Image recognition-based high-speed entrance overload vehicle precise dissuasion business management system

Abstract

The application discloses a high-speed entrance overload vehicle precision dissuasion management system based on image recognition and relates to the field of traffic overload control.The system comprises an image acquisition module, a priori database and the like.The image acquisition module is used for collecting vehicle tire image data, preprocessing the vehicle tire image data and outputting tire size parameters.The a priori database is used for storing priori parameters for judging vehicle overload.The application can accurately extract tire size parameters through multi-view image acquisition and optimization processing, can realize overload judgment by combining priori data comparison, can effectively improve the automation and accuracy of high-speed entrance overload detection, can calibrate errors through a verification mechanism, can reduce misjudgment, and can further improve management efficiency through an electronic signature and information interaction mechanism to standardize the dissuasion process.

Description

Technical Field

[0001] This invention relates to the field of traffic control technology, specifically to a business management system for accurately dissuading overloaded vehicles from entering highways based on image recognition. Background Technology

[0002] Highway overloading control is a management measure to address vehicle overloading. Through weighing and enforcement inspections, it aims to curb overloading, protect highway facilities, ensure traffic safety, maintain transportation order, and reduce traffic accidents and road damage caused by overloading.

[0003] Patent application number 202510232182.3 discloses an intelligent detection and early warning method for a traffic overload control system, comprising: deploying a sensor group in a pre-test area and collecting vehicle data to establish a vehicle dataset; activating load sensors in the test area and configuring the test frequency of the load sensors using vehicle speed data in the vehicle dataset; establishing a load dataset using the configured load sensors, the load dataset having a time stamp; acquiring speed measurement data in the test area based on the time stamp, the speed measurement data including real-time speed data and acceleration data; and inputting the speed measurement data and the load dataset into... The system inputs the load compensation model and establishes load measurement results; it retrieves vehicle information based on license plate data, establishes a first warning based on the vehicle information retrieval results and the load measurement results, performs image recognition on the vehicle loading image, performs linkage analysis based on the image recognition results and the load measurement results, establishes a second warning, and intelligently detects and reports anomalies based on the first and second warnings. This application aims to solve the problem that "in the existing technology, fixed weighbridge detection stations have low detection efficiency, limited coverage, and lack effective monitoring of dynamic changes in vehicle loading status, resulting in overloading behavior not being identified in a timely, comprehensive, and accurate manner, further affecting the efficiency and safety of road traffic management."

[0004] However, in the current highway overload control scenario, vehicle overloading is mostly managed by dynamic and static weighing, and the existing technologies that apply image recognition technology to the detection of vehicle overloading are still in the minority.

[0005] To address this, we propose an image recognition-based business management system for accurately dissuading overloaded vehicles from entering highways. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a business management system for accurately dissuading overloaded vehicles from entering highways based on image recognition, which can effectively solve the problems of the existing technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0008] This invention discloses a business management system for accurately dissuading overloaded vehicles from entering highways based on image recognition, comprising:

[0009] The system comprises the following modules: an image acquisition module for acquiring vehicle tire image data, preprocessing the data to output tire size parameters; a priori database for storing prior parameters for determining vehicle overload; a selection and comparison module for selecting appropriate prior parameters from the database to match the tire size parameters output by the image acquisition module, and comparing the tire size parameters with the prior parameters to determine if the vehicle from which the tire size parameters originated is overloaded; a refresh module for re-entering the image acquisition module's operation phase, using a new vehicle as the acquisition target, and re-initializing the system; a confirmation module for providing electronic signature permissions to the vehicle driver; a verification module for verifying the vehicle's status and estimating the tire size parameter error, calibrating the original tire size parameters output by the image acquisition module based on the error, and updating the tire size parameters output by the image acquisition module; and an interaction module for receiving the operation results from the selection and comparison module and the confirmation module, generating electronic messages, and transmitting them to a preset receiving end.

[0010] Furthermore, the image acquisition module is integrated with several sets of high-definition industrial cameras. The several sets of high-definition industrial cameras operate synchronously to perform vehicle tire image data acquisition operations. The vehicle tire image data acquisition angle of the several sets of high-definition industrial cameras is the side and front of the vehicle tire.

[0011] The vehicle tire image data includes vehicle tire side image data and vehicle tire front image data. The vehicle tire front image data points to the tire surface that is in contact with the road surface. The preprocessing operations performed on the vehicle tire image data include: image optimization, image background segmentation, and tire contour width recognition in the image.

[0012] Among them, the tire profile width recognition result in the image is also the tire size parameter.

[0013] Furthermore, the image optimization and image background segmentation objects are vehicle tire side image data, and the image optimization, image background segmentation, and tire contour width recognition objects are vehicle tire front image data.

[0014] The vehicle tire side image data includes a global image of the vehicle tire side;

[0015] The image optimization operation is represented as follows:

[0016] ;

[0017] In the formula: For optimized pixel values; Here, 'c' represents the original pixel value, and 'c' represents the color channel. For enhancement coefficient; The Laplace gradient; The Laplace gradient adjustment parameter; This is the spatial attenuation coefficient; The Euclidean distance from the pixel to the center of the image; For color channel adaptive coefficients;

[0018] Specifically, the above formula is used to process each pixel in the vehicle tire sidewall image data and the vehicle tire frontal image data to output optimized vehicle tire sidewall image data and vehicle tire frontal image data. This represents the sigmoid cutoff function. The output value is set to 64 when x=0 and 210 when x=255.

[0019] Furthermore, the image background segmentation operation for the vehicle tire frontal image data and the vehicle tire side image data is as follows:

[0020] The optimized vehicle tire sidewall image data and vehicle tire front image data are converted into grayscale images. A grayscale value range for the vehicle tire is set. Based on the set grayscale value range, the image data is traversed. Pixels in the image data that do not conform to the grayscale value range of the vehicle tire are deleted as deletion targets. The local image composed of the remaining pixels is denoted as the vehicle tire front image and the vehicle tire sidewall image.

[0021] The tire profile width recognition operation in the vehicle tire frontal image data is as follows:

[0022] Pick the most distant, directly opposite pixel pairs on the left and right sides of the frontal image of the vehicle tire, identify the distance between the pixel pairs, and determine the actual distance between the pixel pairs, i.e., the tire profile width, based on the ratio of the size of the frontal image of the vehicle tire to the actual size of the corresponding area on the vehicle tire.

[0023] Furthermore, during the stage of acquiring vehicle tire image data, the image acquisition module simultaneously identifies the vehicle license plate number and obtains vehicle specification information based on the vehicle license plate number;

[0024] The prior database stores vehicle tire size parameters when a vehicle is overloaded, corresponding to various vehicle specification information. Each vehicle tire size parameter when overloaded is marked with vehicle specification information and corresponding tire pressure. The marked vehicle tire size parameters when overloaded are the prior parameters.

[0025] Furthermore, during the operation of the selection and comparison module, when selecting prior parameters from the prior database, the module refers to the vehicle specification information corresponding to the tire size parameters output by the image acquisition module. Based on the vehicle specification information, it selects prior parameters marked with the same vehicle specification information from the prior database, and then performs the operation to determine whether the vehicle is overloaded.

[0026] ;

[0027] In the formula: The tire size parameters output by the image acquisition module; Prior parameters selected for the operation of the selection and comparison module; To determine the threshold;

[0028] If the above formula is true, the vehicle is determined to be overloaded; otherwise, the vehicle is determined not to be overloaded. When the vehicle is determined not to be overloaded, the refresh module is triggered to run; when the vehicle is determined to be overloaded, the confirmation module is triggered to run.

[0029] Furthermore, the confirmation module is integrated into a mobile computer device, which stores a preset electronic overload driving confirmation certificate. When the confirmation module triggers the operation phase, the vehicle driver obtains electronic signature permissions.

[0030] When a vehicle driver performs an electronic signature operation, it indicates that the driver acknowledges the overloading behavior, accepts the advice to turn back at the highway exit, and drives towards the preset destination. Simultaneously, the system jumps to the refresh module to run.

[0031] When a vehicle driver refuses to perform the electronic signature operation, a redirect is triggered, leading to the verification module's operation phase, where the system continues to run.

[0032] Furthermore, the verification module is integrated with a tire pressure regulating device and a calibration unit. The tire pressure regulating device is used to sequentially connect each vehicle tire and perform tire pressure detection on each vehicle tire. When the tire pressure is detected to be inconsistent with the tire pressure marked by the prior parameter selected in the selection and comparison module, the tire pressure of the vehicle tire is adjusted to be consistent with the tire pressure marked by the prior parameter.

[0033] The calibration unit is used to receive the vehicle tire sidewall image after image optimization and image background segmentation, identify whether there are cracks in the vehicle tire sidewall image, set the calibration value to 0 when there are no cracks, and calculate the calibration value through the following logic when there are cracks, and then perform the calibration operation to update the output tire size parameters, and apply them again to the judgment in the selection and comparison module.

[0034] The calibration value is:

[0035] ;

[0036] In the formula: The area of ​​the image with the largest crack; This represents the total area of ​​the vehicle tire sidewall image; The shortest distance from the center point of the maximum crack image to the outer contour of the vehicle tire sidewall image and the shortest distance to the inner contour of the vehicle tire sidewall image; The outer contour radius and inner contour radius of the vehicle tire sidewall image;

[0037] in, , The number of pixels contained in the crack image and the vehicle tire sidewall image is used to determine the updated tire size parameters, which are represented as the original tire size parameters minus the calibration value.

[0038] Furthermore, when identifying cracks in the vehicle tire sidewall image, the system user manually marks them on the tire sidewall image, or follows the following rules:

[0039] Manually set the crack detection threshold And a recognition window, which slides across the vehicle tire sidewall image to calculate the crack discrimination index under each window. The crack discrimination index and crack judgment threshold under each window are compared. Compare and obtain the crack detection threshold (greater than or equal to). The crack discrimination index points to the window, and then the crack image is obtained from the window by a threshold-based segmentation method;

[0040] ;

[0041] In the formula: The crack discrimination index is the image of the region within the recognition window after the nth slide; , , , These are the weighting coefficients; To identify the mean grayscale difference between the candidate crack area and the background in the image within the window; To identify the rate of change of edge gradients detected in the image within the window region; To identify the texture disorder of an image region within a window; To identify the pixel connected component length of the candidate region of the image within the window; To identify the standard deviation of pixel brightness in the image region within the window; To identify the difference between the average gradient magnitude of the region image within the window and the surrounding 20-pixel neighborhood, and to normalize this value to the range [0,1];

[0042] Among them, the weighting coefficient , , , All are positive numbers, and their sum is 1, and they follow the rules of... > > > The candidate crack area in the image within the recognition window is the central region of the image within the recognition window, and the area of ​​the central region is half of the area of ​​the image within the recognition window.

[0043] The Euclidean distance between the tire surface texture feature vector and the preset standard texture feature vector within the region is calculated and then normalized.

[0044] Furthermore, the image acquisition module is interactively connected to the prior database and the refresh module via a wireless network. The prior database is interactively connected to the selection and comparison module via a wireless network. The selection and comparison module is interactively connected to the refresh module and the confirmation module via a wireless network. The confirmation module is interactively connected to the refresh module and the verification module via a wireless network. The verification module is connected to the interaction module via a wireless network.

[0045] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0046] This invention provides a management system for accurately dissuading overloaded vehicles from entering highways based on image recognition. During operation, the system accurately extracts tire size parameters through multi-view image acquisition and optimization processing, and determines overload by comparing with prior data. This effectively improves the automation and accuracy of overload detection at highway entrances. Furthermore, it calibrates errors through a verification mechanism to reduce misjudgments, and standardizes the dissuasion process through electronic signatures and information interaction mechanisms to further improve efficiency. This system avoids the errors of manual intervention in traditional detection and ensures continuous adaptation to different vehicles through dynamic refreshing and calibration, effectively guaranteeing highway traffic safety and promoting the intelligent, refined, and standardized development of overload dissuasion services. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0048] Figure 1 A schematic diagram of the structure of the image recognition-based highway entrance overloaded vehicle precise dissuasion and return business management system;

[0049] Figure 2This is a schematic diagram illustrating an example of the acquisition perspective used by the image acquisition module in this invention when acquiring vehicle tire image data;

[0050] Figure 3 The parameters used in the calibration value acquisition stage of this invention The diagram shows the instructions. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0052] The present invention will be further described below with reference to embodiments.

[0053] Example:

[0054] This embodiment describes a business management system for accurately dissuading overloaded vehicles from entering highways based on image recognition. Figure 1 As shown, it includes:

[0055] The image acquisition module is used to acquire vehicle tire image data and preprocess the vehicle tire image data to output tire size parameters.

[0056] The image acquisition module is integrated with several sets of high-definition industrial cameras. These sets of high-definition industrial cameras operate synchronously to acquire vehicle tire image data. The vehicle tire image data acquisition angles of these sets of high-definition industrial cameras are the side and front of the vehicle tires.

[0057] Vehicle tire image data includes vehicle tire side image data and vehicle tire front image data. The vehicle tire front image data points to the tire surface that is in contact with the road surface. The preprocessing operations performed on the vehicle tire image data include: image optimization, image background segmentation, and tire contour width recognition in the image.

[0058] Among them, the tire profile width recognition result in the image is also the tire size parameter;

[0059] Image optimization and image background segmentation target vehicle tire side image data, while image optimization, image background segmentation, and tire contour width recognition target vehicle tire front image data.

[0060] The vehicle tire sidewall image data contains a global image of the vehicle tire sidewall.

[0061] Image optimization operations are represented as:

[0062] ;

[0063] In the formula: For optimized pixel values; Here, 'c' represents the original pixel value, and 'c' represents the color channel. For enhancement coefficient; The Laplace gradient; The Laplace gradient adjustment parameter; This is the spatial attenuation coefficient; The Euclidean distance from the pixel to the center of the image; For color channel adaptive coefficients;

[0064] Specifically, the above formula is used to process each pixel in the vehicle tire sidewall image data and the vehicle tire frontal image data to output optimized vehicle tire sidewall image data and vehicle tire frontal image data. This represents the sigmoid cutoff function. The output value is set to 64 when x=0 and 210 when x=255.

[0065] It should be noted that:

[0066] Enhancement coefficient:

[0067] Dynamic adjustment logic: Adaptively selects values ​​based on the overall average brightness of the image: When the overall average brightness of the image is <80 (dark image, such as nighttime shooting), the enhancement coefficient is 3.0-5.0; when 80 ≤ overall average brightness of the image ≤180 (normal brightness), the enhancement coefficient is 1.5-3.0; when the overall average brightness of the image is >180 (overly bright image, such as midday strong light), α is 0.5-1.5.

[0068] Laplace gradient:

[0069] An improved 3×3 Laplacian operator [0101-41010] is adopted, which focuses more on the gradient response in the vertical / horizontal direction compared with the standard operator, and matches the main direction of the tire tread (mostly the lateral direction);

[0070] Laplace gradient tuning parameters:

[0071] Based on the typical width of tire tread (10-20 pixels), a large number of samples were statistically determined to be λ=10. At this value, the enhancement effect on the edge of the tread is the best for tread with a width of ≥5 pixels, and it has a natural suppression effect on noise (such as road gravel reflection) with a width of less than 3 pixels.

[0072] Spatial attenuation coefficient:

[0073] Joint setting based on the aspect ratio r of the tire detection frame and the image resolution Res: When r ≤ 1.2 and Res ≥ 50, take 0.01 - 0.03; when 1.2 < r ≤ 2.0 and 30 ≤ Res < 50, take 0.03 - 0.06; when r > 2.0 and Res < 30, take 0.06 - 0.09. When the tires overlap, the spatial attenuation coefficient of the overlapping area is temporarily increased by 0.02. Its magnitude determines the steepness of the weight attenuation, which can match the feature distributions under different tire morphologies and shooting resolutions, reducing the interference of redundant information;

[0074] Color channel adaptive coefficient:

[0075] c = 1, 2, 3 correspond to the red, green, and blue channels respectively. There is no specific calculation formula, which is set according to the color feature differences between the tire and the background. The values are: γ(1) = 1.2 for the red channel, γ(2) = 0.8 for the green channel, and γ(3) = 1.0 for the blue channel. When rain is detected in the image (the blue channel value is abnormally high), (3) is automatically reduced to 0.6.

[0076] The above formula breaks through the traditional linear combination framework through the cross - domain fusion of physical fields (sine functions), spatial topology (attenuation factors), and biological vision (sigmoid responses), and can adaptively mine the hidden features of the tire such as "alternating light and dark patterns - edge gradient mutations - color channel deviations", so as to achieve the purpose of image optimization; [[ID=!]]

[0077] The image background segmentation operation for the vehicle tire front - view image data and the vehicle tire side - view image data is as follows:

[0078] Convert the optimized vehicle tire side - view image data and vehicle tire front - view image data into grayscale images, set the vehicle tire grayscale value range, traverse the image data based on the set vehicle tire grayscale value range, and perform a deletion operation on the pixels in the image data that do not conform to the vehicle tire grayscale value range. The local image composed of the remaining pixels is recorded as the vehicle tire front - view image and the vehicle tire side - view image;

[0079] The operation of identifying the tire contour width in the vehicle tire front - view image data is as follows:

[0080] Pick up the pixel pairs that are directly opposite and farthest apart on the left and right sides of the vehicle tire front - view image, identify the distance between the pixel pairs, and determine the actual spacing distance of the pixel pairs based on the ratio of the size of the vehicle tire front - view image to the actual size of the corresponding area on the vehicle tire, that is, the tire contour width;

[0081] The prior database is used to store the prior parameters for determining vehicle overloading;

[0082] During the stage of collecting vehicle tire image data, the image acquisition module synchronously identifies the vehicle license plate number and obtains the vehicle specification information based on the vehicle license plate number;

[0083] The prior database stores the tire size parameters of vehicles when they are overloaded, and each tire size parameter of an overloaded vehicle is marked with the vehicle specification information and the corresponding tire pressure. The marked tire size parameters of an overloaded vehicle are the prior parameters.

[0084] The selection and comparison module is used to select prior parameters from the prior database that are suitable for the tire size parameters output by the image acquisition module, and to determine whether the vehicle from which the tire size parameters are derived is overloaded based on the comparison between the tire size parameters and the prior parameters.

[0085] During the selection and comparison module's operation phase, when selecting prior parameters from the prior database, the system refers to the vehicle specification information corresponding to the tire size parameters output by the image acquisition module. Based on the vehicle specification information, prior parameters marked with the same vehicle specification information are selected from the prior database, and then the operation to determine whether the vehicle is overloaded is performed.

[0086] ;

[0087] In the formula: The tire size parameters output by the image acquisition module; Prior parameters selected for the operation of the selection and comparison module; To determine the threshold;

[0088] If the above formula is true, the vehicle is determined to be overloaded; otherwise, the vehicle is determined not to be overloaded. When the vehicle is determined not to be overloaded, the refresh module is triggered to run; when the vehicle is determined to be overloaded, the confirmation module is triggered to run.

[0089] The refresh module is used to jump to the image acquisition module's running phase, using a new vehicle as the acquisition target, and the control system initializes and runs again.

[0090] The confirmation module is used to provide electronic signature permissions to vehicle drivers.

[0091] The confirmation module is integrated into the mobile computer device, which stores a preset electronic overload driving confirmation certificate. When the confirmation module is triggered, the vehicle driver obtains electronic signature permissions.

[0092] When a vehicle driver performs an electronic signature operation, it indicates that the driver acknowledges the overloading behavior, accepts the advice to turn back at the highway exit, and drives towards the preset destination. Simultaneously, the system jumps to the refresh module to run.

[0093] When a vehicle driver refuses to perform the electronic signature operation, a redirect is triggered, leading to the verification module's operation phase, where the system continues to run.

[0094] The verification module is used to verify the vehicle status and estimate the error of the vehicle's tire size parameters. Based on the error, the tire size parameters output by the original image acquisition module are calibrated to update the tire size parameters output by the image acquisition module.

[0095] The verification module is integrated with a tire pressure regulating device and a calibration unit. The tire pressure regulating device is used to connect the tires of each vehicle in sequence and perform tire pressure detection on each vehicle tire. When the tire pressure is detected to be inconsistent with the tire pressure marked by the prior parameter in the selection and comparison module, the tire pressure of the vehicle tire is adjusted to be consistent with the tire pressure marked by the prior parameter.

[0096] The calibration unit is used to receive the vehicle tire sidewall image after image optimization and image background segmentation, identify whether there are cracks in the vehicle tire sidewall image, set the calibration value to 0 when there are no cracks, and calculate the calibration value through the following logic when there are cracks, and then perform the calibration operation to update the output tire size parameters, and apply them again to the judgment in the selection and comparison module.

[0097] The calibration value is:

[0098] ;

[0099] In the formula: The area of ​​the image with the largest crack; This represents the total area of ​​the vehicle tire sidewall image; The shortest distance from the center point of the maximum crack image to the outer contour of the vehicle tire sidewall image and the shortest distance to the inner contour of the vehicle tire sidewall image; The outer contour radius and inner contour radius of the vehicle tire sidewall image;

[0100] in, , The number of pixels contained in the crack image and the vehicle tire sidewall image is used to determine the updated tire size parameters, which are represented by the original tire size parameters minus the calibration value.

[0101] The above formula quantifies the impact of cracks on tire size parameters based on the area ratio of the largest crack in the tire sidewall image (the ratio of the area of ​​the largest crack image to the total area of ​​the tire sidewall image), the shortest distance from the crack center point to the inner and outer contours of the tire, and the radii of the inner and outer contours of the tire. It then generates calibration values ​​to correct the original tire size parameters, thereby offsetting the measurement errors caused by cracks. This effectively transforms the physical defect of tire cracks into a quantifiable calibration indicator, breaking through the traditional static parameter output mode that relies solely on image recognition results. By modeling the correlation between crack features and size errors, it dynamically optimizes tire size parameters and improves the accuracy of the underlying data for overload determination.

[0102] When identifying cracks in vehicle tire sidewall images, the system user manually marks them on the tire sidewall image, or follows the following rules:

[0103] Manually set the crack detection threshold And a recognition window, which slides across the vehicle tire sidewall image to calculate the crack discrimination index under each window. The crack discrimination index and crack judgment threshold under each window are compared. Compare and obtain the crack detection threshold (greater than or equal to). The crack discrimination index points to the window, and then the crack image is obtained from the window by a threshold-based segmentation method;

[0104] ;

[0105] In the formula: The crack discrimination index is the image of the region within the recognition window after the nth slide; , , , These are the weighting coefficients; To identify the mean grayscale difference between the candidate crack area and the background in the image within the window; To identify the rate of change of edge gradients detected in the image within the window region; To identify the texture disorder of an image region within a window; To identify the pixel connected component length of the candidate region of the image within the window; To identify the standard deviation of pixel brightness in the image region within the window; To identify the difference between the average gradient magnitude of the region image within the window and the surrounding 20-pixel neighborhood, and to normalize this value to the range [0,1];

[0106] Among them, the weighting coefficient , , , All are positive numbers, and their sum is 1, and they follow the rules of... > > > The candidate crack area in the image within the recognition window is the central region of the image within the recognition window, and the area of ​​the central region is half of the area of ​​the image within the recognition window.

[0107] The result is obtained by calculating the Euclidean distance between the tire surface texture feature vector and the preset standard texture feature vector within the region, and then normalizing the Euclidean distance.

[0108] The above formula integrates multi-dimensional features such as the mean gray-level difference, edge gradient change rate, texture disorder, and pixel connected component length within the recognition window, and combines them with weight coefficients to generate a crack discrimination index, thereby achieving accurate identification of tire sidewall cracks. By assigning weights, it highlights key features such as gray-level difference and texture disorder, while weakening the influence of secondary features. This solves the problem that crack identification based on single features is easily affected by stains and lighting interference, and significantly improves the accuracy and anti-interference ability of tire crack identification.

[0109] The interaction module is used to receive the operation results of the selection and comparison module and the confirmation module, generate electronic messages, and transmit them to the preset receiving end.

[0110] The image acquisition module interacts with the prior database and the refresh module via a wireless network. The prior database interacts with the selection and comparison module via a wireless network. The selection and comparison module interacts with the refresh module and the confirmation module via a wireless network. The confirmation module interacts with the refresh module and the verification module via a wireless network. The verification module connects with the interaction module via a wireless network.

[0111] In this embodiment, the image acquisition module collects vehicle tire image data, preprocesses the data to output tire size parameters, and stores prior parameters for determining vehicle overload in a priori database. The selection and comparison module further selects prior parameters from the priori database that are compatible with the tire size parameters output by the image acquisition module. Based on the comparison between the tire size parameters and the prior parameters, it determines whether the vehicle from which the tire size parameters originate is overloaded. Then, the refresh module jumps back to the image acquisition module's operation phase, using a new vehicle as the acquisition target. The control system re-initializes and runs. The confirmation module provides electronic signature permissions for the vehicle driver during post-processing and verifies the vehicle status and estimates the tire size parameter error through the verification module. Based on the error, the tire size parameters output by the original image acquisition module are calibrated to update the tire size parameters output by the image acquisition module. Finally, the interaction module receives the operation results from the selection and comparison module and the confirmation module, generates an electronic message, and transmits it to a preset receiving end.

[0112] The aforementioned system can efficiently determine overloading by accurately collecting and analyzing vehicle tire images and combining them with preset parameters, reducing manual intervention. The verification mechanism can calibrate data errors and improve judgment accuracy. The driver's electronic signature confirmation or objection verification process ensures standardized handling, and the interactive transmission function enables efficient information flow. This allows for the precise dissuasion of overloaded vehicles, reducing highway safety hazards, improving entrance efficiency, standardizing enforcement procedures, balancing management rigor with operational convenience, and contributing to the intelligentization of highway traffic management.

[0113] See Figure 2As shown in the diagram, the arrows further illustrate the perspective of the high-definition industrial camera in the image acquisition module when acquiring vehicle tire image data.

[0114] See Figure 3 As shown, the figure represents the sidewall image of the vehicle tire from a planar perspective using two circles. It also shows the shortest distance from the center point of the largest crack image (the area traversed by the dashed line in the figure) to the outer contour of the sidewall image of the vehicle tire and the shortest distance to the inner contour of the sidewall image of the vehicle tire based on the dashed line. The center of the sidewall image of the vehicle tire (i.e., the area defined by the two circles in the figure) should be above the extension line of the dashed line in the figure.

[0115] The following is an application example of the system described in the above embodiments:

[0116] At 10:00 a.m. on Wednesday, a silver-gray medium-sized truck slowly drove into the entrance of the C expressway, triggering the operation of the image recognition-based expressway entrance overloaded vehicle precise dissuasion and return business management system.

[0117] The image acquisition module responded immediately, with two sets of high-definition industrial cameras capturing images of the truck's four tires: one set of cameras faced the side of the tires, acquiring a global side view of each tire; the other set of cameras faced the front of the tires (the tread in contact with the road surface), acquiring a frontal image. After acquisition, the system preprocessed the images: for the side view images, image optimization (enhancing tire texture clarity) and background segmentation (removing background pixels with abnormal grayscale values) preserved the complete tire side profile; for the frontal images, after optimization and segmentation, the system identified the profile width—picking the farthest pixel pairs on the left and right sides of the frontal image and converting them at a 1:500 ratio to the actual size, yielding the size parameters of the four tires (average width of 0.22 meters). Simultaneously, the system obtained vehicle specification information ("medium-sized truck, rated load capacity 10 tons") through license plate recognition.

[0118] The selection and comparison module is activated, and the corresponding prior parameters are matched from the prior database based on the vehicle specifications: the standard tire width for this type of truck when overloaded is 0.20 meters (corresponding to a tire pressure of 7 bar). The system calculates according to the judgment rules: the current tire width (0.22 meters) is greater than the product of the prior parameter (0.20 meters) and the judgment threshold (1.05) (0.20 × 1.05 = 0.21 meters), and the vehicle is judged to be overloaded.

[0119] As a result, the confirmation module was triggered, and the toll station staff generated an electronic "Overload Dissuasion Confirmation Form" on their handheld terminal. After reviewing it, the driver believed that the judgment was wrong and explicitly refused to sign the electronic form.

[0120] The system then switched to the calibration module: The operator used the tire pressure regulator to check the tire and found the current tire pressure to be 6.8 bar, slightly lower than the pre-defined parameter of 7 bar. The tire pressure was then adjusted to 7 bar. Next, the calibration unit performed crack detection on the tire sidewall image: by calculating the crack discrimination index for each area using a sliding recognition window (combining grayscale difference, edge changes, and other indicators), a minor crack was found on the sidewall of the left front tire. Based on the crack area (accounting for 1.5% of the total sidewall image area), the shortest distance from the crack center point to the inner and outer contours (3cm and 6cm respectively), and the tire's inner and outer contour radii (25cm and 35cm respectively), the system calculated a calibration value of 0.001 meters. The original tire width parameter (0.22 meters) was calibrated to 0.219 meters.

[0121] The calibrated data was resubmitted to the selection and comparison module for re-comparison: 0.219 meters > 0.21 meters, still indicating overloading. Staff showed the driver the calibrated test report, the driver accepted the result, and finally signed the electronic confirmation form, confirming acceptance of the turnaround and leaving the highway entrance.

[0122] The interaction module integrates the judgment result, verification process, and driver signature information into an electronic message and sends it to the highway management center. Simultaneously, the refresh module activates, resetting the system to its initial state, awaiting the next vehicle to enter the inspection area, thus completing a full cycle.

[0123] In summary, the system in the above embodiments accurately extracts tire size parameters through multi-view image acquisition and optimization processing, and achieves overload determination by comparing with prior data. This effectively improves the automation and accuracy of overload detection at highway entrances. Furthermore, it calibrates errors through a verification mechanism to reduce misjudgments, and standardizes the dissuasion process through electronic signatures and information interaction mechanisms, further improving efficiency. This not only avoids the errors of manual intervention in traditional detection, but also ensures continuous adaptation to different vehicles through dynamic refreshing and calibration, effectively guaranteeing highway traffic safety and promoting the intelligent, refined, and standardized development of overload dissuasion services.

[0124] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A highway entrance overloaded vehicle precise dissuasion and return business management system based on image recognition, characterized in that, include: The image acquisition module is used to acquire vehicle tire image data and preprocess the vehicle tire image data to output tire size parameters. The image acquisition module is integrated with several sets of high-definition industrial cameras. The sets of high-definition industrial cameras operate synchronously to perform vehicle tire image data acquisition operations. The vehicle tire image data acquisition angles of the sets of high-definition industrial cameras are the side and front of the vehicle tires. The vehicle tire image data includes vehicle tire side image data and vehicle tire front image data. The vehicle tire front image data points to the tire surface that is in contact with the road surface. The preprocessing operations performed on the vehicle tire image data include: image optimization, image background segmentation, and tire contour width recognition in the image. Among them, the tire profile width recognition result in the image is also the tire size parameter; The image optimization and image background segmentation objects are vehicle tire side image data, and the image optimization, image background segmentation, and tire contour width recognition objects are vehicle tire front image data. The vehicle tire side image data includes a global image of the vehicle tire side; The image optimization operation is represented as follows: ； In the formula: For optimized pixel values; Here, 'c' represents the original pixel value, and 'c' represents the color channel. For enhancement coefficient; The Laplace gradient; The Laplace gradient adjustment parameter; This is the spatial attenuation coefficient; The Euclidean distance from the pixel to the center of the image; For color channel adaptive coefficients; Specifically, the above formula is used to process each pixel in the vehicle tire sidewall image data and the vehicle tire frontal image data to output optimized vehicle tire sidewall image data and vehicle tire frontal image data. This represents the sigmoid cutoff function. The output value is set to 64 when x=0 and 210 when x=255. A priori database is used to store priori parameters for determining vehicle overloading; The selection and comparison module is used to select prior parameters from the prior database that are suitable for the tire size parameters output by the image acquisition module, and to determine whether the vehicle from which the tire size parameters are derived is overloaded based on the comparison between the tire size parameters and the prior parameters. The refresh module is used to jump to the image acquisition module's running phase, using a new vehicle as the acquisition target, and the control system initializes and runs again. The confirmation module is used to provide electronic signature permissions to vehicle drivers. The verification module is used to verify the vehicle status and estimate the error of the vehicle's tire size parameters. Based on the error, the tire size parameters output by the original image acquisition module are calibrated to update the tire size parameters output by the image acquisition module. The interaction module is used to receive the operation results of the selection and comparison module and the confirmation module, generate electronic messages, and transmit them to the preset receiving end.

2. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, The image background segmentation operation for the vehicle tire frontal image data and the vehicle tire side image data is as follows: The optimized vehicle tire sidewall image data and vehicle tire front image data are converted into grayscale images. A grayscale value range for the vehicle tire is set. Based on the set grayscale value range, the image data is traversed. Pixels in the image data that do not conform to the grayscale value range of the vehicle tire are deleted as deletion targets. The local image composed of the remaining pixels is denoted as the vehicle tire front image and the vehicle tire sidewall image. The tire profile width recognition operation in the vehicle tire frontal image data is as follows: Pick the most distant, directly opposite pixel pairs on the left and right sides of the frontal image of the vehicle tire, identify the distance between the pixel pairs, and determine the actual distance between the pixel pairs, i.e., the tire profile width, based on the ratio of the size of the frontal image of the vehicle tire to the actual size of the corresponding area on the vehicle tire.

3. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, During the stage of acquiring vehicle tire image data, the image acquisition module simultaneously identifies the vehicle license plate number and obtains vehicle specification information based on the vehicle license plate number. The prior database stores vehicle tire size parameters when a vehicle is overloaded, corresponding to various vehicle specification information. Each vehicle tire size parameter when overloaded is marked with vehicle specification information and corresponding tire pressure. The marked vehicle tire size parameters when overloaded are the prior parameters.

4. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, During the operation of the selection and comparison module, when selecting prior parameters from the prior database, the module refers to the vehicle specification information corresponding to the tire size parameters output by the image acquisition module. Based on the vehicle specification information, it selects prior parameters marked with the same vehicle specification information from the prior database, and then performs the operation to determine whether the vehicle is overloaded. ； In the formula: The tire size parameters output by the image acquisition module; Prior parameters selected for the operation of the selection and comparison module; To determine the threshold; If the above formula is true, the vehicle is determined to be overloaded; otherwise, the vehicle is determined not to be overloaded. When the vehicle is determined not to be overloaded, the refresh module is triggered to run; when the vehicle is determined to be overloaded, the confirmation module is triggered to run.

5. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, The confirmation module is integrated into a mobile computer device, which stores a preset electronic overload driving confirmation certificate. When the confirmation module is triggered to run, the vehicle driver obtains electronic signature permissions. When a vehicle driver performs an electronic signature operation, it indicates that the driver acknowledges the overloading behavior, accepts the advice to turn back at the highway exit, and drives towards the preset destination. Simultaneously, the system jumps to the refresh module to run. When a vehicle driver refuses to perform the electronic signature operation, a redirect is triggered, leading to the verification module's operation phase, where the system continues to run.

6. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, The verification module is integrated with a tire pressure regulating device and a calibration unit. The tire pressure regulating device is used to connect each vehicle tire in sequence and perform tire pressure detection on each vehicle tire. When the tire pressure is detected to be inconsistent with the tire pressure marked by the prior parameter selected in the selection and comparison module, the tire pressure of the vehicle tire is adjusted to be consistent with the tire pressure marked by the prior parameter. The calibration unit is used to receive the vehicle tire sidewall image after image optimization and image background segmentation, identify whether there are cracks in the vehicle tire sidewall image, set the calibration value to 0 when there are no cracks, and calculate the calibration value through the following logic when there are cracks, and then perform the calibration operation to update the output tire size parameters, and apply them again to the judgment in the selection and comparison module. The calibration value is: ； In the formula: The area of ​​the image with the largest crack; This represents the total area of ​​the vehicle tire sidewall image; The shortest distance from the center point of the maximum crack image to the outer contour of the vehicle tire sidewall image and the shortest distance to the inner contour of the vehicle tire sidewall image; The outer contour radius and inner contour radius of the vehicle tire sidewall image; in, , The number of pixels contained in the crack image and the vehicle tire sidewall image is used to determine the updated tire size parameters, which are represented as the original tire size parameters minus the calibration value.

7. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 6, characterized in that, When identifying cracks in the vehicle tire sidewall image, the system user manually marks them on the tire sidewall image, or follows the following rules: Manually set the crack detection threshold And a recognition window, which slides across the vehicle tire sidewall image to calculate the crack discrimination index under each window. The crack discrimination index and crack judgment threshold under each window are compared. Compare and obtain the crack detection threshold (greater than or equal to). The crack discrimination index points to the window, and then the crack image is obtained from the window by a threshold-based segmentation method; ； In the formula: The crack discrimination index is the image of the region within the recognition window after the nth slide; , , , These are the weighting coefficients; To identify the mean grayscale difference between the candidate crack area and the background in the image within the window; To identify the rate of change of edge gradients detected in the image within the window region; To identify the texture disorder of an image region within a window; To identify the pixel connected component length of the candidate region of the image within the window; To identify the standard deviation of pixel brightness in the image region within the window; To identify the difference between the average gradient magnitude of the region image within the window and the surrounding 20-pixel neighborhood, and to normalize this value to the range [0,1]; Among them, the weighting coefficient , , , All are positive numbers, and their sum is 1, and they follow the rules of... > > > The candidate crack area in the image within the recognition window is the central region of the image within the recognition window, and the area of ​​the central region is half of the area of ​​the image within the recognition window. The Euclidean distance between the tire surface texture feature vector and the preset standard texture feature vector within the region is calculated and then normalized.

8. The image recognition-based highway entrance overloaded vehicle precise dissuasion and return management system according to claim 1, characterized in that, The image acquisition module is interactively connected to the prior database and the refresh module via a wireless network. The prior database is interactively connected to the selection and comparison module via a wireless network. The selection and comparison module is interactively connected to the refresh module and the confirmation module via a wireless network. The confirmation module is interactively connected to the refresh module and the verification module via a wireless network. The verification module is connected to the interaction module via a wireless network.

Images