A high-speed green vehicle fee-avoiding identification method, system and medium

Abstract

The application provides a high-speed green-pass vehicle escape fee identification method, system and medium. The method comprises the following steps: obtaining vehicle pass record data of a green-pass lane, preprocessing the vehicle pass record data, obtaining optimized vehicle pass record data, constructing a green-pass vehicle escape fee identification rule set, performing green-pass vehicle escape fee identification, obtaining a suspected escape fee vehicle state, if the suspected escape fee vehicle state is obtained, analyzing and processing the suspected escape fee vehicle state by a preset escape fee confidence processing method, obtaining an escape fee confidence, and finally performing threshold comparison, if the escape fee confidence is less than a preset escape fee confidence threshold, activating manual review, if the escape fee confidence is greater than or equal to the preset escape fee confidence threshold, generating a green-pass vehicle escape fee identification report through a preset green-pass escape fee identification template. The application realizes accurate identification of high-speed green-pass vehicle escape fee by constructing image data analysis and core mining, performing multi-dimensional analysis on vehicle pass record data of a green-pass lane, and accurately identifying various escape fee behaviors.

Description

Technical Field

[0001] This application relates to the field of highway toll management technology, and more specifically, to a method, system, and medium for identifying toll evasion by green channel vehicles on highways. Background Technology

[0002] Currently, toll evasion by vehicles using the green channel mainly involves using fake truck registration certificates to transport goods within the green channel and evading tolls, or disguising non-green channel goods as green channel goods to evade tolls. Traditional manual inspection has many limitations, such as requiring highly skilled inspectors who are proficient in both inspection procedures and computer technologies like data retrieval, resulting in low efficiency (averaging only 100-200 data entries per day), and the need for manual evidence preparation, which is time-consuming and labor-intensive. Therefore, there is an urgent need for an efficient and accurate method to identify toll evasion by green channel vehicles to address the shortcomings of existing technologies.

[0003] Effective technical solutions are urgently needed to address the above problems. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, and medium for identifying toll evasion by vehicles using highway green channels. By constructing image data analysis and core mining, the system can perform multi-dimensional analysis of vehicle passage record data in green channel lanes, accurately identify various toll evasion behaviors, and thus achieve accurate identification of toll evasion by vehicles using highway green channels.

[0005] Firstly, this application provides a method for identifying toll evasion by vehicles using highway green channels, including the following steps:

[0006] Obtain vehicle passage record data from green channel lanes and preprocess it to obtain optimized vehicle passage record data;

[0007] Construct a rule set for identifying toll evasion by green channel vehicles;

[0008] Based on the optimized data of the vehicle passage records and the rule set for identifying toll evasion of green channel vehicles, the status of suspected toll evading vehicles is obtained, including suspected toll evading vehicles or non-suspected toll evading vehicles.

[0009] If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring will be performed;

[0010] If the suspected toll evasion vehicle is in the status of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle.

[0011] The confidence level of evasion is compared with a preset confidence threshold for evasion.

[0012] If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated.

[0013] If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated using the preset green channel evasion identification template and sent to the terminal for display.

[0014] Optionally, in the method for identifying toll evasion by vehicles using the green channel lane described in this application, the step of acquiring vehicle passage record data of the green channel lane and preprocessing it to obtain optimized vehicle passage record data includes:

[0015] Obtain vehicle passage record data for green channel lanes, including vehicle license plate numbers, cargo category characteristic data, vehicle registration certificate images, and vehicle images;

[0016] The vehicle license plate, cargo category feature data, vehicle registration certificate image, and vehicle image are formatted, correlated, and preprocessed to obtain optimized vehicle passage record data, including standard vehicle license plate, standard cargo category feature data, denoised vehicle registration certificate image, and denoised vehicle image.

[0017] Optionally, the method for identifying toll evasion by expressway green channel vehicles described in this application further includes:

[0018] Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data;

[0019] Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

[0020] Optionally, in the method for identifying toll evasion by green channel vehicles described in this application, the step of constructing a rule set for identifying toll evasion by green channel vehicles includes:

[0021] Construct a rule set for identifying toll evasion by green channel vehicles;

[0022] The set of rules for identifying toll evasion by green channel vehicles includes rules for identifying vehicles with the same license plate but different driver's licenses, rules for identifying abnormal vehicle license marking sizes, rules for identifying vehicles with counterfeit license plates, rules for identifying abnormal passage records of trailers, and rules for identifying abnormal payment records of vehicles with the same license plate.

[0023] Optionally, in the method for identifying toll evasion by green channel vehicles described in this application, the step of identifying green channel vehicles for toll evasion based on the optimized data of the vehicle passage records combined with the rule set for identifying green channel vehicles for toll evasion, and obtaining the status of suspected toll evading vehicles, includes:

[0024] Based on the standard vehicle license plate, query the corresponding denoised vehicle license image, and process the vehicle license text data corresponding to the denoised vehicle license image by matching the recognition rules of different vehicle licenses with the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0025] The vehicle width data corresponding to the denoised vehicle license image is combined with the vehicle license text data to match the vehicle license mark size anomaly recognition rules and obtain the status of suspected toll evasion vehicles.

[0026] Based on the standard vehicle license plate, query the corresponding denoised vehicle image's front outline image, headlight position data, and body color data, and process them according to the rules for identifying cloned vehicles to obtain the status of suspected toll-evading vehicles.

[0027] Extract the trailer's license plate from the denoised vehicle image, query the corresponding first green channel payment record for the trailer, and process the trailer's abnormal passage record by matching the first green channel payment record with the trailer's abnormal passage record identification rules to obtain the status of the suspected toll evasion vehicle.

[0028] Based on the standard vehicle license plate, query the second green channel payment record of the preset passage time sequence, and process it according to the anomaly identification rules of the payment record of the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0029] Optionally, in the method for identifying toll evasion by expressway green channel vehicles described in this application, if the suspected toll evasion vehicle is in the state of suspected toll evasion, then the suspected toll evasion vehicle is analyzed and processed according to a preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle, including:

[0030] The seasonal matching rate is obtained by analyzing and processing the cargo category characteristic data.

[0031] Obtain the image recognition accuracy score of suspected toll evading vehicles, the matching degree score of green channel vehicle toll evasion recognition rules, and the data integrity score;

[0032] The image recognition accuracy score, the matching score of the green channel vehicle toll evasion identification rule, and the data integrity score are weighted and summed, and then divided by the seasonal matching rate to obtain the toll evasion confidence of the suspected toll evading vehicle.

[0033] Secondly, this application provides a system for identifying toll evasion by expressway green channel vehicles. The system includes a memory and a processor. The memory includes a program for identifying toll evasion by expressway green channel vehicles. When the program for identifying toll evasion by expressway green channel vehicles is executed by the processor, it performs the following steps:

[0034] Obtain vehicle passage record data from green channel lanes and preprocess it to obtain optimized vehicle passage record data;

[0035] Construct a rule set for identifying toll evasion by green channel vehicles;

[0036] Based on the optimized data of the vehicle passage records and the rule set for identifying toll evasion of green channel vehicles, the status of suspected toll evading vehicles is obtained, including suspected toll evading vehicles or non-suspected toll evading vehicles.

[0037] If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring is performed;

[0038] If the suspected toll evasion vehicle is in the status of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle.

[0039] The confidence level of evasion of payment is compared with a preset confidence threshold for evasion of payment.

[0040] If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated.

[0041] If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated using the preset green channel evasion identification template and sent to the terminal for display.

[0042] Optionally, in the highway green channel vehicle toll evasion identification system described in this application, the step of acquiring vehicle passage record data of the green channel and preprocessing it to obtain optimized vehicle passage record data includes:

[0043] Obtain vehicle passage record data for green channel lanes, including vehicle license plate numbers, cargo category characteristic data, vehicle registration certificate images, and vehicle images;

[0044] The vehicle license plate, cargo category feature data, vehicle registration certificate image, and vehicle image are formatted, correlated, and preprocessed to obtain optimized vehicle passage record data, including standard vehicle license plate, standard cargo category feature data, denoised vehicle registration certificate image, and denoised vehicle image.

[0045] Optionally, the toll evasion identification system for highway green channel vehicles described in this application further includes:

[0046] Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data;

[0047] Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

[0048] Thirdly, this application also provides a computer-readable storage medium storing a program for identifying toll evasion by high-speed green channel vehicles. When the program is executed by a processor, it implements the steps of the method for identifying toll evasion by high-speed green channel vehicles as described in any of the above claims.

[0049] As can be seen from the above, the method, system and medium for identifying toll evasion by green channel vehicles provided in this application conduct multi-dimensional analysis of green channel vehicle passage record data through image data analysis and core mining, accurately identify various toll evasion behaviors, and thus achieve accurate identification of toll evasion by green channel vehicles.

[0050] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart illustrating a method for identifying toll evasion by vehicles using green channels on highways, as provided in this application embodiment;

[0053] Figure 2 A flowchart illustrating the process of obtaining optimized vehicle passage record data for a method of identifying toll evasion by expressway green channel vehicles provided in this application embodiment;

[0054] Figure 3 A flowchart illustrating the method for identifying toll evasion by vehicles using highway green channels, provided in this application embodiment, shows how to obtain confidence in the toll evasion of suspected toll evading vehicles.

[0055] Figure 4 This is a high-level flowchart of the methods of various embodiments of this application. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0057] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0058] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying toll evasion by expressway green channel vehicles, as described in some embodiments of this application. This method is used in terminal devices, such as computers and mobile terminals. The method includes the following steps:

[0059] S11. Obtain vehicle passage record data for the green channel and preprocess it to obtain optimized vehicle passage record data;

[0060] S12. Construct a rule set for identifying toll evasion by green channel vehicles;

[0061] S13. Based on the optimized data of the vehicle passage record and combined with the green channel vehicle toll evasion identification rule set, green channel vehicle toll evasion identification is performed to obtain the status of suspected toll evasion vehicles, including suspected toll evasion vehicles or non-suspected toll evasion vehicles.

[0062] S141. If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring is performed;

[0063] S142. If the suspected toll evasion vehicle status is a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle.

[0064] S15. Compare the confidence level of evasion with the preset confidence threshold for evasion;

[0065] S161. If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated.

[0066] S162. If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated through the preset green channel evasion identification template and sent to the terminal for display.

[0067] It should be noted that, in order to identify suspicious green channel vehicles from all green channel vehicle flow records and compile relevant evidence for inspectors' reference, the process first involves acquiring and preprocessing vehicle passage record data for a preset time period (e.g., one calendar month) to obtain optimized vehicle passage record data. Simultaneously, a multi-dimensional set of rules for identifying green channel vehicle toll evasion is constructed. The optimized vehicle passage record data is then deeply mined and matched with the constructed set of rules for identifying green channel vehicle toll evasion. Based on the matching results, the status of suspected toll evading vehicles is determined. If a vehicle is suspected of toll evasion, its toll evasion confidence is further assessed. Finally, accurate identification is achieved through threshold comparison. Different processing strategies are applied in real time based on the threshold comparison results. If the threshold is less than the preset toll evasion confidence threshold, a suspected toll evading vehicle list is generated, and manual review is activated for further confirmation. Conversely, if the threshold is higher, a green channel vehicle toll evasion identification report is generated based on the preset green channel toll evasion identification template, including vehicle license plate number, suspicious type, passage record, and identification basis.

[0068] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the process of obtaining optimized vehicle passage record data in a method for identifying toll evasion by vehicles using green lanes, as described in some embodiments of this application. According to embodiments of the present invention, obtaining vehicle passage record data from green lanes and performing preprocessing to obtain optimized vehicle passage record data includes:

[0069] S21. Obtain vehicle passage record data for green channel lanes, including vehicle license plate, cargo category characteristic data, vehicle registration certificate image, and vehicle image;

[0070] S22. Perform format unification, data association, and data cleaning preprocessing on the vehicle license plate, transport cargo category feature data, vehicle registration certificate image, and vehicle image to obtain optimized vehicle passage record data, including standard vehicle license plate, standard transport cargo category feature data, denoised vehicle registration certificate image, and denoised vehicle image.

[0071] It should be noted that data from different sources (such as local databases at toll stations, lane monitoring systems, and manual registration terminals) are acquired in batches through API interfaces and direct database connections to form a raw data pool. Text data (such as vehicle license plates) is uniformly converted into structured data formats and standardized, for example, converting Lu A123bp to Lu A123BP. Image data (such as vehicle registration certificate images) is uniformly stored in JPG or PNG formats and named according to the rule of "vehicle license plate + passage time + image type" (such as "Lu A123BP_20200507223922_vehicle registration certificate.jpg") and associated with the corresponding text data entries of the vehicle. At the same time, incomplete, noisy, and inconsistent data are removed to ensure the integrity and accuracy of the data. The standard transport cargo category refers to the cargo category enjoying green channel access in the catalog of fresh agricultural products. The characteristic data of the standard transport cargo category is represented by a unique identifier.

[0072] According to an embodiment of the present invention, it further includes:

[0073] Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data;

[0074] Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

[0075] It should be noted that, in order to deeply mine the data required for toll evasion identification, the corresponding vehicle width data and vehicle registration text data were obtained by using an image analysis engine and AI image recognition technology based on the denoised vehicle image. Data extraction was performed on the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and body color data. Among them, the headlight position data includes the headlight position data, side headlight position data, and rear headlight position data, and the body color data includes trailer color data and trailer color data.

[0076] According to an embodiment of the present invention, the construction of the rule set for identifying toll evasion by green channel vehicles includes:

[0077] Construct a rule set for identifying toll evasion by green channel vehicles;

[0078] The set of rules for identifying toll evasion by green channel vehicles includes rules for identifying vehicles with the same license plate but different driver's licenses, rules for identifying abnormal vehicle license marking sizes, rules for identifying vehicles with counterfeit license plates, rules for identifying abnormal passage records of trailers, and rules for identifying abnormal payment records of vehicles with the same license plate.

[0079] It should be noted that, in order to optimize the data from the in-depth mining of vehicle passage records for green channel toll evasion identification from multiple dimensions, a multi-dimensional set of rules for identifying green channel vehicle toll evasion has been constructed. Among these, the rule for identifying different vehicle registration certificates for the same license plate refers to identifying non-unique registration certificates for the same license plate or any registration certificate containing the phrase "only for transporting non-detachable objects." The purpose is to target certain vehicles that use genuine certificates on less strictly managed road sections while using fake certificates on other sections, or that use transport vehicles that do not meet the requirements for green channel transport vehicles. The rule for identifying abnormal registration certificate dimensions involves extracting the vehicle outline dimensions from the registration certificate and determining the vehicle width parameter (i.e., vehicle width data) using an edge detection algorithm. If the width value is greater than a preset width threshold, and the registration certificate text does not contain the phrase "only for transporting non-detachable objects," the purpose is to identify vehicles that consistently use fake registration certificates. In such cases, the image certificates stored in the system may all be identical, and there will be no non-unique registration certificates. However, these are fake license plates. The rules for identifying cloned vehicles refer to comparing the vehicle's front outline image, headlight position data, and body color data from multiple passage records of the same vehicle license plate using a feature point matching algorithm to calculate image similarity. The purpose is to identify multiple vehicles using the same license plate and the same driver's license to evade tolls by using the green channel. The rules for identifying abnormal trailer passage records refer to whether the same trailer used by different trailers simultaneously has both green channel payment and free green channel records. The purpose is to identify whether the same trailer has both green channel payment and free green channel records simultaneously. The rules for identifying abnormal payment records for the same vehicle license plate refer to the same vehicle's passage time sequence where the preceding record is free and the subsequent record is paid, or where there are both paid and free records within the same time period. The purpose is to identify certain vehicles that enjoy free passage through the green channel in poorly managed sections, but pay the actual toll when using the green channel in other sections. To adapt to the dynamic development of toll evasion identification, the set of rules for identifying green channel vehicle toll evasion can be dynamically modified by those skilled in the art.

[0080] According to an embodiment of the present invention, the step of identifying green channel vehicles for toll evasion based on the optimized vehicle passage record data and the green channel vehicle toll evasion identification rule set, and obtaining the status of suspected toll evading vehicles, includes:

[0081] Based on the standard vehicle license plate, query the corresponding denoised vehicle license image, and process the vehicle license text data corresponding to the denoised vehicle license image by matching the recognition rules of different vehicle licenses with the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0082] The vehicle width data corresponding to the denoised vehicle license image is combined with the vehicle license text data to match the vehicle license mark size anomaly recognition rules and obtain the status of suspected toll evasion vehicles.

[0083] Based on the standard vehicle license plate, query the corresponding denoised vehicle image's front outline image, headlight position data, and body color data, and process them according to the rules for identifying cloned vehicles to obtain the status of suspected toll-evading vehicles.

[0084] Extract the trailer's license plate from the denoised vehicle image, query the corresponding first green channel payment record for the trailer, and process the trailer's abnormal passage record by matching the first green channel payment record with the trailer's abnormal passage record identification rules to obtain the status of the suspected toll evasion vehicle.

[0085] Based on the standard vehicle license plate, query the second green channel payment record of the preset passage time sequence, and process it according to the anomaly identification rules of the payment record of the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0086] It should be noted that the rule for identifying different vehicle license plates based on matching the text data of the denoised vehicle license image with the same license plate means that if the denoised vehicle license image corresponding to the standard vehicle license plate is not unique, or if any denoised vehicle license image contains the phrase "only for transporting non-removable objects," then the vehicle suspected of evading tolls is classified as a suspected toll evader; otherwise, it is classified as not a suspected toll evader. The rule for identifying abnormal vehicle license mark sizes based on matching the vehicle width data corresponding to the denoised vehicle license image with the text data means that the vehicle width data determined by the edge detection algorithm is compared with a preset width threshold (e.g., 2550mm for ordinary trucks). If it is greater than the preset width threshold and there is no "only for transporting non-removable objects," then the vehicle suspected of evading tolls is classified as a suspected toll evader; otherwise, it is classified as not a suspected toll evader. This rule is used to identify fake vehicle license plates. To identify vehicles with cloned license plates, the front outline image, headlight position data, and body color of the denoised vehicle image from multiple passage records are retrieved based on the standard vehicle license plate. The system analyzes color data and matches it with rules for identifying vehicles using counterfeit license plates. This involves using a feature point matching algorithm to compare images and calculate image similarity. If the similarity is less than a preset threshold, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle. The system also extracts the trailer's license plate from the denoised vehicle image and queries the first green channel payment record for the same trailer. This first green channel payment record is then matched against rules for identifying abnormal trailer passage records. If both a green channel payment record and a green channel free passage record exist simultaneously, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle. For integrated ordinary trucks, the system queries the second green channel payment record based on a preset passage time sequence using the standard vehicle license plate. The preset passage time sequence refers to two passage records with a time interval less than a preset period (e.g., 2 hours) and a continuous route. If both a green channel payment record and a green channel free passage record exist simultaneously, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle.

[0087] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the process of obtaining the toll evasion confidence of a suspected toll evading vehicle in a method for identifying toll evasion by vehicles using highway green channels, as described in some embodiments of this application. According to an embodiment of the present invention, if the suspected toll evading vehicle is classified as such, then the suspected toll evading vehicle is analyzed and processed using a preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evading vehicle, including:

[0088] S31. Analyze and process the transported goods category characteristic data to obtain the seasonal matching rate;

[0089] S32. Obtain the image recognition accuracy score, green channel vehicle toll evasion recognition rule matching score, and data integrity score of suspected toll evasion vehicles;

[0090] S33. The image recognition accuracy score, the matching score of the green channel vehicle toll evasion identification rule, and the data integrity score are weighted and summed, and then divided by the seasonal matching rate to obtain the toll evasion confidence of the suspected toll evasion vehicle.

[0091] It should be noted that the seasonal matching rate is obtained by analyzing and processing the characteristic data of the transported goods category. The seasonal matching rate is evaluated by those skilled in the art based on the transported goods category and the corresponding seasonal transport cycle. For example, the lychee season is generally from May to August. During this period, the seasonal matching rate is evaluated as 100%. If the season exceeds the time limit, the seasonal matching rate is lowered according to the preset gradient based on the time exceeded. If it is lower than the preset seasonal matching rate, it is calculated according to the preset seasonal matching rate, and a warning response is output simultaneously. The image recognition accuracy score is based on the text recognition confidence score output by the image analysis engine (for the recognition results of key text such as "only for transporting non-disassembled objects" on the vehicle registration certificate, the image analysis engine will output a confidence score of 0-100, such as 90 points when the recognition is clear and 60 points when it is blurry) and the size extraction accuracy (obtained based on the vehicle width extraction error rate, such as the actual width of 2600mm, the recognition result is 2580mm, and the error rate is 2600mm). The image recognition accuracy score (0.77%, corresponding to 95 points; when the error rate is >5%, it drops to 60 points) and the vehicle image matching score (the similarity of multiple passage image features of the same vehicle license plate; for example, in license plate counterfeiting, an 85% matching score corresponds to 85 points, and <70% is considered a mismatch, corresponding to 50 points) are obtained by weighted summation. The green channel vehicle toll evasion identification rule matching score (each identification rule is assigned a weight value; a successful match is scored as 100 points, otherwise 0 points, and all identification rules are weighted summation) and the data integrity score (different scores are assigned to vehicle license plate, transported goods category feature data, vehicle registration certificate image, and vehicle image; the corresponding score is deducted according to the missing information) are obtained by weighted summation of the obtained image recognition accuracy score, green channel vehicle toll evasion identification rule matching score, and data integrity score, and divided by the seasonal matching rate to obtain the toll evasion confidence of suspected toll evading vehicles. The weight values ​​are preset according to the specific application and can be dynamically adjusted.

[0092] Please refer to Figure 4 , Figure 4 This is a high-level flowchart of various embodiments of the methods in this application, which can be used in methods for identifying toll evasion by vehicles using the green channel. According to embodiments of the present invention, firstly, vehicle passage record data of the green channel is acquired and preprocessed to obtain optimized vehicle passage record data. Simultaneously, a multi-dimensional set of rules for identifying toll evasion by green channel vehicles is constructed. Both are processed together to identify suspected toll evading vehicles. To prevent false judgments, a toll evasion confidence assessment is performed, and a threshold comparison is used to finally determine whether toll evasion behavior exists. This enables timely detection of new toll evasion methods such as alternating use of genuine and fake driver's licenses and short-distance loop toll evasion.

[0093] It is worth mentioning that, according to embodiments of the present invention, it further includes:

[0094] The denoised vehicle registration image is input into a preset image tampering detection model for monitoring, and the probability value of vehicle registration image tampering is obtained and compared with a preset vehicle registration tampering probability threshold.

[0095] If the probability value of the vehicle registration certificate image being tampered with is greater than or equal to the preset probability threshold for vehicle registration certificate tampering, then an early warning response is output.

[0096] If the probability value of the vehicle registration certificate image being tampered with is less than the preset probability threshold for vehicle registration certificate tampering, then data extraction is performed based on the denoised vehicle registration certificate image to obtain the extracted curb weight and load capacity.

[0097] The curb weight and load capacity are compared with the corresponding parameters in the preset vehicle management database to obtain the curb weight deviation rate and load capacity deviation rate.

[0098] Acquire vehicle trajectory analysis data and cargo characteristic data within the preset itinerary. The trajectory analysis data includes the driving speed fluctuation coefficient, the number of sudden decelerations, and the frequency of green channel round trips at stations. The cargo characteristic data includes cargo consistency verification data and loading volume ratio.

[0099] Based on the curb weight deviation rate and load capacity deviation rate, as well as the driving speed fluctuation coefficient, number of sudden decelerations and frequency of green channel round trips at stations, combined with the cargo consistency verification data and loading volume ratio, a feature vector for identifying toll evasion of green channel vehicles is constructed.

[0100] The feature vector for identifying toll evasion of green channel vehicles is input into a preset toll evasion risk identification model for processing to obtain the probability value of toll evasion risk.

[0101] The probability value of evasion risk is compared with the preset probability threshold of evasion risk.

[0102] If the probability value of the toll evasion risk is less than or equal to the preset toll evasion risk probability threshold, it is determined to be a normal green channel vehicle;

[0103] If the probability value of toll evasion risk is greater than the preset toll evasion risk probability threshold, the vehicle is identified as an abnormal green channel vehicle, and a manual inspection response is activated.

[0104] It should be noted that, in order to achieve real-time identification of toll evasion risks by green channel vehicles, when a green channel vehicle enters the highway, the data collection equipment at the entrance toll station immediately acquires a photo of the vehicle's driver's license. During the vehicle's journey, the gantry system collects its driving trajectory data in real time. At the exit toll station, staff inspect the cargo from different angles (front, rear, sides, and top of the vehicle) and collect relevant photos or videos. First, a pre-set image tampering detection model based on deep learning (such as a tampering detection model based on generative adversarial networks) is used to analyze the denoised driver's license image, detect Photoshop traces, and generate a tampered driver's license image. The probability value is modified, and a threshold comparison is used to determine whether tampering has occurred. If no tampering is found, the curb weight and load capacity are further extracted from the denoised vehicle registration image and compared with the corresponding parameters in the preset vehicle management database to obtain the curb weight deviation rate and load capacity deviation rate. The preset image tampering detection model is trained by acquiring a large number of historical denoised vehicle registration images and their corresponding tampering probability values. The curb weight deviation rate is the ratio of the absolute value of the difference between the curb weight and the curb weight registration data in the vehicle management database to the curb weight registration data in the vehicle management database. The load capacity deviation rate... The rate refers to the ratio of the absolute value of the difference between the registered load capacity and the registered load capacity data in the vehicle management database to the registered load capacity data in the vehicle management database; the speed fluctuation coefficient refers to the ratio of the standard deviation of the driving speed within a preset journey (such as from toll station A to toll station B or from gantry A to gantry B) to the average driving speed; the number of sudden decelerations refers to the number of times the speed change rate exceeds 20 km / h / min (i.e., the acceleration or deceleration exceeds 20 km / h within 1 minute) exceeds 3 times within a preset time period (such as 1 hour) within a preset journey. Frequent sudden decelerations may be related to vehicles deliberately evading inspection. For example, when a vehicle approaches a toll station or inspection point, it may suddenly slow down to observe the strictness of the on-site inspection and take the opportunity to choose whether to use fake documents, disguise goods, or other means to evade tolls. The frequency of green channel round trips refers to the number of times the same vehicle license plate travels to and from the same pair of toll stations within a preset time period (such as one week) and applies for green channel access. The cargo consistency verification data includes whether the cargo is consistent or inconsistent. It is obtained by extracting features from cargo photos or videos, combining them with deep learning models (such as convolutional neural networks) to identify the cargo category, and comparing it with the declared category. The loading volume ratio refers to the ratio of the cargo loading volume to the vehicle's approved cargo volume.The obtained curb weight deviation rate, load capacity deviation rate, speed fluctuation coefficient, number of sudden decelerations, and frequency of green channel round trips to stations are integrated with cargo consistency verification data and loading volume ratio to form a feature vector. Each feature is assigned a different weight based on its impact on toll evasion risk (the weights are trained using historical toll evasion case data). This integrated green channel vehicle toll evasion identification feature vector is then input into a pre-set toll evasion risk identification model for processing to obtain a toll evasion risk probability value. The toll evasion risk is identified by threshold comparison, thus achieving the integration of vehicle basic information, driving trajectory, and multi-source cargo data to extract toll evasion risk features from multiple dimensions. Compared to traditional identification methods that rely solely on cargo information, this approach can more comprehensively and accurately identify toll evasion behavior and effectively reduce the false positive rate.

[0105] This invention also discloses a highway green channel vehicle toll evasion identification system, including a memory and a processor. The memory includes a highway green channel vehicle toll evasion identification method program. When the processor executes the highway green channel vehicle toll evasion identification method program, it performs the following steps:

[0106] Obtain vehicle passage record data from green channel lanes and preprocess it to obtain optimized vehicle passage record data;

[0107] Construct a rule set for identifying toll evasion by green channel vehicles;

[0108] Based on the optimized data of the vehicle passage records and the rule set for identifying toll evasion of green channel vehicles, the status of suspected toll evading vehicles is obtained, including suspected toll evading vehicles or non-suspected toll evading vehicles.

[0109] If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring will be performed;

[0110] If the suspected toll evasion vehicle is in the status of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle.

[0111] The confidence level of evasion is compared with a preset confidence threshold for evasion.

[0112] If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated.

[0113] If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated using the preset green channel evasion identification template and sent to the terminal for display.

[0114] It should be noted that, in order to identify suspicious green channel vehicles from all green channel vehicle flow records and compile relevant evidence for inspectors' reference, the process first involves acquiring and preprocessing vehicle passage record data for a preset time period (e.g., one calendar month) to obtain optimized vehicle passage record data. Simultaneously, a multi-dimensional set of rules for identifying green channel vehicle toll evasion is constructed. The optimized vehicle passage record data is then deeply mined and matched with the constructed set of rules for identifying green channel vehicle toll evasion. Based on the matching results, the status of suspected toll evading vehicles is determined. If a vehicle is suspected of toll evasion, its toll evasion confidence is further assessed. Finally, accurate identification is achieved through threshold comparison. Different processing strategies are applied in real time based on the threshold comparison results. If the threshold is less than the preset toll evasion confidence threshold, a suspected toll evading vehicle list is generated, and manual review is activated for further confirmation. Conversely, if the threshold is higher, a green channel vehicle toll evasion identification report is generated based on the preset green channel toll evasion identification template, including vehicle license plate number, suspicious type, passage record, and identification basis.

[0115] According to an embodiment of the present invention, the step of acquiring vehicle passage record data of green channel lanes and performing preprocessing to obtain optimized vehicle passage record data includes:

[0116] Obtain vehicle passage record data for green channel lanes, including vehicle license plate numbers, cargo category characteristic data, vehicle registration certificate images, and vehicle images;

[0117] The vehicle license plate, cargo category feature data, vehicle registration certificate image, and vehicle image are formatted, correlated, and preprocessed to obtain optimized vehicle passage record data, including standard vehicle license plate, standard cargo category feature data, denoised vehicle registration certificate image, and denoised vehicle image.

[0118] It should be noted that data from different sources (such as local databases at toll stations, lane monitoring systems, and manual registration terminals) are acquired in batches through API interfaces and direct database connections to form a raw data pool. Text data (such as vehicle license plates) is uniformly converted into structured data formats and standardized, for example, converting Lu A123bp to Lu A123BP. Image data (such as vehicle registration certificate images) is uniformly stored in JPG or PNG formats and named according to the rule of "vehicle license plate + passage time + image type" (such as "Lu A123BP_20200507223922_vehicle registration certificate.jpg") and associated with the corresponding text data entries of the vehicle. At the same time, incomplete, noisy, and inconsistent data are removed to ensure the integrity and accuracy of the data. The standard transport cargo category refers to the cargo category enjoying green channel access in the catalog of fresh agricultural products. The characteristic data of the standard transport cargo category is represented by a unique identifier.

[0119] According to an embodiment of the present invention, it further includes:

[0120] Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data;

[0121] Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

[0122] It should be noted that, in order to deeply mine the data required for toll evasion identification, the corresponding vehicle width data and vehicle registration text data were obtained by using an image analysis engine and AI image recognition technology based on the denoised vehicle image. Data extraction was performed on the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and body color data. Among them, the headlight position data includes the headlight position data, side headlight position data, and rear headlight position data, and the body color data includes trailer color data and trailer color data.

[0123] According to an embodiment of the present invention, the construction of the rule set for identifying toll evasion by green channel vehicles includes:

[0124] Construct a rule set for identifying toll evasion by green channel vehicles;

[0125] The set of rules for identifying toll evasion by green channel vehicles includes rules for identifying vehicles with the same license plate but different driver's licenses, rules for identifying abnormal vehicle license marking sizes, rules for identifying vehicles with counterfeit license plates, rules for identifying abnormal passage records of trailers, and rules for identifying abnormal payment records of vehicles with the same license plate.

[0126] It should be noted that, in order to optimize the data from the in-depth mining of vehicle passage records for green channel toll evasion identification from multiple dimensions, a multi-dimensional set of rules for identifying green channel vehicle toll evasion has been constructed. Among these, the rule for identifying different vehicle registration certificates for the same license plate refers to identifying non-unique registration certificates for the same license plate or any registration certificate containing the phrase "only for transporting non-detachable objects." The purpose is to target certain vehicles that use genuine certificates on less strictly managed road sections while using fake certificates on other sections, or that use transport vehicles that do not meet the requirements for green channel transport vehicles. The rule for identifying abnormal registration certificate dimensions involves extracting the vehicle outline dimensions from the registration certificate and determining the vehicle width parameter (i.e., vehicle width data) using an edge detection algorithm. If the width value is greater than a preset width threshold, and the registration certificate text does not contain the phrase "only for transporting non-detachable objects," the purpose is to identify vehicles that consistently use fake registration certificates. In such cases, the image certificates stored in the system may all be identical, and there will be no non-unique registration certificates. However, these are fake license plates. The rules for identifying cloned vehicles refer to comparing the vehicle's front outline image, headlight position data, and body color data from multiple passage records of the same vehicle license plate using a feature point matching algorithm to calculate image similarity. The purpose is to identify multiple vehicles using the same license plate and the same driver's license to evade tolls by using the green channel. The rules for identifying abnormal trailer passage records refer to whether the same trailer used by different trailers simultaneously has both green channel payment and free green channel records. The purpose is to identify whether the same trailer has both green channel payment and free green channel records simultaneously. The rules for identifying abnormal payment records for the same vehicle license plate refer to the same vehicle's passage time sequence where the preceding record is free and the subsequent record is paid, or where there are both paid and free records within the same time period. The purpose is to identify certain vehicles that enjoy free passage through the green channel in poorly managed sections, but pay the actual toll when using the green channel in other sections. To adapt to the dynamic development of toll evasion identification, the set of rules for identifying green channel vehicle toll evasion can be dynamically modified by those skilled in the art.

[0127] According to an embodiment of the present invention, the step of identifying green channel vehicles for toll evasion based on the optimized vehicle passage record data and the green channel vehicle toll evasion identification rule set, and obtaining the status of suspected toll evading vehicles, includes:

[0128] Based on the standard vehicle license plate, query the corresponding denoised vehicle license image, and process the vehicle license text data corresponding to the denoised vehicle license image by matching the recognition rules of different vehicle licenses with the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0129] The vehicle width data corresponding to the denoised vehicle license image is combined with the vehicle license text data to match the vehicle license mark size anomaly recognition rules and obtain the status of suspected toll evasion vehicles.

[0130] Based on the standard vehicle license plate, query the corresponding denoised vehicle image's front outline image, headlight position data, and body color data, and process them according to the rules for identifying cloned vehicles to obtain the status of suspected toll-evading vehicles.

[0131] Extract the trailer's license plate from the denoised vehicle image, query the corresponding first green channel payment record for the trailer, and process the trailer's abnormal passage record by matching the first green channel payment record with the trailer's abnormal passage record identification rules to obtain the status of the suspected toll evasion vehicle.

[0132] Based on the standard vehicle license plate, query the second green channel payment record of the preset passage time sequence, and process it according to the anomaly identification rules of the payment record of the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

[0133] It should be noted that the rule for identifying different vehicle license plates based on matching the text data of the denoised vehicle license image with the same license plate means that if the denoised vehicle license image corresponding to the standard vehicle license plate is not unique, or if any denoised vehicle license image contains the phrase "only for transporting non-removable objects," then the vehicle suspected of evading tolls is classified as a suspected toll evader; otherwise, it is classified as not a suspected toll evader. The rule for identifying abnormal vehicle license mark sizes based on matching the vehicle width data corresponding to the denoised vehicle license image with the text data means that the vehicle width data determined by the edge detection algorithm is compared with a preset width threshold (e.g., 2550mm for ordinary trucks). If it is greater than the preset width threshold and there is no "only for transporting non-removable objects," then the vehicle suspected of evading tolls is classified as a suspected toll evader; otherwise, it is classified as not a suspected toll evader. This rule is used to identify fake vehicle license plates. To identify vehicles with cloned license plates, the front outline image, headlight position data, and body color of the denoised vehicle image from multiple passage records are retrieved based on the standard vehicle license plate. The system analyzes color data and matches it with rules for identifying vehicles using counterfeit license plates. This involves using a feature point matching algorithm to compare images and calculate image similarity. If the similarity is less than a preset threshold, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle. The system also extracts the trailer's license plate from the denoised vehicle image and queries the first green channel payment record for the same trailer. This first green channel payment record is then matched against rules for identifying abnormal trailer passage records. If both a green channel payment record and a green channel free passage record exist simultaneously, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle. For integrated ordinary trucks, the system queries the second green channel payment record based on a preset passage time sequence using the standard vehicle license plate. The preset passage time sequence refers to two passage records with a time interval less than a preset period (e.g., 2 hours) and a continuous route. If both a green channel payment record and a green channel free passage record exist simultaneously, the vehicle is considered a suspected toll evader; otherwise, it is considered a legitimate vehicle.

[0134] According to an embodiment of the present invention, if the suspected toll evasion vehicle is in the state of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to a preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle, including:

[0135] The seasonal matching rate is obtained by analyzing and processing the cargo category characteristic data.

[0136] Obtain the image recognition accuracy score of suspected toll evading vehicles, the matching degree score of green channel vehicle toll evasion recognition rules, and the data integrity score;

[0137] The image recognition accuracy score, the matching score of the green channel vehicle toll evasion identification rule, and the data integrity score are weighted and summed, and then divided by the seasonal matching rate to obtain the toll evasion confidence of the suspected toll evading vehicle.

[0138] It should be noted that the seasonal matching rate is obtained by analyzing and processing the characteristic data of the transported goods category. The seasonal matching rate is evaluated by those skilled in the art based on the transported goods category and the corresponding seasonal transport cycle. For example, the lychee season is generally from May to August. During this period, the seasonal matching rate is evaluated as 100%. If the season exceeds the time limit, the seasonal matching rate is lowered according to the preset gradient based on the time exceeded. If it is lower than the preset seasonal matching rate, it is calculated according to the preset seasonal matching rate, and a warning response is output simultaneously. The image recognition accuracy score is based on the text recognition confidence score output by the image analysis engine (for the recognition results of key text such as "only for transporting non-disassembled objects" on the vehicle registration certificate, the image analysis engine will output a confidence score of 0-100, such as 90 points when the recognition is clear and 60 points when it is blurry) and the size extraction accuracy (obtained based on the vehicle width extraction error rate, such as the actual width of 2600mm, the recognition result is 2580mm, and the error rate is 2600mm). The image recognition accuracy score (0.77%, corresponding to 95 points; when the error rate is >5%, it drops to 60 points) and the vehicle image matching score (the similarity of multiple passage image features of the same vehicle license plate; for example, in license plate counterfeiting, an 85% matching score corresponds to 85 points, and <70% is considered a mismatch, corresponding to 50 points) are obtained by weighted summation. The green channel vehicle toll evasion identification rule matching score (each identification rule is assigned a weight value; a successful match is scored as 100 points, otherwise 0 points, and all identification rules are weighted summation) and the data integrity score (different scores are assigned to vehicle license plate, transported goods category feature data, vehicle registration certificate image, and vehicle image; the corresponding score is deducted according to the missing information) are obtained by weighted summation of the obtained image recognition accuracy score, green channel vehicle toll evasion identification rule matching score, and data integrity score, and divided by the seasonal matching rate to obtain the toll evasion confidence of suspected toll evading vehicles. The weight values ​​are preset according to the specific application and can be dynamically adjusted.

[0139] According to an embodiment of the present invention, firstly, vehicle passage record data of green channel lanes is acquired and preprocessed to obtain optimized vehicle passage record data. At the same time, a multi-dimensional set of rules for identifying green channel vehicle toll evasion is constructed. Both are processed together to identify suspected toll evasion vehicles. To prevent misjudgment, toll evasion confidence is assessed, and the existence of toll evasion behavior is finally determined by threshold comparison. This enables timely detection of new toll evasion methods such as alternating use of genuine and fake driver's licenses and short-distance loop toll evasion.

[0140] It is worth mentioning that, according to embodiments of the present invention, it further includes:

[0141] The denoised vehicle registration image is input into a preset image tampering detection model for monitoring, and the probability value of vehicle registration image tampering is obtained and compared with a preset vehicle registration tampering probability threshold.

[0142] If the probability value of the vehicle registration certificate image being tampered with is greater than or equal to the preset probability threshold for vehicle registration certificate tampering, then an early warning response is output.

[0143] If the probability value of the vehicle registration certificate image being tampered with is less than the preset probability threshold for vehicle registration certificate tampering, then data extraction is performed based on the denoised vehicle registration certificate image to obtain the extracted curb weight and load capacity.

[0144] The curb weight and load capacity are compared with the corresponding parameters in the preset vehicle management database to obtain the curb weight deviation rate and load capacity deviation rate.

[0145] Acquire vehicle trajectory analysis data and cargo characteristic data within the preset itinerary. The trajectory analysis data includes the driving speed fluctuation coefficient, the number of sudden decelerations, and the frequency of green channel round trips at stations. The cargo characteristic data includes cargo consistency verification data and loading volume ratio.

[0146] Based on the curb weight deviation rate and load capacity deviation rate, as well as the driving speed fluctuation coefficient, number of sudden decelerations and frequency of green channel round trips at stations, combined with the cargo consistency verification data and loading volume ratio, a feature vector for identifying toll evasion of green channel vehicles is constructed.

[0147] The feature vector for identifying toll evasion of green channel vehicles is input into a preset toll evasion risk identification model for processing to obtain the probability value of toll evasion risk.

[0148] The probability value of evasion risk is compared with the preset probability threshold of evasion risk.

[0149] If the probability value of the toll evasion risk is less than or equal to the preset toll evasion risk probability threshold, it is determined to be a normal green channel vehicle;

[0150] If the probability value of toll evasion risk is greater than the preset toll evasion risk probability threshold, the vehicle is identified as an abnormal green channel vehicle, and a manual inspection response is activated.

[0151] It should be noted that, in order to achieve real-time identification of toll evasion risks by green channel vehicles, when a green channel vehicle enters the highway, the data collection equipment at the entrance toll station immediately acquires a photo of the vehicle's driver's license. During the vehicle's journey, the gantry system collects its driving trajectory data in real time. At the exit toll station, staff inspect the cargo from different angles (front, rear, sides, and top of the vehicle) and collect relevant photos or videos. First, a pre-set image tampering detection model based on deep learning (such as a tampering detection model based on generative adversarial networks) is used to analyze the denoised driver's license image, detect Photoshop traces, and generate a tampered driver's license image. The probability value is modified, and a threshold comparison is used to determine whether tampering has occurred. If no tampering is found, the curb weight and load capacity are further extracted from the denoised vehicle registration image and compared with the corresponding parameters in the preset vehicle management database to obtain the curb weight deviation rate and load capacity deviation rate. The preset image tampering detection model is trained by acquiring a large number of historical denoised vehicle registration images and their corresponding tampering probability values. The curb weight deviation rate is the ratio of the absolute value of the difference between the curb weight and the curb weight registration data in the vehicle management database to the curb weight registration data in the vehicle management database. The load capacity deviation rate... The rate refers to the ratio of the absolute value of the difference between the registered load capacity and the registered load capacity data in the vehicle management database to the registered load capacity data in the vehicle management database; the speed fluctuation coefficient refers to the ratio of the standard deviation of the driving speed within a preset journey (such as from toll station A to toll station B or from gantry A to gantry B) to the average driving speed; the number of sudden decelerations refers to the number of times the speed change rate exceeds 20 km / h / min (i.e., the acceleration or deceleration exceeds 20 km / h within 1 minute) exceeds 3 times within a preset time period (such as 1 hour) within a preset journey. Frequent sudden decelerations may be related to vehicles deliberately evading inspection. For example, when a vehicle approaches a toll station or inspection point, it may suddenly slow down to observe the strictness of the on-site inspection and take the opportunity to choose whether to use fake documents, disguise goods, or other means to evade tolls. The frequency of green channel round trips refers to the number of times the same vehicle license plate travels to and from the same pair of toll stations within a preset time period (such as one week) and applies for green channel access. The cargo consistency verification data includes whether the cargo is consistent or inconsistent. It is obtained by extracting features from cargo photos or videos, combining them with deep learning models (such as convolutional neural networks) to identify the cargo category, and comparing it with the declared category. The loading volume ratio refers to the ratio of the cargo loading volume to the vehicle's approved cargo volume.The obtained curb weight deviation rate, load capacity deviation rate, speed fluctuation coefficient, number of sudden decelerations, and frequency of green channel round trips to stations are integrated with cargo consistency verification data and loading volume ratio to form a feature vector. Each feature is assigned a different weight based on its impact on toll evasion risk (the weights are trained using historical toll evasion case data). This integrated green channel vehicle toll evasion identification feature vector is then input into a pre-set toll evasion risk identification model for processing to obtain a toll evasion risk probability value. The toll evasion risk is identified by threshold comparison, thus achieving the integration of vehicle basic information, driving trajectory, and multi-source cargo data to extract toll evasion risk features from multiple dimensions. Compared to traditional identification methods that rely solely on cargo information, this approach can more comprehensively and accurately identify toll evasion behavior and effectively reduce the false positive rate.

[0152] A third aspect of the present invention provides a readable storage medium storing a program for identifying toll evasion by high-speed green channel vehicles. When the program is executed by a processor, it implements the steps of the method for identifying toll evasion by high-speed green channel vehicles as described in any of the preceding claims.

[0153] This invention discloses a method, system, and medium for identifying toll evasion by vehicles traveling on highway green channels. By constructing image data analysis and core mining, it performs multi-dimensional analysis on the vehicle passage record data of green channel lanes, accurately identifies various toll evasion behaviors, and thus achieves accurate identification of toll evasion by vehicles traveling on highway green channels.

[0154] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0155] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0156] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0157] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0158] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for identifying toll evasion by vehicles using highway green channels, characterized in that, Includes the following steps: Obtain vehicle passage record data from green channel lanes and preprocess it to obtain optimized vehicle passage record data; Construct a rule set for identifying toll evasion by green channel vehicles; Based on the optimized data of the vehicle passage records and the rule set for identifying toll evasion of green channel vehicles, the status of suspected toll evading vehicles is obtained, including suspected toll evading vehicles or non-suspected toll evading vehicles. If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring is performed; If the suspected toll evasion vehicle is in the status of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle. The confidence level of evasion of payment is compared with a preset confidence threshold for evasion of payment. If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated. If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated using the preset green channel evasion identification template and sent to the terminal for display. The process of acquiring vehicle passage record data for green lanes and preprocessing it to obtain optimized vehicle passage record data includes: Obtain vehicle passage record data for green channel lanes, including vehicle license plate numbers, cargo category characteristic data, vehicle registration certificate images, and vehicle images; The vehicle license plate, cargo category feature data, vehicle registration certificate image and vehicle image are formatted, linked and cleaned preprocessed to obtain optimized vehicle passage record data, including standard vehicle license plate, standard cargo category feature data, denoised vehicle registration certificate image and denoised vehicle image. If the suspected toll evasion vehicle is classified as a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed using a preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle, including: The seasonal matching rate is obtained by analyzing and processing the cargo category characteristic data. Obtain the image recognition accuracy score of suspected toll evading vehicles, the matching degree score of green channel vehicle toll evasion recognition rules, and the data integrity score; The image recognition accuracy score, the matching score of the green channel vehicle toll evasion identification rule, and the data integrity score are weighted and summed, and then divided by the seasonal matching rate to obtain the toll evasion confidence of the suspected toll evading vehicle.

2. The method for identifying toll evasion by expressway green channel vehicles according to claim 1, characterized in that, Also includes: Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data; Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

3. The method for identifying toll evasion by expressway green channel vehicles according to claim 2, characterized in that, The rule set for identifying toll evasion by green channel vehicles includes: Construct a rule set for identifying toll evasion by green channel vehicles; The set of rules for identifying toll evasion by green channel vehicles includes rules for identifying vehicles with the same license plate but different driver's licenses, rules for identifying abnormal vehicle license marking sizes, rules for identifying vehicles with counterfeit license plates, rules for identifying abnormal passage records of trailers, and rules for identifying abnormal payment records of vehicles with the same license plate.

4. The method for identifying toll evasion by expressway green channel vehicles according to claim 3, characterized in that, The step of identifying toll evasion by optimizing the vehicle passage record data and combining it with the green channel vehicle toll evasion identification rule set to obtain the status of suspected toll evading vehicles includes: Based on the standard vehicle license plate, query the corresponding denoised vehicle license image, and process the vehicle license text data corresponding to the denoised vehicle license image by matching the recognition rules of different vehicle licenses with the same vehicle license plate to obtain the status of suspected toll evasion vehicles. The vehicle width data corresponding to the denoised vehicle license image is combined with the vehicle license text data to match the vehicle license mark size anomaly recognition rules and obtain the status of suspected toll evasion vehicles. Based on the standard vehicle license plate, query the corresponding denoised vehicle image's front outline image, headlight position data, and body color data, and process them according to the rules for identifying cloned vehicles to obtain the status of suspected toll-evading vehicles. Extract the trailer's license plate from the denoised vehicle image, query the corresponding first green channel payment record for the trailer, and process the trailer's abnormal passage record by matching the first green channel payment record with the trailer's abnormal passage record identification rules to obtain the status of the suspected toll evasion vehicle. Based on the standard vehicle license plate, query the second green channel payment record of the preset passage time sequence, and process it according to the anomaly identification rules of the payment record of the same vehicle license plate to obtain the status of suspected toll evasion vehicles.

5. A system for identifying toll evasion by vehicles using highway green channels, characterized in that, The system includes a memory and a processor. The memory contains a program for identifying toll evasion by vehicles using the expressway green channel. When the processor executes the program for identifying toll evasion by vehicles using the expressway green channel, it performs the following steps: Obtain vehicle passage record data from green channel lanes and preprocess it to obtain optimized vehicle passage record data; Construct a rule set for identifying toll evasion by green channel vehicles; Based on the optimized data of the vehicle passage records and the rule set for identifying toll evasion of green channel vehicles, the status of suspected toll evading vehicles is obtained, including suspected toll evading vehicles or non-suspected toll evading vehicles. If the suspected toll evasion vehicle status is changed to not suspected toll evasion vehicle, then normal monitoring is performed; If the suspected toll evasion vehicle is in the status of a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed according to the preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle. The confidence level of evasion is compared with a preset confidence threshold for evasion. If the confidence level of evasion is less than the preset confidence level threshold for evasion, a suspected list is generated and manual review is activated. If the confidence level of evasion is greater than or equal to the preset confidence level threshold for evasion, a green channel vehicle evasion identification report is generated using the preset green channel evasion identification template and sent to the terminal for display. The process of acquiring vehicle passage record data for green lanes and preprocessing it to obtain optimized vehicle passage record data includes: Obtain vehicle passage record data for green channel lanes, including vehicle license plate numbers, cargo category characteristic data, vehicle registration certificate images, and vehicle images; The vehicle license plate, cargo category feature data, vehicle registration certificate image and vehicle image are formatted, linked and cleaned preprocessed to obtain optimized vehicle passage record data, including standard vehicle license plate, standard cargo category feature data, denoised vehicle registration certificate image and denoised vehicle image. If the suspected toll evasion vehicle is classified as a suspected toll evasion vehicle, then the suspected toll evasion vehicle is analyzed and processed using a preset toll evasion confidence processing method to obtain the toll evasion confidence of the suspected toll evasion vehicle, including: The seasonal matching rate is obtained by analyzing and processing the cargo category characteristic data. Obtain the image recognition accuracy score of suspected toll evading vehicles, the matching degree score of green channel vehicle toll evasion recognition rules, and the data integrity score; The image recognition accuracy score, the matching score of the green channel vehicle toll evasion identification rule, and the data integrity score are weighted and summed, and then divided by the seasonal matching rate to obtain the toll evasion confidence of the suspected toll evading vehicle.

6. The highway green channel vehicle toll evasion identification system according to claim 5, characterized in that, Also includes: Data is extracted from the denoised vehicle registration certificate image to obtain vehicle width data and vehicle registration certificate text data; Data is extracted from the denoised vehicle image to obtain the vehicle front outline image, headlight position data, and vehicle body color data.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program for identifying toll evasion by high-speed green channel vehicles. When the program is executed by a processor, it implements the steps of a method for identifying toll evasion by high-speed green channel vehicles as described in any one of claims 1 to 4.

Images