A Facial Recognition-Based Logistics Weighing Safety Control System

Abstract

The application discloses a logistics weighing safety control system based on face recognition and relates to the technical field of logistics transportation management.The system comprises a collection module, which is used for collecting the images of vehicles and drivers when the vehicles are weighed, and sends the processed images to a recognition module; the recognition module receives the images sent by the collection module, identifies the face images and vehicle images therefrom, extracts face feature information and vehicle information, and sends the face feature information and the vehicle information to a comparison module; and a storage module, which is used for storing vehicle and face registration information and receiving the vehicle information and the face feature information transmitted by the comparison module.The logistics weighing safety control system based on face recognition has significant technical advantages and practical value in improving the identity verification strength, safety prevention and control capability and management intelligent level of the logistics weighing link.

Description

Technical Field

[0001] This invention relates to the field of logistics transportation management technology, specifically to a logistics weighing safety control system based on facial recognition. Background Technology

[0002] In modern logistics and transportation, cargo weighing is a crucial step in ensuring logistics management, contract execution, and cargo safety. Truck scales are typically used to weigh the cargo carried by transport vehicles, and the weighing results serve as important data in the logistics process. Especially during the entry and exit of transport vehicles from factory areas, loading and unloading, and weighing, it is often necessary to verify the driver's identity and business information to ensure the authenticity and validity of the weighing data.

[0003] However, certain management loopholes still exist in the current logistics management process. For example, when some transport vehicles weigh goods at weigh stations, verification is based solely on entry and exit registration information, lacking an effective anti-fraud mechanism. This makes it easy for individuals to cheat by changing license plates or manipulating vehicle tare weight, resulting in repeated transport, understating net weight, or even theft of goods for illegal profit. This behavior not only seriously interferes with logistics companies' accurate control over the flow of goods but also harms the interests of cargo owners and carriers. Furthermore, with the continuous increase in logistics operation processes, the information management of vehicles and drivers in the transportation process has become increasingly complex. During vehicle transfer operations, logistics companies often need to re-register new vehicles and conduct identity verification and tracking to ensure the continuity and accuracy of cargo transportation. This not only increases the complexity of operations but also the workload of data management, making it easy for oversights to occur in information verification and cargo identification, affecting overall logistics efficiency and increasing the risk of errors. Summary of the Invention

[0004] The purpose of this invention is to provide a logistics weighing safety control system based on facial recognition, thereby solving the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a logistics weighing safety control system based on facial recognition, the system comprising:

[0006] The data acquisition module is used to acquire images of the vehicle and driver when the logistics vehicle is being weighed, and then send the processed images to the recognition module.

[0007] The recognition module receives images sent by the acquisition module, identifies face images and vehicle images from them, extracts facial feature information and vehicle information, and sends the facial feature information and vehicle information to the comparison module. This includes acquiring three consecutive frames of images, calculating the similarity of facial features between consecutive frames, obtaining the standard deviation, and taking the reciprocal of the standard deviation as the stability of face feature recognition. Similarly, the vehicle image recognition results are processed to obtain the stability of vehicle image recognition. The stability ratio of face recognition to vehicle recognition is calculated to determine whether to prioritize face results or vehicle recognition results, and the information is merged and sent to the comparison module.

[0008] The comparison module queries the registration information of vehicle and driver faces recorded in the storage module. When no matching vehicle and driver face information is found, the current vehicle information and facial feature information are processed and sent to the storage module for storage. When matching vehicle and driver face information is found, the received vehicle and driver face information is compared with the vehicle and driver facial feature information corresponding to the previously stored vehicle and driver face information. The comparison result is then sent to the decision module.

[0009] The judgment module makes a judgment on the received comparison results and decides whether to allow the vehicle to pass based on the actual situation. If the judgment fails, it is transferred to the manual processing and then returned to the acquisition module.

[0010] The storage module is used to store vehicle and driver facial registration information, as well as receive vehicle information and facial feature information transmitted by the comparison module.

[0011] Preferably, when the driver swipes their ID card, the acquisition module acquires images of the vehicle and the person, calls an image enhancement algorithm to estimate the improvement in recognition success rate after image quality processing, and simultaneously calculates the increased computational resource consumption to improve image quality. It also calculates the marginal utility of image quality. If the marginal utility of image quality is greater than 1, advanced image enhancement is performed and the image is sent to the recognition module; otherwise, the original image is sent directly to save resources.

[0012] Preferably, the specific formula for calculating the marginal utility of image quality is: A = B / C;

[0013] Where A represents the marginal utility of image quality, B represents the improvement in recognition success rate after image quality processing, and C represents the increased computational resource consumption required to improve image quality.

[0014] Preferably, the specific formula for calculating the stability ratio of face and vehicle recognition is: R = H / D;

[0015] Where R represents the face and vehicle recognition stability ratio, H represents the reciprocal of the standard deviation of the face similarity change in consecutive frames, and D represents the vehicle image recognition stability.

[0016] Preferably, the comparison module monitors the total number of comparison requests currently being processed and the number of comparisons completed at the current time, sets a maximum comparison capacity, calculates the current comparison efficiency growth trend, and if the current comparison efficiency growth trend is less than 0, switches to a lightweight comparison strategy; otherwise, it uses complete features for high-precision comparison and sends the comparison results to the decision module.

[0017] Preferably, the specific formula for calculating the current comparison efficiency growth trend is as follows:

[0018] dV / dt=βV(1-M / C;

[0019] Where dV / dt represents the current alignment efficiency growth trend, β represents the alignment efficiency coefficient, V represents the number of alignments completed at the current time, M represents the total number of alignment requests currently being processed, and C represents the set maximum alignment capacity.

[0020] Preferably, the decision module obtains the similarity score returned by the comparison module, queries the driver's historical passage records in the database, calculates the current driver's historical passage success rate, evaluates the image recognizability score based on the current image acquisition time and lighting conditions, sets weights using experience, calculates the total decision score, and if the total decision score is greater than a preset threshold, it automatically allows passage; otherwise, it refuses passage and issues an alarm.

[0021] Preferably, the specific formula for calculating the total judgment score is: S = w1 × E + w2 × F + w3 × G;

[0022] Where S represents the total judgment score, w1, w2 and w3 represent weights, satisfying w1 + w2 + w3 = 1, E represents the similarity score, F represents the current driver's historical success rate, and G represents the image recognizability score based on the current image acquisition time and lighting conditions.

[0023] Preferably, the storage module scans the access frequency of all driver identities in the database at regular intervals every day, calculates the frequency of occurrence of each driver identity, and calculates the identity data distribution entropy value as an indicator of identity diversity distribution. If the identity data distribution entropy value decreases, an abnormal data cleanup mechanism is activated; if the identity data distribution entropy value increases, the index depth and cache space are automatically increased.

[0024] Preferably, the specific method for calculating the identity data distribution entropy value includes calculating the relative frequency of each registered independent identity, multiplying each frequency by the logarithm of the frequency, summing the product results and taking the opposite number, and finally obtaining the value as the identity data distribution entropy value.

[0025] As can be seen from the above technical solution, the present invention has the following beneficial effects:

[0026] This logistics weighing safety control system based on facial recognition integrates facial recognition with vehicle image information comparison to achieve dual verification of driver identity and vehicle. This effectively avoids the risk of misuse associated with relying solely on registration information, significantly improving the credibility of weighing data. By comparing currently collected facial features with existing registration records, it identifies whether the same driver or vehicle frequently engages in abnormal weighing operations, preventing illegal activities such as smuggling or undercounting goods by exploiting tare weight differences. In the logistics vehicle changeover process, the system automatically identifies the characteristics of new vehicles and their drivers, eliminating the need for manual data entry. It can quickly complete identity verification and data updates, reducing manual intervention costs and information entry errors. Through modular collection, recognition, comparison, and storage mechanisms, the system can form a complete identity trajectory and weighing record chain, facilitating subsequent traceability and risk analysis, enhancing enterprises' control over cargo flow. The automatic information comparison and judgment mechanism significantly reduces the burden of manual verification, avoids human error, and improves the automation level and overall operational efficiency of logistics operations. This invention has significant technical advantages and practical value in improving the strength of identity verification, security control capabilities, and intelligent management level in the logistics weighing process. Attached Figure Description

[0027] Figure 1 This is a connection diagram of the system modules of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] like Figure 1 As shown, the present invention provides a technical solution: a logistics weighing safety control system based on facial recognition, the system comprising:

[0030] The data acquisition module is used to acquire images of the vehicle and driver when the logistics vehicle is being weighed, and then send the processed images to the recognition module.

[0031] The recognition module receives images sent by the acquisition module, identifies face images and vehicle images from them, extracts facial feature information and vehicle information, and sends the facial feature information and vehicle information to the comparison module. This includes acquiring three consecutive frames of images, calculating the similarity of facial features between consecutive frames, obtaining the standard deviation, and taking the reciprocal of the standard deviation as the stability of face feature recognition. Similarly, the vehicle image recognition results are processed to obtain the stability of vehicle image recognition. The stability ratio of face recognition to vehicle recognition is calculated to determine whether to prioritize face results or vehicle recognition results, and the information is merged and sent to the comparison module.

[0032] The comparison module queries the registration information of vehicle and driver faces recorded in the storage module. When no matching vehicle and driver face information is found, the current vehicle information and facial feature information are processed and sent to the storage module for storage. When matching vehicle and driver face information is found, the received vehicle and driver face information is compared with the vehicle and driver facial feature information corresponding to the previously stored vehicle and driver face information. The comparison result is then sent to the decision module.

[0033] The judgment module makes a judgment on the received comparison results and decides whether to allow the vehicle to pass based on the actual situation. If the judgment fails, it is transferred to the manual processing and then returned to the acquisition module.

[0034] The storage module is used to store vehicle and driver facial registration information, as well as receive vehicle information and facial feature information transmitted by the comparison module.

[0035] This system employs a modular design to automatically identify and verify the identity information of vehicles and drivers during the logistics weighing process. When a logistics vehicle is being weighed, the acquisition module activates and simultaneously acquires images of the front of the vehicle and the driver's cab, thereby obtaining image data containing vehicle license plate information and driver facial information. The acquisition module preprocesses the raw images, including image denoising, edge enhancement, and contrast adjustment, to improve the accuracy of subsequent recognition.

[0036] Subsequently, the recognition module analyzes the processed images, performing face detection and license plate localization based on deep learning algorithms such as convolutional neural networks. Keypoint localization algorithms extract facial feature points and license plate character outline regions, calculating and extracting the corresponding feature vectors. To improve recognition stability, the system continuously acquires three frames of images and extracts features from each frame. Then, it calculates the cosine similarity or Euclidean distance between the feature vectors of adjacent frames, obtaining a set of similarity values. The system further calculates the standard deviation of this set of similarity values, taking its reciprocal as a recognition stability index; higher recognition stability indicates better stability of the target within the image sequence.

[0037] The system then compares the stability of face and vehicle recognition, using a set threshold or weighted algorithm to determine which recognition result should be used as the primary basis for identification. The selected recognition information (face or license plate features) and the other auxiliary information are sent to the comparison module. The comparison module retrieves historical registration information from the storage module and builds a feature index library, then matches the received image information using a feature vector comparison algorithm. If no matching record is found, the current image features are automatically bound to the ID card information to generate a new file record; if a matching record exists, feature comparison is performed to confirm whether the driver's identity matches the vehicle.

[0038] The comparison results are sent to the decision module, which determines whether to allow the vehicle to pass based on factors such as the similarity score, recognition stability, and past recognition history, using fuzzy logic reasoning or rule-based decision trees. If the decision result is "not passed," the system marks the event as abnormal and automatically sends it back to the data acquisition module, while simultaneously issuing a prompt for manual review to prevent impersonation or substitution.

[0039] This invention improves the security and accuracy of driver and vehicle identity verification in logistics weighing processes. It enhances the system's robustness in complex environments by utilizing a multi-frame image recognition stabilization mechanism, and further improves recognition efficiency by optimizing the priority of face or vehicle recognition through a dynamic comparison strategy. The system's automatic comparison and decision-making mechanism significantly reduces the frequency of manual intervention and improves weighing efficiency. Simultaneously, the system supports historical storage and reuse of vehicle information, enhancing its information management capabilities during long-term operation and demonstrating good scalability and application prospects.

[0040] When a logistics vehicle is being weighed, the acquisition module collects images of the vehicle and personnel, calls an image enhancement algorithm to estimate the improvement in recognition success rate after image quality processing, and calculates the increased computational resource consumption to improve image quality. It also calculates the marginal utility of image quality. If the marginal utility of image quality is greater than 1, advanced image enhancement is performed and the image is sent to the recognition module; otherwise, the original image is sent directly to save resources.

[0041] When a vehicle enters the weighing area, the acquisition module is triggered and begins acquiring images of the scene. The acquisition module simultaneously captures images containing the vehicle's exterior and the driver's face using a configured high-definition camera, and calls preset image enhancement algorithms (such as Retinex, CLAHE, or deep neural network-based image super-resolution reconstruction models) to assess the quality of the acquired images.

[0042] First, the system comprehensively evaluates the acquired image based on indicators such as image sharpness, brightness uniformity, and edge detail preservation. Combining this with historical image recognition data, it estimates the potential increase in recognition success rate after image quality improvement. Next, the system calls the performance analysis module to assess the computational resources required to execute the image enhancement algorithm, including CPU / GPU time usage, memory consumption, and I / O bandwidth usage, quantifying its cost.

[0043] The system uses the improvement in recognition success rate as a benefit indicator and resource consumption as a cost indicator, calculating the marginal utility of image quality improvement (i.e., the ratio of recognition accuracy improvement per unit of resource cost). If the marginal utility value is greater than 1, the system determines that the current resource investment has a positive benefit, executes image optimization processing with a higher enhancement level, and transmits the enhanced image to the recognition module. If the marginal utility is less than or equal to 1, the system considers the image optimization input-output ratio to be low, and directly sends the original image to the recognition module, thereby saving computing resources and improving the overall processing efficiency of the system. This mechanism essentially establishes a real-time dynamic balance between image processing benefits and resource costs, ensuring that recognition accuracy is maintained without resource waste, making it particularly suitable for edge computing or resource-constrained scenarios.

[0044] This implementation effectively improves the intelligence and resource utilization efficiency of the image acquisition stage. By using an image quality marginal utility evaluation method, the system can intelligently decide whether to perform image enhancement operations. This avoids unnecessary image processing when image quality is high, reducing system load and improving response speed; while prioritizing recognition accuracy during critical recognition moments despite lower image quality, it enhances system reliability and adaptability. This mechanism is highly adaptable and particularly suitable for practical applications with frequent image quality fluctuations or limited device computing resources. It contributes to the long-term stable operation of the system and provides a fundamental guarantee for the stability of subsequent recognition accuracy.

[0045] The specific formula for calculating the marginal utility of image quality is: A = B / C;

[0046] Where A represents the marginal utility of image quality, B represents the improvement in recognition success rate after image quality processing, and C represents the increased computational resource consumption required to improve image quality.

[0047] During the image acquisition and preprocessing stage, the system evaluates the image enhancement effect and computational resource consumption of the acquired images, and then uses a marginal utility model to make decisions. The marginal utility of image quality is calculated by the formula A = B / C, where B is the improvement in recognition success rate brought by the image after the enhancement algorithm compared to the original image. The system can estimate this value by comparing the difference in recognition confidence between the original image and the enhanced image under existing model recognition or the historical accuracy improvement. C is the computational resource cost required to execute the enhancement algorithm under current hardware conditions, including but not limited to image processing time, percentage of CPU / GPU resources used, memory usage, etc., expressed as a unified resource consumption index using a weighted summation method. The system calculates the cost-effectiveness A value of the image enhancement operation using the above formula, that is, the degree of improvement in recognition accuracy that can be obtained by unit computational resource consumption. If the A value is greater than 1, it indicates that the enhancement operation has a positive benefit under the current conditions, and the system will execute the image enhancement operation; if the A value is less than or equal to 1, it is considered that the resource input-output ratio is not good, and the system chooses to skip the enhancement process and directly use the original image to improve the overall resource utilization efficiency and response speed.

[0048] By introducing a marginal utility calculation model, the system achieves quantitative analysis and intelligent decision-making in image processing strategies, avoiding the blind consumption of resources. While ensuring recognition quality, it effectively reduces the computational burden, improves the flexibility of the image processing stage, and enhances the stability of system operation. The use of explicit formula modeling provides a quantitative foundation for subsequent system optimization and resource scheduling algorithm integration, enabling real-time control over the benefits and costs of image processing and improving the transparency and feasibility of algorithmic strategies.

[0049] The specific formula for calculating the stability ratio of face and vehicle recognition is: R = H / D;

[0050] Where R represents the face and vehicle recognition stability ratio, H represents the reciprocal of the standard deviation of the face similarity change in consecutive frames, and D represents the vehicle image recognition stability.

[0051] The system incorporates a recognition stability ratio (R) calculation in its recognition module to dynamically assess whether face recognition or vehicle image recognition should be prioritized in the recognition task. Specifically, after receiving three consecutive frames of image data from the acquisition module, the system extracts features from both the face and vehicle images, calculating the image feature similarity between adjacent frames. For the face image similarity results, the system calculates the standard deviation, denoted as S, and then takes its reciprocal, 1 / S, to obtain the face recognition stability H, which measures the consistency of face features across consecutive frames. The vehicle image processing flow is similar, calculating the vehicle recognition stability D. Subsequently, the system uses the formula R = H / D to obtain the face-to-vehicle recognition stability ratio (R). This value characterizes the relative stability of face recognition compared to vehicle recognition, guiding the selection of the priority criterion. When the R value is greater than a set threshold (e.g., 1.0), it indicates that face recognition is more stable than vehicle recognition, and the system will use the face recognition result as the primary recognition criterion, supplemented by vehicle image information; otherwise, the vehicle recognition result will be the primary criterion for subsequent comparison operations. This dynamic adjustment mechanism based on the R value significantly improves the adaptive capability of the recognition module, ensuring that reasonable recognition judgments can be made based on data stability in different scenarios.

[0052] Introducing the recognition stability ratio R as a decision parameter helps the system objectively prioritize information from multiple sources. This method avoids erroneous judgments caused by differences in image quality or changes in the recognition environment, improving the robustness and accuracy of face and vehicle recognition processes. By quantifying recognition stability using the reciprocal of the standard deviation and further constructing the ratio R, the system obtains a real-time adaptive decision logic that requires no external annotation, demonstrating high engineering practicality and algorithm scalability. Furthermore, this method facilitates algorithm optimization and hardware resource allocation adjustments, providing a quantitative basis for intelligent upgrades to recognition strategies.

[0053] The comparison module monitors the total number of comparison requests currently being processed and the number of comparisons completed at the current time, sets the maximum comparison capacity, calculates the current comparison efficiency growth trend, and if the current comparison efficiency growth trend is less than 0, switches to a lightweight comparison strategy; otherwise, it uses complete features for high-precision comparison and sends the comparison results to the decision module.

[0054] This system incorporates a dynamic adjustment mechanism in its comparison module, aiming to adjust the comparison strategy based on real-time system load and processing performance, thereby optimizing the overall recognition process's response efficiency. During operation, the comparison module monitors the total number of comparison requests entering the queue and the number of comparison operations completed per unit time in real time. It calculates the comparison processing rate based on time series data and derives its first derivative, representing the comparison efficiency growth trend. The system presets a maximum comparison capacity, representing the optimal parallel processing capability under the current hardware environment. When the comparison efficiency growth trend is positive, it indicates that the system's processing capacity is still improving or remaining stable. The module continues to use complete features (such as 128-dimensional facial vectors and vehicle image depth features) for high-precision comparison to ensure recognition accuracy. When the growth trend is less than 0, indicating a decrease in processing efficiency per unit time, it suggests a potential performance bottleneck or task backlog. The module immediately switches to a lightweight comparison strategy, such as using low-dimensional feature extraction, downsampling images, and feature template matching to reduce computational load and alleviate system pressure. After comparison, regardless of the strategy used, the results are sent in real-time to the decision module for subsequent passage determination operations. Through this mechanism, the comparison module can flexibly adjust between ensuring accuracy and efficiency, thereby improving the overall processing stability and concurrent processing capabilities of the system.

[0055] By incorporating the comparison efficiency growth trend as the basis for strategy switching, the system possesses adaptive computational load adjustment capabilities, effectively mitigating the impact of sudden traffic spikes or hardware performance fluctuations on the identification process. Under high load, automatically activating a lightweight strategy significantly shortens response time and prevents system congestion caused by comparison task backlog; under manageable load, high-precision features are used to ensure identification accuracy, thus achieving a balance between resource utilization and security control. Furthermore, this strategy adjustment mechanism is simple, quantifiable, and real-time, facilitating integration into various heterogeneous systems and demonstrating good engineering feasibility.

[0056] The specific formula for calculating the current comparison efficiency growth trend is as follows:

[0057] dV / dt=βV(1-M / C;

[0058] Where dV / dt represents the current alignment efficiency growth trend, β represents the alignment efficiency coefficient, V represents the number of alignments completed at the current time, M represents the total number of alignment requests currently being processed, and C represents the set maximum alignment capacity.

[0059] To achieve dynamic adjustment of the comparison strategy, the comparison module uses the formula dV / dt = βV(1 - M / C) to calculate the current comparison efficiency growth trend. Here, dV / dt represents the rate of change of comparison efficiency per unit time, reflecting the impact of the system's current load on its comparison capabilities. V is the number of successfully completed comparisons per unit time at the current moment, a direct indicator of the current actual processing capacity. M is the total number of comparison requests currently waiting to be processed, reflecting the current system load intensity; C is the maximum comparison capacity, representing the upper limit of the system's processing capacity under stable operation, typically determined by the system's hardware capabilities and concurrent processing strategy; β is the comparison efficiency coefficient, a positive real number used to adjust the system's sensitivity, optimized based on historical operating data. This model employs a structure similar to the logistic growth model. When M / C approaches 1, the parenthetical terms approach 0, the rate of change in alignment efficiency slows down, indicating that the system is in a saturated state. If M / C < 1, the system still has spare capacity, dV / dt is positive, and the alignment efficiency continues to increase. If M / C > 1, dV / dt is negative, indicating that the system load exceeds the capacity limit, and processing efficiency will decrease. The alignment module dynamically determines whether to switch to a lightweight alignment strategy based on the positive or negative state of dV / dt, thereby controlling the system load while ensuring response efficiency.

[0060] By introducing a clear mathematical model, the system can quantify the trend of comparison efficiency changes in real time, providing a mathematical basis and automated logic for switching comparison strategies, avoiding reliance on manually set experience values ​​or single threshold judgments. This model can provide early warnings of system bottlenecks, effectively preventing task backlog and decreased processing efficiency. Based on the model results, the system adaptively optimizes the recognition strategy, helping to improve the balance between processing stability, efficiency, and accuracy. Its simple form, adjustable parameters, and good generalization ability make it valuable for engineering deployment.

[0061] The decision module obtains the similarity score returned by the comparison module, queries the driver's historical passage records in the database, calculates the current driver's historical passage success rate, evaluates the image recognizability score based on the current image acquisition time and lighting conditions, sets weights using experience, calculates the total decision score, and if the total decision score is greater than the preset threshold, it automatically allows passage; otherwise, it refuses passage and issues an alarm.

[0062] The system incorporates a multi-factor fusion decision-making mechanism in its decision-making module to improve the overall accuracy and adaptability of vehicle and driver identification. After the comparison module completes the comparison between the image and the database information, it sends the similarity score to the decision-making module, which then triggers a multi-dimensional scoring process upon receiving the score.

[0063] First, the decision module retrieves historical passage records from the storage module to query the current driver's passage history, including past recognition success rates and the presence of any abnormal records. Based on these records and the total number of attempts, the system calculates the historical passage success rate, which serves as one of the key parameters for determining the driver's credibility. Next, the system evaluates the identifiability of the current image. This evaluation is based on the time period of image acquisition (e.g., daytime or nighttime) and ambient lighting information (e.g., brightness values, contrast distribution) collected simultaneously by the system, quantifying and generating an image identifiability score to characterize the impact of image input quality on the recognition result.

[0064] The decision-making module sets weight parameters for each factor based on experience or training data. For example, it assigns different weights to similarity scores, historical success rates, and image recognizability scores, and calculates the total decision score through a weighted summation. If the score is higher than the set safety threshold, the system automatically determines "identity passed," triggers a release signal, and controls the equipment to allow the vehicle to pass. If the score is lower than the threshold, the system marks the current event as abnormal, refuses vehicle passage, and simultaneously sends an alarm signal to the backend system, prompting manual intervention or review. This decision-making mechanism combines static information with dynamic environmental factors to construct a more comprehensive identity determination logic.

[0065] This invention significantly improves the intelligence of the decision-making module and the reliability of the recognition results by introducing historical behavioral data and image environment adaptability analysis. The system no longer relies solely on a single similarity value to determine access permissions, but instead integrates multi-source data for comprehensive analysis, effectively reducing false recognition and false negative rates, and maintaining stable judgment capabilities even in complex scenarios such as insufficient lighting and blurred images. By setting empirical weights and a total decision score threshold, the system can flexibly adapt to different scenarios and user security strategies, improving overall operational efficiency and security. Furthermore, the alarm mechanism enables real-time feedback of abnormal events, providing data support for system management and event tracing.

[0066] The specific formula for calculating the total judgment score is: S = w1 × E + w2 × F + w3 × G;

[0067] Where S represents the total judgment score, w1, w2 and w3 represent weights, satisfying w1 + w2 + w3 = 1, E represents the similarity score, F represents the current driver's historical success rate, and G represents the image recognizability score based on the current image acquisition time and lighting conditions.

[0068] When determining identity in the decision-making module, the system integrates three types of information indicators: current recognition similarity E, driver's historical success rate F, and image recognizability score G. A weighted model is used to calculate the total decision score S to determine whether to allow passage. The formula S = w1×E + w2×F + w3×G is the core calculation model of this module, where w1, w2, and w3 are preset weight coefficients, and their sum is always equal to 1, representing the importance of each factor in the overall judgment. The similarity score E is generated by the comparison module based on the matching results between the current image and identity templates in the database, typically based on a confidence value of 0-1 converted from the distance between feature vectors. The historical success rate F is calculated by statistically analyzing the percentage of successful recognitions in the driver's past passage records, reflecting the credibility of the driver's identity. The image recognizability score G is calculated by combining the current image acquisition time (e.g., daytime or nighttime) and parameters such as the simultaneously acquired light intensity and image contrast, reflecting the degree of environmental influence on image recognition. The decision module performs score fusion by setting reasonable weights (e.g., w1=0.5, w2=0.3, w3=0.2) and compares the results with preset thresholds. When the S value is higher than the threshold, the system recognizes the identity as trustworthy and automatically triggers a passage signal; if the S value is lower than the threshold, an alarm process is triggered and the system switches to manual processing. This model implements a clear and responsive passage control logic.

[0069] By introducing a weighted scoring model, this decision mechanism avoids over-reliance on a single factor (such as similarity), improving the reliability of the system's judgments in complex and non-ideal environments. The formula structure has good adjustability; the weights can be adjusted based on historical data and actual scenarios, flexibly adapting to different recognition environments and security strategies. The system can achieve refined management of the judgment strategy by controlling the weight coefficients and thresholds, effectively reducing the false recognition rate and false negative rate, and improving the accuracy and rationality of the judgments. It is particularly suitable for logistics weighing scenarios requiring strong security guarantees.

[0070] The storage module scans the access frequency of all driver identities in the database every day, calculates the frequency of each driver identity, and calculates the identity data distribution entropy value as an indicator of identity diversity distribution. If the identity data distribution entropy value decreases, the abnormal data cleaning mechanism is activated; if the identity data distribution entropy value increases, the index depth and cache space are automatically increased.

[0071] This implementation introduces an information entropy model to dynamically monitor and optimize the distribution of identity data in the storage module. The storage module is configured with a daily scheduled task to scan the call frequency of all driver identity records in the database. By statistically analyzing the number of times each identity is accessed within a certain time window and standardizing the access frequency of all identities, the probability distribution p(i) of each identity is calculated. The system further uses the information entropy formula H = -Σp(i)log(p(i)) to calculate the overall identity distribution entropy value H, which measures the diversity and distribution balance of identity calls. This entropy value, serving as an indicator of the dynamic structural state of identity data, has the following application logic: When the entropy value continuously decreases, it indicates that some identities are appearing frequently, potentially indicating abnormal concentrated access phenomena such as malicious face scanning, identity theft, or operational errors. At this time, the system automatically triggers an abnormal data cleanup mechanism, marking, isolating, or transferring low-frequency or abnormal behavior records to a manual review process to ensure the health and security of the database data structure. When the entropy value increases, it indicates that the distribution of identity access is becoming more uniform, and the system judges it as a data growth state. To ensure access performance, it automatically increases the database index depth (such as enabling multidimensional hash indexes, prefix tree indexes, etc.) and cache space allocation, thereby improving data retrieval speed and concurrent processing capabilities. This mechanism uses information entropy as the core driving factor to achieve self-awareness of data state and dynamic adjustment of resource allocation, ensuring that the database can maintain high performance and data structure stability even under high-concurrency access.

[0072] By introducing the identity data distribution entropy value as a core indicator for database management, the system achieves quantitative perception and decision-making basis for the data structure status. Compared with traditional static cleaning or mean judgment mechanisms, the information entropy model can reflect more complex distribution characteristics and anomaly concentration phenomena, and its response is more sensitive and reliable. The system can proactively clean up potentially abnormal data when identity access is concentrated, avoiding redundant expansion and database pollution; at the same time, it automatically adjusts caching and indexing strategies when identity access expands in a balanced manner, improving response speed and scalability. The overall mechanism enhances the system's data security, structural controllability, and operational efficiency.

[0073] The specific method for calculating the entropy value of identity data distribution includes calculating the relative frequency of each registered independent identity, multiplying each frequency by the logarithm of the frequency, summing the products and taking the opposite value, and finally obtaining the value as the entropy value of identity data distribution.

[0074] In the daily scheduled data monitoring task executed by the storage module, the system scans all unique driver IDs registered in the database and counts the number of times each ID is accessed within a specified time window (e.g., 24 hours). Then, the access count for each ID is divided by the total access count to calculate its relative frequency p(i), forming a complete ID access frequency distribution. The system uses the information entropy calculation formula H = -Σp(i)log(p(i)) to calculate the ID data distribution entropy value. Here, p(i) is the relative frequency of the i-th ID, and the log function is generally base 2 (i.e., bit-based) or the natural logarithm ln, chosen according to the system implementation. For each ID, the system calculates the value of p(i) × log(p(i)), sums all results, and takes the negative of the sum; this is the ID data distribution entropy H for that time period.

[0075] A higher entropy value indicates a more even distribution of identity access; conversely, a lower entropy value indicates that some identities appear frequently and their concentration is increased. The system uses this entropy value to determine the diversity of the identity data structure and drives cleanup or expansion mechanisms. This entropy value provides a scientific computational basis for database structure health assessment and dynamic response.

[0076] This calculation method has a clear mathematical principle, is easy to operate, and has good feasibility and scalability. By statistically analyzing relative frequencies and calculating the sum of the logarithms of their products, the system not only monitors the distribution of identity access behavior but also quantifies its structural diversity trends, providing a scientific basis for subsequent system cleanup, optimization, and index adjustment. Compared to traditional mean or counting threshold judgments, information entropy is more sensitive and accurate in handling distributed anomaly identification, making it particularly suitable for handling large-scale, dynamically changing identity data scenarios.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A logistics weighing safety control system based on face recognition, characterized in that, The system comprises: The acquisition module is used for acquiring images of the vehicle and the driver when the logistics vehicle is weighed, and sending the processed images to the identification module; The identification module receives the images sent by the acquisition module, identifies the face image and the vehicle image therefrom, extracts the facial feature information and the vehicle information, and sends the facial feature information and the vehicle information to the comparison module, including acquiring three consecutive images, calculating the similarity of the face features between the consecutive images respectively, obtaining the standard deviation and taking the reciprocal of the standard deviation as the face feature recognition stability, and similarly processing the vehicle image recognition result to obtain the vehicle image recognition stability, calculating the face and vehicle recognition stability ratio, deciding whether to preferentially use the face result or the vehicle recognition result, and merging the information and sending it to the comparison module; The comparison module queries the registration information of the vehicle and the driver face recorded in the storage module, processes the current vehicle information and the facial feature information when the current acquired vehicle and driver face information does not exist, and sends it to the storage module for storage, and when the current acquired vehicle and driver face information exists, compares the received vehicle and driver face information with the vehicle and driver face feature information corresponding to the already stored vehicle and driver face information, and then sends the comparison result to the decision module; The decision module judges the received comparison result, decides whether to allow the vehicle to pass according to the actual situation, and when the decision is not passed, transfers to manual processing and returns to the acquisition module; Specifically, the decision module obtains the similarity score returned by the comparison module, queries the historical passing records of the driver in the database, calculates the historical passing success rate of the current driver, evaluates the image recognizability score according to the current image acquisition time and light condition, sets the weight using experience, calculates the total decision score, and if the total decision score is greater than a preset threshold, automatically releases, otherwise refuses to pass and issues an alarm; The specific formula for calculating the total decision score is: S = w1xE + w2xF + w3xG; Wherein, S represents the total decision score, w1, w2 and w3 represent the weight, w1+w2+w3=1, E represents the similarity score, F represents the historical passing success rate of the current driver, and G represents the image recognizability score of the current image acquisition time and light condition; The storage module is used for storing the registration information of the vehicle and the driver face, and transmitting the vehicle information and the facial feature information received by the comparison module.

2. The logistics weighing safety control system based on face recognition according to claim 1, characterized in that: When the logistics vehicle is weighed, the acquisition module acquires the vehicle and personnel images, calls the image enhancement algorithm to estimate the recognition success rate improvement value after image quality processing, simultaneously measures the calculation resource consumption increased for improving the image quality, calculates the marginal utility of the image quality, and if the marginal utility of the image quality is greater than 1, executes advanced image enhancement and sends it to the identification module; otherwise, directly sends the original image to save resources.

3. The logistics weighing safety control system based on face recognition according to claim 2, characterized in that: The specific formula for calculating the marginal utility of the image quality is: A = B / C; Wherein, A represents the marginal utility of the image quality, B represents the recognition success rate improvement value after image quality processing, and C represents the calculation resource consumption increased for improving the image quality.

4. The logistics weighing safety control system based on face recognition according to claim 1, characterized in that: The specific formula of the face and vehicle recognition stability ratio is R=H / D. Wherein, R represents the face and vehicle recognition stability ratio, H represents the reciprocal of the standard deviation of the face similarity change in the continuous frame, and D represents the vehicle image recognition stability.

5. The logistics weighing safety control system based on face recognition according to claim 1, characterized in that: The comparison module monitors the total amount of current processing comparison requests and the number of completed comparisons at the current time, sets the maximum comparison capacity, calculates the current comparison efficiency growth trend, and if the current comparison efficiency growth trend is less than 0, switches to a lightweight comparison strategy, otherwise uses complete features for high-precision comparison, and sends the comparison result to the judgment module.

6. The logistics weighing safety control system based on face recognition according to claim 5, characterized in that: The specific formula for calculating the current comparison efficiency growth trend is: dV / dt=βV(1-M / C); Wherein, dV / dt represents the current comparison efficiency growth trend, β represents the comparison efficiency coefficient, V represents the number of completed comparisons at the current time, M represents the total amount of current processing comparison requests, and C represents the set maximum comparison capacity.

7. The logistics weighing safety control system based on face recognition according to claim 1, characterized in that: The storage module scans the access frequency of all driver identities in the database at a fixed time every day, calculates the frequency of each driver identity, calculates the identity data distribution entropy value as an identity diversity distribution index, if the identity data distribution entropy value decreases, an abnormal data cleaning mechanism is enabled, and if the identity data distribution entropy value increases, the index depth and cache space are automatically increased.

8. The logistics weighing safety control system based on face recognition according to claim 7, characterized in that: The specific method for calculating the identity data distribution entropy value includes calculating the relative frequency of each identity for all registered independent identities, then multiplying each frequency by the logarithm value of the frequency, then summing the product results and taking the inverse, and finally the resulting value is taken as the identity data distribution entropy value.

Images