology to process sensitive data, and provides stable and reliable services through efficient data management methods.

[0068] Generally, the attendance record storage module receives the identification results from the deep learning identification module, which outputs the user's attendance status, identification confidence, clock-in time and other information. The attendance record storage module stores this data in a structured format in the database and provides efficient query functions when needed. As user data continues to increase, the module will also perform regular backup and archiving operations to ensure the long-term preservation and security of data.

[0069] In this embodiment, the attendance record storage module uses a relational database system (such as MySQL or PostgreSQL) for data storage. The database contains multiple tables, each of which records different types of attendance data. Each attendance record contains the user's identity information, attendance time, attendance status, recognition confidence, and other information.

[0070] Specifically, the storage structure of attendance records is as follows: ; in: A unique identifier for the user (such as a user ID or palm vein template ID) to ensure that each attendance record can be associated with a specific user; Attendance time, which records the specific time when the user punches in, usually expressed in UTC time format; Attendance status, indicating whether the attendance is successful, such as "successful" or "failed"; The confidence of recognition indicates the confidence of the recognition module in matching the identity, usually expressed as a floating value between 0 and 1; It is a device identifier that records the device information that initiates the attendance request to ensure that the source of the attendance data can be traced; Geolocation data records the location of the attendance device, which helps prevent fraud (for example, remote clocking in).

[0071] Each attendance record is stored in the database through Unique identification ensures the integrity and uniqueness of data storage.

[0072] Considering the privacy protection and data security of users, the attendance record storage module in this embodiment uses an encryption algorithm to encrypt and store attendance data. Sensitive information such as user ID, recognition confidence, and attendance time will be encrypted using a symmetric encryption algorithm (such as AES). Specifically, each attendance record is encrypted using the AES-256 encryption standard to ensure the security of data during storage and transmission.

[0073] The encryption algorithm process is as follows: ; in: is the encrypted data (ciphertext); Plain text data (such as user ID, attendance status, and other sensitive information); is the encryption key (the key length is 256 bits).

[0074] To ensure the security of the key, this module uses a key management system (KMS) to regularly update the key and dynamically generate the key in each encryption operation. By adopting a secure key management strategy, the long-term security of the key is guaranteed.

[0075] In the database, the encrypted data stored can only be decrypted and viewed by authorized users or administrators after authentication. This mechanism ensures that attendance data will not be accessed or leaked without authorization.

[0076] The attendance record storage module relies on an efficient database management system to handle a large number of concurrent read and write operations. In some embodiments, a database engine that supports transaction processing (such as InnoDB) is used to ensure data consistency and persistence when multiple users operate simultaneously.

[0077] This module follows the ACID (atomicity, consistency, isolation, and persistence) characteristics to ensure that the write operation of each attendance record is atomic, that is, the operation is either completely successful or completely failed. Specifically, the data storage operation uses transaction control.

[0078] Before a transaction is committed, any unfinished operations will not affect existing data in the database. In the event of an error or system crash, the database will roll back the transaction and restore it to the state before the operation, thus avoiding data loss or errors.

[0079] The attendance record storage module supports efficient query functions, and administrators can quickly retrieve attendance records based on different query conditions (such as time, status, user ID, etc.). The query function supports flexible query statements based on SQL, and can extract data according to needs.

[0080] In some embodiments, this module also provides a query cache mechanism to store the results of recent queries to improve the response speed of frequent queries. The query can quickly return all successful attendance records within a certain time range.

[0081] In order to facilitate the later analysis and management of attendance data, the attendance record storage module supports automatic report generation. Managers can automatically generate daily or monthly attendance statistics reports according to the set rules. The generated reports can be exported to PDF or Excel format for easy archiving and sharing.

[0082] In order to ensure the security and long-term availability of data, in this embodiment, the attendance record storage module provides a regular backup function. The system will automatically back up the database every day or every week and save all attendance records in a secure storage medium. The backed-up data will be compressed and its content will be protected by encryption technology.

[0083] The data recovery process can be carried out by following the steps below: Unzip and decrypt the backup file; Use SQL scripts to restore backup data to the database; Verify the integrity of recovered data to ensure no data loss.

[0084] The backup mechanism and disaster recovery capability provided by the system can effectively ensure that attendance data can be quickly restored in the event of a system failure or catastrophic event.

[0085] In this embodiment, the control and communication module is responsible for coordinating the work of each submodule in the system and ensuring the data flow and communication between each module. Specifically, the control and communication module effectively processes the data from the palm vein data acquisition module, the image optimization processing module, the deep learning recognition module, and the attendance record storage module. In this embodiment, the control and communication module uses a standard communication protocol for data transmission to ensure efficient and reliable data transmission. Commonly used communication protocols such as TCP / IP, HTTP, WebSocket, etc. are widely used in the data interaction of this module. Specifically, the data transmission between the module and each subsystem adopts the TCP / IP protocol to ensure the stable transmission of data in the network.

[0086] In some embodiments, the control and communication module communicates with other system modules through a RESTful-API interface. RESTful-API provides a simple and lightweight way of exchanging data. The control module transmits information between modules through API requests and responses.

[0087] Considering the security and privacy of data, the control and communication module adopts multi-level data encryption technology. During the transmission process, the module uses HTTPS (encrypted HTTP protocol) to ensure the security of data transmission. All sensitive data (such as user ID, attendance status, etc.) are transmitted after encryption to prevent man-in-the-middle attacks or data leakage.

[0088] During data transmission, the SSL / TLS protocol is used to encrypt the communication to ensure that the data will not be eavesdropped or tampered with when it is transmitted on the network. Specifically, the SSL / TLS encryption protocol encrypts the communication by using public and private keys to avoid unauthorized access.

[0089] The encryption process is as follows: ; in: is the encrypted data; The original data to be encrypted (such as attendance records); is the encryption key (using the AES-256 encryption algorithm).

[0090] In a possible implementation, to ensure the security of the key, the control and communication module manages the key by means of periodic updating and dynamic key exchange, so as to reduce the risk of encryption key leakage.

[0091] The control and communication module is not only responsible for data transmission, but also needs to coordinate the execution of other modules. The task scheduling function of the module ensures that each submodule is executed in a predetermined order. For example, the control module is responsible for managing the time nodes of data acquisition and recognition to ensure that the image acquisition and recognition processes are carried out synchronously. According to the attendance requirements, the control module can monitor the entire attendance process in real time to ensure the smooth flow of data.

[0092] In some embodiments, the control and communication module schedules the work of each module through a timed task. For example, at a specific time of the day, the control module can automatically start the attendance process, collect the attendance data of all users, and upload the recognition results to the attendance database. The scheduling of this process is triggered by time, avoiding the need for manual intervention.

[0093] In some embodiments, the control and communication module also supports data synchronization between multiple devices, especially in scenarios where multiple users' clock-in data needs to be processed simultaneously. The system synchronizes the attendance data between devices through a local area network (LAN) or a wide area network (WAN), thereby ensuring the consistency and accuracy of attendance records.

[0094] To this end, the control module will synchronize device data regularly or in real time, and aggregate attendance information on different devices to the central server through the data synchronization mechanism. This mechanism is particularly suitable for distributed attendance systems, such as scenarios where multiple class cards or access control devices are distributed in different locations. In specific implementation, the module will encrypt the data of each attendance punch-in and transmit it to the central server for unified management and storage.

[0095] The control and communication module is not only responsible for data transmission, but also has a real-time response function. In some embodiments, the control and communication module supports real-time feedback to the user device, such as displaying the attendance results in real time through the class card device. When the user completes the clock-in operation and is successfully recognized, the system will immediately feedback the user's attendance status, such as displaying "Card-in successful" or "Recognition failed" on the screen. This real-time feedback can ensure that the user has a clear understanding of the attendance process.

[0096] During the response process, the control and communication module will generate a corresponding message according to the data received from the identification module, and feed it back to the user equipment through the communication interface.

[0097] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An electronic attendance card system for palm vein recognition, characterized in that: include: A palm vein data acquisition module, used to illuminate the user's palm through a near-infrared light source and obtain a palm vein image; Image optimization processing module, used for optical modeling optimization, noise removal and quality enhancement of the collected palm vein images; Deep learning recognition module, used to extract palm vein features and perform identity matching; Attendance record storage module, used to store and manage user attendance data; Control and communication module, used to manage the collaborative work of each module and exchange data with the external attendance database; Among them, the image optimization processing module calculates the propagation path of light in skin tissue based on the light transmission equation and adjusts the image contrast to optimize the visibility of palm vein features.

2. The electronic attendance card system for palm vein recognition according to claim 1 is characterized in that: The palm vein data acquisition module uses a near-infrared optical camera to illuminate the user's palm with near-infrared light of a wavelength of 760nm to 850nm, and receives transmission or reflection signals through a photoelectric sensor to form a palm vein image.

3. The electronic attendance card system for palm vein recognition according to claim 1 is characterized in that: The image optimization processing module is based on the light transmission equation: ； in, Pixel The light intensity at is the incident light intensity, is the light absorption coefficient, is the propagation path length of light in skin tissue, is the base of natural logarithms; The image optimization processing module calculates the transmittance based on the equation and optimizes the contrast of the palm vein image by adjusting the illumination parameters and the image enhancement algorithm.

4. The electronic attendance card system for palm vein recognition according to claim 1, characterized in that: The image optimization processing module performs denoising and edge preservation on the palm vein image based on a nonlinear variational method, and its optimization objective function is defined as: ； in, For the optimized image, is the original collected image, To control the parameters of edge preservation ability, is the regularization weight, is the image domain, Representing the image domain A small area on the surface is used to calculate the energy function , by solving the variational equation, the image quality is optimized and the edge structure of the palm vein is enhanced.

5. The electronic attendance card system for palm vein recognition according to claim 4 is characterized in that: The nonlinear variational method uses a gradient descent method for numerical solution, and performs geometric optimization on the palm vein image based on the Laplace-Beltrami operator.

6. The electronic attendance card system for palm vein recognition according to claim 1, characterized in that: The deep learning recognition module adopts a neural network model including a convolutional neural network and a Transformer structure. The convolutional neural network is used to extract local features of palm veins, and the Transformer structure is used to model global patterns to complete palm vein identity matching.

7. The electronic attendance card system for palm vein recognition according to claim 1 is characterized in that: The Transformer structure of the deep learning recognition module calculates the palm vein feature map based on the self-attention mechanism, and the calculation method is: ； in, , , are the query matrix, key matrix and value matrix of the input features respectively, is the feature dimension, is the key matrix The transposition of is used to improve the accuracy of palm vein feature matching through the self-attention mechanism.

8. The electronic attendance card system for palm vein recognition according to claim 1 is characterized in that: The deep learning recognition module performs image optimization based on a generative adversarial network. The adversarial network generator is used to optimize the quality of the palm vein image, and the discriminator is used to evaluate the consistency between the generated image and the real palm vein image. The objective function of the generator is defined as: ； in, is the discriminator’s result, The image generated by the generator, is the real data distribution, is a random noise distribution, For the generator, is the discriminator, is a random variable representing a sample sampled from a certain distribution. For real data The expected value of It represents the expected value operation, which usually refers to solving the average value or weighted average value of a random variable. The generator is optimized through this objective function so that the generated image is closer to the real palm vein image.

9. The electronic attendance card system for palm vein recognition according to claim 1, characterized in that: The attendance record storage module includes a database management system for storing user identity information, attendance time, attendance status, and providing a data synchronization interface to support remote access and management.

10. The electronic attendance card system for palm vein recognition according to claim 1, characterized in that: The control and communication module uses an encrypted communication protocol to protect data transmission and interacts with the attendance management platform via wireless communication.

Images