CN120183056B

CN120183056B - Vehicle-related payment method, vehicle-related payment system and storage medium for passing on expressway

Info

Publication number: CN120183056B
Application number: CN202510596227.5A
Authority: CN
Inventors: 周之安
Original assignee: Welinkpark Shenzhen Technology Co ltd
Current assignee: Welinkpark Shenzhen Technology Co ltd
Priority date: 2025-05-09
Filing date: 2025-05-09
Publication date: 2025-08-01
Anticipated expiration: 2045-05-09
Also published as: CN120183056A

Abstract

The present application relates to payment system technology, and discloses a vehicle-related payment method, vehicle-related payment system, and storage medium for highway travel, including: based on a camera deployed at the highway entrance, collecting vehicle images of vehicles to be passed; using a pre-trained artificial intelligence model to identify license plate numbers; when it is detected that there is an associated vehicle with the license plate number in the vehicle-related payment system, the vehicle-related payment system notifies the barrier device at the highway entrance to lift the barrier and release the vehicle; when the camera deployed at the highway exit collects the vehicle image of the vehicle to be passed and identifies the corresponding license plate number, the vehicle-related payment system calls the billing model to calculate the highway toll; according to the payment method pre-signed by the vehicle associated with the license plate number in the vehicle-related payment system, the highway toll is automatically settled, and the barrier device at the highway exit is notified to lift the barrier and release the vehicle. The present application aims to provide a highway vehicle-related payment system based on license plate recognition to improve the efficiency of vehicle passage at highway entrances and exits.

Description

Vehicle-related payment method, vehicle-related payment system and storage medium for passing on expressway

Technical Field

The present application relates to the technical field of payment systems, and in particular, to a vehicle-related payment method, a vehicle-related payment system, and a computer-readable storage medium for passing on a highway.

Background

At present, the automatic passing mode of vehicles at the expressway entrance is mainly ETC (electronic toll collection) charging, which requires vehicle owners to install ETC vehicle-mounted equipment on the vehicles in advance, and the purchase and maintenance of the ETC vehicle-mounted equipment is an extra expenditure for the vast owners. Particularly for owners who occasionally get on the highway, the use frequency of ETC vehicle-mounted equipment is low, and the cost performance of the arrangement equipment is low, so that the cost of ETC traffic on the highway is particularly outstanding.

With the development of image recognition technology, the vehicle-related payment scene using license plate image recognition is widely applied in life, especially in a parking lot scene. However, the existing license plate image recognition technology still faces many challenges when applied to expressway scenes, on one hand, the running speed of vehicles on roads is often higher, motion blur is more easily caused when images are collected, on the other hand, the illumination conditions on the expressway are complex and changeable, huge interference is brought to license plate image recognition, the license plate can be in a strong light direct or shadow area (the strong light direct can lead to license plate reflection, image overexposure and character information loss are caused when the license plate is in the shadow area), the existing license plate image recognition technology is difficult to meet the accurate recognition requirement under the complex expressway scenes, and the automatic passing of the vehicles at the expressway entrances and exits is difficult to be stably and efficiently supported like an ETC system, so that the wide application of the license plate image recognition technology in the expressway vehicle payment field is limited.

The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art.

Disclosure of Invention

The application mainly aims to provide a vehicle-related payment method, a vehicle-related payment system and a computer-readable storage medium for passing on a highway, and aims to provide a vehicle-related payment system for the highway based on license plate recognition so as to improve the passing efficiency of vehicles at a highway entrance.

In order to achieve the above purpose, the application provides a vehicle-related payment method for passing on a highway, comprising the following steps:

based on a camera deployed at a high-speed entrance, acquiring a vehicle image of a vehicle to be passed;

The method comprises the steps of utilizing a pre-trained artificial intelligent model to identify license plate numbers corresponding to vehicle images, wherein the artificial intelligent model adopts a double-branch network, one branch is used for estimating motion blur parameters of the images, the other branch is used for estimating weights of different spectrum channels of the images, an interaction mechanism is introduced between two branch outputs to enable estimated outputs of each branch to act on characteristic images of other branches for adjustment, and in each branch, the estimated outputs of each branch and the characteristic images after adjustment are utilized to process original images respectively, and the images obtained by processing each branch are fused to generate fusion images after deblurring and spectrum optimization for license plate number identification;

when detecting that the identified license plate number has an associated vehicle in the vehicle-related payment system, the vehicle-related payment system informs a railing device at a high-speed entrance to lift and pass and records corresponding passing information;

When a camera deployed at a high-speed exit collects vehicle images of vehicles to be passed and identifies corresponding license plates by utilizing the artificial intelligent model, the vehicle-related payment system calls a charging model provided by a high-speed management system, and calculates high-speed passing fees according to the passing information of the currently identified license plates at the high-speed entrance and the passing information of the currently high-speed exit;

According to a pre-signed payment mode of a vehicle associated with a license plate number in a vehicle-related payment system, automatically settling high-speed passing fees;

After the high-speed toll settlement is successful, the vehicle-related payment system informs a railing device at the high-speed exit to lift and release.

In order to achieve the aim, the application also provides a vehicle-related payment system which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the computer program is executed by the processor to realize the steps of the vehicle-related payment method for passing on a highway.

To achieve the above object, the present application also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the vehicle-related payment method for passing on a highway as described above.

The vehicle-related payment method, the vehicle-related payment system and the computer-readable storage medium for passing on the expressway adopt a special double-branch network pre-training artificial intelligent model, even if the image motion is fuzzy due to the fact that a vehicle runs at an expressway or the illumination condition on the expressway is complex, a clear and accurate fusion image can be generated for license plate number identification, the accurate identification requirement under the complex scene of the expressway is met, and once the identification is successful, the vehicle-related payment system can rapidly respond, automatically settle and release the cost according to the pre-signed payment mode of the vehicle, manual intervention is not needed, vehicle waiting time is reduced, high-efficiency and convenient high-speed passing payment experience is provided for a vehicle owner, the whole vehicle-related payment passing process does not need to be provided with ETC (electronic toll collection) and other equipment for passing at the expressway in advance, and the cost of the vehicle passing on the expressway is reduced.

Drawings

FIG. 1 is a schematic diagram showing steps of a method for payment for vehicle-related traffic on a highway according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a dual-branch network of an artificial intelligence model according to an embodiment of the present application;

Fig. 3 is a schematic diagram of an internal architecture of a wading payment system according to an embodiment of the application.

The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below are exemplary and intended to illustrate the present application and should not be construed as limiting the application, and all other embodiments, based on the embodiments of the present application, which may be obtained by persons of ordinary skill in the art without inventive effort, are within the scope of the present application.

Furthermore, references to "first," "second," etc. in this disclosure are for descriptive purposes only (e.g., to distinguish between the same or similar features), and are not to be construed as indicating or implying a relative importance or implying any particular order of magnitude of the features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions of the embodiments may be combined with each other, but it is necessary to base that the technical solutions can be realized by those skilled in the art, and when the technical solutions are contradictory or cannot be realized, the combination of the technical solutions should be considered to be absent and not within the scope of protection claimed in the present application.

Referring to fig. 1, in an embodiment, a vehicle-related payment method for passing on a highway includes:

step S10, acquiring a vehicle image of a vehicle to be passed based on a camera deployed at a high-speed entrance;

S20, identifying a license plate number corresponding to a vehicle image by utilizing a pre-trained artificial intelligent model, wherein the artificial intelligent model adopts a double-branch network, one branch is used for estimating motion blur parameters of the image, the other branch is used for estimating weights of different spectrum channels of the image, an interaction mechanism is introduced between two branch outputs, so that the estimated output of each branch acts on the characteristic diagram adjustment of other branches;

Step S30, when the recognized license plate number is detected to have an associated vehicle in the vehicle-related payment system, the vehicle-related payment system informs a railing device at a high-speed entrance to lift and release and records corresponding traffic information;

Step S40, when a camera deployed at a high-speed exit collects vehicle images of vehicles to be passed and identifies corresponding license plates by utilizing the artificial intelligent model, calling a charging model provided by a high-speed management system by a vehicle-related payment system, and calculating high-speed passing fees according to the passing information of the currently identified license plates at the high-speed entrance and the passing information of the currently high-speed exit;

Step S50, automatically settling high-speed toll according to a pre-signed payment mode of a vehicle associated with a license plate number in a vehicle-related payment system;

And S60, after the high-speed toll settlement is successful, the vehicle-related payment system informs a railing device at the high-speed exit to lift and release.

In this embodiment, the terminal system for executing the embodiment may be a vehicle-related payment system, or may be other devices or apparatuses (such as a control apparatus) for controlling the vehicle-related payment system.

As described in step S10, the entrance, exit and section charging points of the expressway have usually been pre-deployed with corresponding cameras. The system can establish communication connection with cameras existing on the highways. Therefore, the existing hardware equipment resources can be fully utilized, and the cost for accessing the existing system is greatly reduced.

Alternatively, the camera is typically mounted in place above or to the side of the high speed doorway lane. When the vehicle head is arranged above, the vehicle head can be shot at an angle obliquely above the vehicle head to obtain a vehicle image comprising the license plate, and when the vehicle head is arranged on the side, the vehicle image comprising the license plate can be shot from the oblique side of the vehicle head.

Optionally, a vehicle sensor, such as a ground coil or an infrared sensor, is installed on the high speed entrance lane. When a vehicle enters the sensing area, the sensor detects the existence of the vehicle and immediately sends a trigger signal to the camera to start an image acquisition program. The camera can also perform image acquisition in a continuous video stream mode, then detect the entry of the vehicle in real time through an image processing algorithm, and intercept image frames containing the vehicle from the video stream once the vehicle is detected to enter the shooting range.

And as shown in step S20, the pre-trained artificial intelligent model is utilized to process the vehicle image acquired by the camera at the high-speed entrance, so that the corresponding license plate number is accurately identified. In order to cope with the problems of motion blur of images and spectrum differences under different illumination conditions possibly caused during the running process of a vehicle, referring to fig. 2, the artificial intelligence model adopts a unique dual-branch network architecture, wherein one branch is a motion blur estimation branch, and the other branch is a spectrum channel estimation branch.

During the running of the vehicle, the camera is likely to generate motion blur when capturing images because it is in motion. The main task of the motion blur estimation branch is to analyze the vehicle image and estimate the motion blur parameters of the image due to motion. These parameters include the direction of the blur (e.g., horizontal, vertical, or oblique directions) and the degree of blur (i.e., the length of the blur). The motion blur estimation branch may perform depth analysis on local features of the image (such as edge sharpness, texture continuity, etc.), and determine motion blur parameters of the current image by comparing with image features of different blur degrees and directions stored in the pre-trained model.

Different lighting conditions (e.g., sunny, cloudy, backlight, etc.) may cause the image to appear differently on the various spectral channels (e.g., red, green, blue channels). The function of the spectrum channel estimation branch is to estimate the weights of different spectrum channels of the image so as to optimize the display effect of the image under different spectrums. The spectrum channel estimation branch can analyze the color distribution, brightness and other characteristics of the image, and allocate proper weights for each spectrum channel of the current image according to the weight allocation rules of the spectrum channels under different lighting conditions learned in the pre-training model.

In order to enable the information of the two branches to be better fused, the processing effect of the image is improved, and an interaction mechanism is introduced between the output of the two branches. This mechanism allows the estimated outputs of the two branches not to be independent of each other, but to be mutually influenced and adjusted.

Optionally, the output of the motion blur estimation branch may be applied to the feature map of the spectral channel estimation branch, and adjusted. For example, if the motion blur parameter indicates that the image has stronger blur in a certain direction, the feature map is corrected accordingly in consideration of the influence of the blur on color perception when adjusting the spectral channel weight. Conversely, the output of the spectral channel estimation branch will also act on the feature map of the motion blur estimation branch in the same way.

Optionally, the artificial intelligence model encodes the estimated output of the motion blur estimation branch to generate a weight matrix, and performs weighted fusion on the weight matrix and the first feature map of the spectrum channel estimation branch to obtain an adjusted first feature map.

Optionally, the artificial intelligence model performs dot product operation on the estimated output of the spectrum channel estimation branch and the second feature map of the motion blur estimation branch to obtain tensors with the same shape as the second feature map;

Carrying out normalization processing on tensors to obtain attention weights;

And multiplying the attention weight with the second feature map element by element in the space dimension to obtain an adjusted second feature map.

Optionally, deblurring is performed on the original image according to the adjusted second feature map and in combination with the previously estimated motion blur parameters, so as to attempt to recover the detail information lost by the image due to motion blur and spectral influence, and make key information such as license plates in the image clearer. The following is the process of generating a deblurred clear image:

The motion blur kernel is constructed from the motion blur parameters (blur direction and blur length) of the motion blur estimation branch output. The blur kernel is a two-dimensional matrix whose size is mainly determined by the blur length and whose shape is related to the blur direction. In the construction process, a zero matrix with a specific size is firstly created, non-zero elements are arranged on the matrix according to the fuzzy direction and the length, and finally normalization processing is carried out on the matrix, so that the sum of all elements is ensured to be 1. This blur kernel simulates the process of generating motion blur for the image and is a key tool for subsequent deblurring operations.

And then, comprehensively considering the noise conditions in the fuzzy core and the image by adopting wiener filtering, and recovering the original clear image by a mode of minimizing the mean square error. And during processing, the adjusted second characteristic diagram is used as guiding information and is merged into the filtering process. According to the characteristic intensity of different areas in the second characteristic diagram, the filtering parameters are dynamically adjusted, so that finer deblurring operation is performed in the areas with obvious characteristics, and a relatively conservative filtering mode is adopted in the areas with weaker characteristics.

Specifically, the noise condition in the original image F ₀ is estimated to obtain a noise power spectrum P _n (u, v), the original image F ₀ is subjected to two-dimensional Discrete Fourier Transform (DFT) to obtain a frequency domain representation G (u, v), the blur kernel is also subjected to two-dimensional discrete fourier transform to obtain H (u, v), and the adjusted second feature map F ₂ is subjected to two-dimensional discrete fourier transform to obtain F ₂' (u, v). Then, a wiener filter function is calculated, and the calculation formula of the wiener filter function is as follows:

;

Wherein H ^* (u, v) is the conjugate complex number of H (u, v), and P _f (u, v) is the power spectrum of the original clear image, and the model is reasonably estimated according to priori knowledge obtained by pre-training because the original clear image is unknown. In the training stage of the model, a large number of clear images and corresponding blurred images are used for training the model, and the model can learn the power distribution rules of different types of images on different frequencies and the influence of image blurring and noise on a power spectrum through learning the data. For example, the model may find similar image samples from training data based on some characteristics of the blurred image (e.g., average brightness, contrast, texture complexity, etc.), and then use the power spectrum of these samples as an estimate of the original sharp image power spectrum of the current blurred image. By the estimation method based on priori knowledge, the model can reasonably estimate P _f (u, v) under the condition that the original clear image is unknown, so that effective wiener filtering deblurring processing is realized.

In order to dynamically adjust the filter parameters according to the feature intensities of the different regions in the second feature map, an adjustment factor α (u, v) is predefined to calculate α (u, v) from the magnitude of F ₂' (u, v):

;

blending the adjustment factor into the wiener filter function to obtain an adjusted wiener filter function:

;

Where k >1 is a constant used to control the use of relatively conservative filtering in the weaker regions of the feature, and its specific value can also be determined based on a priori knowledge learned in advance by the model. When alpha (u, v) is close to 1 (the characteristic obvious region), the filter function is close to the original wiener filter function, and finer deblurring operation is carried out, and when alpha (u, v) is close to 0 (the characteristic weaker region), the denominator in the filter function is increased, and the filter effect is relatively conservative.

The adjusted wiener filter function w ₂ (u, v) is multiplied with the frequency domain representation G (u, v) of the original image to obtain the frequency domain representation F ₀' (u, v) of the restored image. And performing two-dimensional Inverse Discrete Fourier Transform (IDFT) on F ₀' (u, v) to obtain a recovered clear image.

It should be noted that, due to factors such as aberration of the optical system, atmospheric scattering, etc., images of different spectral channels may have different degrees of blurring. Some blurring phenomena may be directly related to spectral characteristics (such as chromatic aberration blurring), chromatic aberration may cause images of different spectral channels to shift spatially, so that the images become blurred, and by adjusting the weights of the spectral channels and combining with a deblurring algorithm, the blur related to the spectrum may be compensated, so that clear details of the images may be recovered more accurately.

When deblurring algorithms such as wiener filtering are adopted, the spectrum channel weight can be used as guiding information for dynamically adjusting filtering parameters, so that finer deblurring operation is performed on spectrum channels with obvious characteristics, and a relatively conservative filtering mode is adopted on channels with weaker characteristics, thereby improving the deblurring effect.

After the adjusted first feature map is obtained, the original image can be processed by using the estimated output of the spectrum channel estimation branch and the first feature map according to the following scheme to obtain a spectrum optimization map with motion blur influence removed:

Because the spectral channel estimation branch has analyzed the color distribution, brightness and other characteristics of the image, the spectral channels (such as red, green and blue channels) of the current image can be allocated with proper weights according to the weight allocation rules of the spectral channels under different illumination conditions learned in the pre-training model. Let the output of the spectral channel estimation branch be W _s=[W_r,W_g,W_b ], where W _r、W_g、W_b is the weight of the red, green, and blue channels, respectively.

The adjusted first feature map F ₁ contains information of the motion blur parameters after adjustment of the spectral channel estimation branch feature map, and the adjusted first feature map F ₁ can be combined with the spectral channel weight W _s to highlight or suppress the features of different spectral channels. Assuming that F ₁ is split into feature maps F _r、F_g、F_b of the three color channels, the weighted feature maps of the three color channels are respectively F_r'=W_r×F_r、F_g'=W_g×F_g、F_b'=W_b×F_b.

Let the original image F ₀ be split into the three-color channel image I _r、I_g、I_b as well, apply the weighted feature map of the three-color channel to the corresponding channel of the original image to obtain the adjusted images I_r'=I_r+β×F_r'、I_g'=I_g+β×F_g'、I_b'=I_b+β×F_b', of each channel respectively, wherein β is an adjustment coefficient used to control the influence degree of the weighted feature map on the original image, and the optimal value of β can be learned in advance in the model training process.

And recombining the three adjusted channel images I _r'、I_g'、I_b' into a complete image, so as to obtain a spectrum optimization diagram after weakening the influence of motion blur.

After the motion blur estimation branch processing is carried out to obtain a clear image after deblurring, and the spectrum channel estimation branch processing is carried out to obtain a spectrum optimization diagram, the images obtained by the two branch processing are fused. The purpose of fusion is to integrate the advantages of the two branch processes to generate a fused image that both removes motion blur and optimizes spectrum.

The pixel values of the fusion image can be obtained by carrying out weighted summation on the values of the corresponding pixel points according to weights by adopting a weighted average method.

Since the motion blur has a greater influence on the accuracy of license plate number recognition than the spectral channel weight optimization, the first weight of the motion blur estimation branch is set to be greater than the second weight of the spectral channel estimation branch (and the sum of the first weight and the second weight is 1). On the premise, the corresponding weights (which can use image quality evaluation indexes (such as peak signal to noise ratio (PSNR), structural Similarity Index (SSIM), etc.) to evaluate the reliability, and the higher the reliability, the greater the weights) can be adjusted according to the reliability of the two branch processing results to evaluate the quality of the two images, and the weights are dynamically allocated according to the evaluation results (the final allocation result still needs to satisfy the first weight being greater than the second weight, and the sum of the two weights is 1).

The conventional processes of deblurring and spectrum optimization for license plate images are often independent of each other, such as deblurring and spectrum optimization, or spectrum optimization and deblurring, and the mutual influence of deblurring and spectrum optimization is ignored. If deblurring processing is performed first, in the process, the blurred image may cause inaccurate color and illumination information, and when spectral optimization is performed later, the image features may be damaged to some extent due to earlier deblurring, so that the optimized image color may be unnatural and have problems such as color cast and the like due to the fact that the earlier deblurring may be difficult to adjust according to the accurate color and illumination features, otherwise, if spectral optimization is performed first, the blurred image may cause the optimization process to fail to accurately judge the actual color distribution and illumination conditions, so that the spectral optimization may be excessive or insufficient, and then when deblurring is performed, the characteristic change caused by the previous spectral optimization affects the recognition of the image edges, textures and other features by a deblurring algorithm, so that the deblurring effect is poor, and the image still has a blurred feeling or has artifacts.

The artificial intelligent model of the application adopts an interaction mechanism introduced by a double-branch network, so that the motion blur estimation branch and the spectrum channel estimation branch can share the characteristic information extracted from each other, thereby avoiding the occurrence of the problems. For example, the features such as image edges and textures extracted by the motion blur estimation branches can provide references for spectral channel weight estimation to help the spectral channel weight estimation to better analyze the color distribution and illumination conditions of the image, and the color features extracted by the spectral channel estimation branches can assist in motion blur parameter estimation to improve the accuracy of judging the blur direction and degree.

After the generated fusion image with deblurring and spectrum optimization, a character recognition algorithm based on deep learning can be adopted to recognize the fusion image so as to recognize the corresponding license plate number. The algorithm can locate and divide the license plate region in the fused image, divide the characters on the license plate into single characters, and then classify and identify each character to finally obtain the complete license plate number.

As described in step S30, when the license plate number corresponding to the vehicle image is successfully identified by using the pre-trained artificial intelligence model, the vehicle-related payment system will immediately compare the identified license plate number with the associated information stored in the system. The vehicle-related payment system stores license plate number information of a series of contracted vehicles (comprising vehicles contracted for online payment and vehicles contracted for pre-payment of passing fees (namely prepaid vehicles)), and owners of the vehicles have a cooperation agreement with the vehicle-related payment system in advance and enter own vehicle information into the system.

If the identified license plate number is detected to have an associated vehicle in the wading payment system, the method means that the owner of the vehicle has completed the necessary signing flow and accords with the payment condition of high-speed passing. At this time, the vehicle-related payment system can rapidly send a lever lifting instruction to the railing device at the high-speed entrance. After receiving the instruction, the railing device can immediately execute a lifting action to allow the vehicles to be passed to smoothly enter the expressway.

The vehicle-related payment system can record the traffic information of the vehicle in detail while informing the railing device to lift the bar for release. Such traffic information includes, but is not limited to, the number of the vehicle's license plate, the time of entry into the highway, the specific entry location, etc. These recorded information will be stored in the database of the vehicle-related payment system as an important basis for the subsequent calculation of high-speed tolls and for the inquiry of the vehicle's passage history.

If the identified license plate number is detected as not having an associated vehicle in the wading payment system, the vehicle may not sign up with the wading payment system. In this case, the vehicle-related payment system may trigger a corresponding prompt mechanism, for example, to display a prompt message on a display screen at the high-speed entrance to guide the vehicle owner to sign up, or the railing device may not lift the vehicle to release, so that the vehicle owner is required to acquire the right of way by adopting other traditional ways of way (such as real card way, ETC).

When the vehicle travels to the exit of the highway, the camera installed at a specific position at the exit immediately starts the operation as shown in step S40. These cameras are carefully deployed to ensure that an image of the vehicle to be passed, particularly the license plate portion, is captured in a comprehensive and clear manner.

After the vehicle image is acquired, the system immediately adopts the license plate recognition mode in the step S20, calls a pre-trained artificial intelligent model, and carries out license plate recognition on the vehicle image.

After the vehicle-related payment system successfully identifies the license plate number of the vehicle at the high-speed exit, the system database can quickly search the traffic information at the high-speed entrance corresponding to the license plate number. These entry traffic information have been accurately recorded as the vehicle enters the highway, including the specific time the vehicle entered the highway, the time in minutes seconds, and the specific entry location, such as a particular toll gate name or number.

Meanwhile, the vehicle-related payment system can call a charging model of the high-speed management system, and the charging model is carefully designed by comprehensively considering various factors through long-term research and practice of relevant engineers of the high-speed management system, and aims to scientifically and reasonably calculate the high-speed passing cost of the vehicle based on a large amount of traffic data, cost analysis and policy regulation.

The vehicle-related payment system integrates the entry traffic information recorded before and the exit traffic information acquired currently. The exit traffic information also contains the exact departure time and the specific exit location. The information is a basic element for calculating the toll, and provides a key basis for determining the driving path and time range of the vehicle.

In addition, in some specific sections of the expressway, an expressway section charging point is set. These charging points are also equipped with cameras with the same function as at the entrance and exit. When the vehicle passes through the charging points, the camera automatically collects the vehicle image and utilizes the same artificial intelligent model to identify the license plate number. If the image of the corresponding vehicle is acquired at the high-speed interval charging point and the license plate number is successfully identified, the traffic information (including the specific time when the vehicle passes through the charging point and the accurate position of the charging point) generated by the vehicle at the high-speed interval charging point is related to the corresponding license plate number.

When the vehicle-related payment system calls the charging model to calculate the high-speed toll of the vehicle, if the corresponding traffic information is generated at the high-speed interval charging points based on the license plate number, the vehicle-related payment system can acquire the traffic information of the vehicle at the high-speed interval charging points and take the traffic information as a calculation factor to be included in the calculation of the high-speed toll.

Then, the called charging model combines the entrance and exit traffic information of the vehicle, comprehensively considers the factors such as the driving mileage, the driving time, the charging standard of different road sections (such as a high-speed section charging point), and the like, and accurately calculates the high-speed traffic cost of the vehicle. For example, if the vehicle passes through a plurality of high-speed section charging points, the charging model calculates the cost according to the charging standard of each section, and then adds up the total cost.

As described in step S50, when the vehicle-related payment system accurately calculates the high-speed toll of the vehicle, the system will immediately perform accurate information matching and searching operations in a huge database according to the identified license plate number. The database stores a large amount of relevant information of vehicles, and each license plate number corresponds to a detailed file which contains key information such as a payment mode of the vehicles signed in advance with the vehicle-related payment system. The system can quickly locate the record corresponding to the license plate number in a short time by utilizing an efficient index algorithm.

Once the corresponding record is found, the system confirms the pre-signed payment mode of the vehicle and automatically settles the corresponding high-speed toll according to the pre-signed payment mode.

The pre-signed payment mode may be online payment (such as using digital rmb, third party paymate, bank card binding, etc.), prepayment (pre-paying a certain amount), etc.

In step S60, if the vehicle payment system successfully and automatically settles the corresponding high-speed toll, a lever raising instruction is sent to the rail device at the high-speed exit. After receiving the lever lifting instruction, the control module of the railing device can drive a motor or other actuating mechanisms to lift the railing. Meanwhile, the railing device feeds back execution result information to the wading payment system, and the information indicates that the lever lifting operation is completed.

And after receiving the feedback of the execution result of the railing device, the vehicle-related payment system confirms that the vehicle is released. The system can record the release time and related operation information of the vehicle, and complete the whole high-speed toll settlement and release flow. Meanwhile, the records can be stored and analyzed as important business data for subsequent operation management and statistical report generation.

If the vehicle-related payment system fails to successfully and automatically settle the corresponding high-speed toll, outputting corresponding prompt information through a display screen of a high-speed outlet or associated equipment (such as vehicle-mounted equipment, mobile equipment and the like) of a vehicle according to the reason of settlement failure so as to prompt a vehicle owner to complement the amount of money or manually settle the toll.

In one embodiment, a special double-branch network pre-training artificial intelligent model is adopted, even if the image motion is fuzzy due to high-speed running of a vehicle or the illumination condition on a highway is complex, a clear and accurate fusion image can be generated for license plate number identification, so that the accurate identification requirement under a complex expressway scene is met, once the identification is successful, a vehicle-related payment system can rapidly respond, and automatically settle and release the cost according to the pre-signed payment mode of the vehicle, manual intervention is not needed, the waiting time of the vehicle is reduced, the high-efficiency and convenient high-speed passing payment experience is provided for a vehicle owner, in addition, the whole vehicle-related payment passing process is not needed, ETC and other equipment for high-speed passing billing is not needed, the cost of passing the vehicle on the expressway is reduced, and the vehicle-related payment system is more economical and practical especially for the vehicle owner with low expressway passing frequency.

In an embodiment, when the wiener filtering is performed in step S20, the power spectrum P _f (u, v) of the original clear image used for performing the wiener filtering is estimated based on a priori knowledge obtained by model pre-training, and a certain error may exist in the estimation. The recovered clear image obtained by the subsequent processing is obtained by one-time deblurring processing, and the power spectrum can more accurately reflect the characteristic distribution condition of the original image in the frequency domain.

Therefore, after the primary image is deblurred to obtain a clear image, the power spectrum P _f (u, v) of the primary clear image in the wiener filter functions w ₁ (u, v) and w ₂ (u, v) can be updated by utilizing the power spectrum of the clear image, and on the basis, the deblurring process is carried out again on the primary image, so that the deblurring effect is further optimized, and the clear image with better quality is obtained.

However, the wiener filtering needs to be performed again, so that a series of calculation is needed, and the calculation complexity is inevitably increased. Therefore, to avoid the problems that may be faced when the restored clear image power spectrum is used to replace the original P _f (u, v) for re-filtering, the power spectrum of the finally generated clear image may be compared with the P _f (u, v) used by the original wiener filtering function to determine whether there is a need to update the power spectrum and regenerate the clear image.

Optionally, after the recovered sharp image is obtained, it is subjected to a two-dimensional Discrete Fourier Transform (DFT) to transform the image from the spatial domain to the frequency domain. Let the restored sharp image be I _rec, obtain its frequency domain representation I _rec (u, v) after DFT, then calculate its power spectrum:

。

To effectively compare P _rec (u, v) with the original P _f (u, v), the mean and standard deviation of the two may be calculated separately, then the mean difference between the two is compared with a first threshold (preset mean difference threshold), and the difference of the standard deviation between the two is compared with a second threshold (preset standard difference threshold). If the difference of the mean values of the two is larger than a first threshold value, the difference of the mean values of the two is obvious, and if the difference of the standard deviation of the two is larger than a second threshold value, the difference of the discrete degrees of the two is obvious.

Alternatively, if both indices indicate that P _rec (u, v) differs significantly from P _f (u, v), indicating that the recovered sharp image power spectrum is significantly different from the original estimated power spectrum, then re-wiener filtering using P _rec (u, v) instead of P _f (u, v) can greatly improve the deblurring effect, and can be considered to regenerate the sharp image. If at least one index indicates that the difference between P _rec (u, v) and P _f (u, v) is not significant, it indicates that the original estimated P _f (u, v) can better reflect the frequency domain characteristics of the image, and the re-filtering may not bring about obvious improvement, or even introduce a new problem, and at this time, it is not recommended to use P _rec (u, v) to replace P _f (u, v) to re-perform wiener filtering.

In an embodiment, based on the above embodiment, the k value in the wiener filter function w ₂ (u, v) adjusted in step S20 may be dynamically adjusted according to the local feature of the image. Image characteristics (e.g., texture complexity, noise level, signal strength, etc.) may be different for different regions, so the filtering mode may be more precisely controlled using different k values for different regions.

Alternatively, the image to be processed is divided into a plurality of small blocks, and for each small block, its characteristic parameters such as variance (reflecting texture complexity), noise estimation value, and the like are calculated. And establishing a mapping relation between the k value and the characteristic parameter according to the characteristic parameter. For example, a regression model (which stores the mapping relationship between the k value and the feature parameter) may be trained in advance, and the feature parameter is input and the k value is output. In the filtering process, the k value is adjusted in real time according to the characteristic parameters of each small block.

The self-adaptive k value adjusting method can dynamically adjust the k value according to the local characteristics of the image, so that good filtering effects can be obtained in different areas.

In an embodiment, based on the foregoing embodiment, the step of automatically settling the high-speed toll according to a payment mode of a vehicle associated with a license plate number signed up in advance in a wading payment system includes:

If the pre-signed payment mode of the license plate number-related vehicle in the vehicle-related payment system is online payment, the corresponding online payment mode is adopted to automatically settle the high-speed toll.

In this embodiment, the vehicle-related payment system queries a database for a payment method for a pre-signed vehicle according to the identified license plate number. If the query result shows online payment, the system further obtains detailed information of the subscription account.

Different online payment types, the information extracted by the system is different:

(1) And the digital RMB account is used for extracting the identification of the digital RMB wallet, the information of the operating mechanism and the like.

(2) And the third party payment platform account (such as WeChat and Payment treasury) is used for acquiring the information such as the bound third party payment account number, platform identification and the like.

(3) The bank card binds the account by reading the information of the bank card number (a part of the possible desensitization process), the issuer, the card type (debit or credit card), etc.

Optionally, for digital rmb payment, the wading payment system establishes connection with a payment interface of the digital rmb operation mechanism, and packages necessary data such as payment amount, order number, license plate number, digital rmb wallet information, etc. into a payment request to be sent to the operation mechanism. After the digital RMB operation mechanism receives the payment request, the validity of the request, the balance of the wallet and the like are verified. If the verification is passed, the operation mechanism can deduct corresponding amount from the digital RMB wallet of the vehicle owner to the high-speed charging account.

Optionally, for payment of the third party payment platform, the system invokes a payment interface of the third party payment platform and transmits information such as payment amount, order number, license plate number, and bound third party payment account number to the payment platform. The third party paymate verifies the accuracy of the payment information and the available balance or credit of the account. If the payment condition is met, the platform deducts the corresponding fee from the user account and settles the funds to the designated account for the high-speed charge.

Optionally, for bank card payment, the wading payment system communicates with the bank system through a payment gateway and sends a payment instruction containing contents such as a bank card number, a payment amount, an order number and the like. The bank system performs various verification on the validity, balance, password (part of transaction needs) and the like of the bank card. After the verification is passed, the bank deducts corresponding high-speed toll from the bank card account and settles funds to the appointed charging account.

In either online payment mode, after the payment is completed, the payment mechanism (the digital RMB operation mechanism, the third party payment platform or the bank) feeds back the payment result (success or failure) to the vehicle-related payment system.

If the payment is successful, the vehicle-related payment system records detailed information of the successful payment, such as payment time, payment amount, payment mode, transaction serial number, etc. And meanwhile, updating the passing record of the vehicle, marking that the cost is clear, and controlling the railing to lift up to allow the vehicle to pass.

If the payment fails, the vehicle-related payment system immediately marks the transaction as a failure state and records the failure reason. Meanwhile, the vehicle is forbidden to pass, and corresponding prompt information such as 'insufficient balance', prompt to recharge or replace a payment mode ',' failure of a payment system ', prompt to wait or contact staff' and the like are output according to failure reasons.

Therefore, the vehicle owner does not need to park in a toll station for changing or carry out complicated cash transaction, and the vehicle can quickly pass through the toll station, so that the passing time is greatly saved, the vehicle passing efficiency at a high-speed intersection is improved, and the traffic jam is reduced. Meanwhile, the whole process of online payment operation is automatic, manual intervention is not needed, a vehicle owner only needs to sign up in advance, and subsequent toll is automatically deducted, so that travel is greatly facilitated.

In an embodiment, on the basis of the foregoing embodiment, the step of automatically settling the high-speed toll fee according to a payment mode of a vehicle associated with a license plate number signed up in advance in a wading payment system includes:

if the pre-signed payment mode of the license plate number-associated vehicle in the vehicle-related payment system is prepaid, corresponding high-speed toll is deducted from the prepaid amount;

if the prepaid amount is deducted by the full amount, the high-speed toll settlement is judged to be successful, and if the prepaid amount has an excessive part, the excessive amount is returned in an original way;

If the prepaid amount is not deducted by the full amount, outputting prompt information of the vehicle owner to manually pay the residual amount.

In this embodiment, the vehicle-related payment system searches the prepaid account information of the vehicle pre-signed in the database according to the identified license plate number, and obtains the current prepaid amount. The system compares the calculated high speed transit fee with the prepaid amount and attempts to deduct the corresponding fee from the prepaid amount.

If the prepaid amount is enough to pay the high-speed toll, the vehicle-related payment system judges that the settlement of the high-speed toll is successful. If the prepaid amount has excess after deducting the fee, the system will return the excess in the original way according to the pre-signed payment channel. For example, if the owner of the vehicle is prepaid through a bank card, the excess will be returned to the bank card, and if prepaid through a third party paymate, the excess will be returned to the corresponding third party payment account. Meanwhile, the system can record information such as the amount of rollback, time, transaction serial number and the like.

When the prepaid amount is insufficient to pay the high-speed toll, the vehicle-related payment system can immediately output prompt information. The prompt message can be displayed in a striking text through a display screen at the high-speed exit, and simultaneously, the voice broadcasting device informs the owner that the content is approximately 'insufficient prepayment amount of you, please pay the rest X-ray toll manually'.

At this time, the owner needs to manually select other payment modes according to the prompt to finish the payment of the residual amount, and after the payment is successful, the system can update the pass record and release the vehicle.

Therefore, the car owners only need to prepay fees in advance, pay fees automatically in passing, stop and pay fees are not needed, time is saved, passing efficiency of the high-speed intersection is improved, and congestion is reduced.

In an embodiment, on the basis of the foregoing embodiment, before the step of deducting the corresponding high-speed toll from the prepaid amount if the payment mode of the license plate number-related vehicle signed up in advance by the wading payment system is prepaid, the method further includes:

when the vehicle-related payment system receives preset passing information sent by the associated equipment of the vehicle with the prepaid payment mode, the charging model is called, and corresponding estimated passing cost is calculated;

And sending prepaid information to the associated equipment of the corresponding vehicle according to the estimated toll so as to prompt the owner to pay corresponding amount in advance.

In this embodiment, when the association device (such as the mobile phone APP of the owner, the vehicle-mounted intelligent terminal, etc.) of the vehicle with the prepaid payment mode sends the predetermined traffic information to the vehicle-related payment system, the system captures the information in time. The predetermined traffic information may contain key contents of a start place, a destination, a predicted traffic time, a vehicle type, etc. where the vehicle is planned to travel.

The vehicle-related payment system calls a charging model provided by the high-speed management system according to the received preset traffic information, and accurately calculates estimated traffic cost of the preset traffic of the vehicle by combining various parameters in the preset traffic information, such as driving mileage, charging standards corresponding to vehicle types and the like.

And the vehicle-related payment system generates prepaid information according to the calculated estimated toll. The information can comprise important contents such as specific amount of estimated toll, payment mode guidance, payment deadline and the like. The system then sends the prepaid information to the associated device of the corresponding vehicle to prompt the vehicle owner to prepay for the corresponding amount. After receiving the prepaid information, the vehicle owner can complete prepaid operation according to the prompt, and is ready for subsequent automatic settlement of high-speed toll.

Therefore, the car owner predicts the estimated passing cost in advance and finishes the prepayment, so that the planning of the outgoing line is clearer, the complicated operation of cash or other payment modes during passing is avoided, and the time is saved. The toll collection system and the high-speed management party can lock the toll in advance, ensure smooth toll collection and reduce toll disputes. In addition, the advanced payment is beneficial to optimizing fund management, improving the operation efficiency of the whole high-speed charging system, and enabling the high-speed passing experience to be smoother and more convenient.

In an embodiment, on the basis of the above embodiment, the predetermined traffic information includes any one of the following:

the association equipment sends the selected information of the expressway entrance before the vehicle enters the expressway;

The associated device sends selected information of the high-speed exit after the vehicle enters the expressway, wherein the estimated traffic Fei Yongshi is calculated, and the traffic information of the vehicle at the high-speed entrance is also called;

The associated equipment sends notification information when the vehicle arrives at the high-speed payment square, wherein the calculation of estimated traffic Fei Yongshi also calls traffic information of the vehicle at the high-speed entrance and high-speed exit information associated with the high-speed payment square;

and the association equipment sends a navigation planning route before the vehicle enters the expressway, wherein when the vehicle-related payment system receives the navigation planning route, the corresponding expressway traffic track is extracted to calculate estimated traffic cost.

In this embodiment, before the vehicle owner goes out or enters the expressway, the expressway may be selected through an association device (such as a mobile phone APP) before the vehicle enters the expressway. At this time, after the vehicle-related payment system receives the information, a corresponding charging model is called, and the estimated toll is calculated rapidly by combining the selected entrance and exit.

Or after the vehicle has entered the highway, the owner may select the highway exit through the associated device en route to a service station, rest area, etc. At this time, the system, besides obtaining the exit selection information, also calls the traffic information (such as entrance time, entrance toll gate number, etc.) recorded by the vehicle at the high-speed entrance, and calculates the estimated traffic fee by integrating the entrance information and the toll standard.

Or after the vehicle arrives at the high-speed payment square, the vehicle owner can send notification information through the associated equipment. The system not only acquires the notification, but also calls the traffic information of the vehicle at the high-speed entrance and the high-speed exit information associated with the high-speed payment square, and the estimated traffic cost is accurately calculated according to the complete data.

Or the vehicle owner sends a navigation planning route through the associated equipment before entering the expressway, the vehicle-related payment system extracts an expressway passing track from the route after receiving the route, and invokes a charging model, and the estimated passing cost is calculated according to the information of the expressway section, mileage and the like related to the track and the charging standard.

In an embodiment, the flexibility of various pre-charging modes is high, whether the vehicle enters a high-speed front-selection entrance, sends a navigation planning route, enters a high-speed rear-selection exit and arrives at a payment square to send a notice, and a vehicle owner can select a proper mode according to actual conditions to attach different travel scenes. The system can calculate the cost by combining the entrance information, the exit information, the traffic track and other multidimensional data, so that the estimated cost is more close to the actual cost, and multi-charge or charge missing is avoided. And know in advance and estimate the expense and can let the car owner have a number in mind, can make payment preparation in advance, reduce the traffic latency, promote the convenience and the fluency of high-speed traffic, optimize whole trip experience.

In an embodiment, on the basis of the foregoing embodiment, after the step of detecting that the identified license plate number has an associated vehicle in the wading payment system, the wading payment system notifies the rail device at the high-speed entrance to raise the bar and record corresponding traffic information, the method further includes:

The vehicle-related payment system is based on a satellite positioning module, receives positioning signals generated by associated equipment of a vehicle on an expressway, and generates a vehicle track;

wherein the vehicle track is taken as one of the calculation factors when calculating the high-speed toll.

In this embodiment, after the vehicle-related payment system detects that the identified license plate number has an associated vehicle, notifies the rail device at the high-speed entrance to raise the bar and pass, and records the traffic information, the vehicle-related payment system can continuously receive the positioning signal sent by the vehicle-related device (such as vehicle navigation, mobile phone APP with positioning function, etc.) by means of the satellite positioning module. The positioning signals contain geographic position information of the vehicle at different moments on the expressway. The system accurately depicts the running path of the vehicle on the expressway according to the received series of positioning signals, so that the vehicle track is generated.

In calculating high speed tolls, vehicle trajectories become a critical calculation factor. Different highway section charging criteria may be different, and through the vehicle track, the system can clearly know which road sections the vehicle specifically travels through and which specific areas pass. By combining the detailed road section information and the corresponding charging rules, the system can more accurately calculate the cost of the vehicle passing at high speed, avoid cost errors caused by inaccurate estimation and ensure the fairness and accuracy of charging. For example, some road segments may be charged more due to high construction cost, large traffic flow, etc., and the vehicle track may determine whether to pass through these special road segments, thereby reasonably calculating the cost.

Optionally, a plurality of edge computing nodes are reasonably deployed along the expressway, and the nodes can receive and process satellite positioning signals sent by vehicle association equipment in real time. The edge computing nodes form an ad hoc network through the internet of things technology, and can automatically adjust a communication link and a data forwarding path according to real-time data transmission requirements and network conditions, so that efficient data sharing and collaborative processing are realized. And the data transmission quantity is reduced and the response speed of the system is improved by carrying out preliminary screening and processing on the positioning data.

In an embodiment, based on the foregoing embodiment, the artificial intelligence model encodes an estimated output of the motion blur estimation branch, generates a weight matrix, and performs weighted fusion on the weight matrix and the first feature map of the spectral channel estimation branch to obtain an adjusted first feature map.

In this embodiment, the motion blur estimation branch receives an original vehicle image as an input, and analyzes and estimates the motion blur condition of the image through a series of processing of neural network layers such as a convolution layer and a pooling layer. Finally, a motion blur related parameter vector p is output, wherein the vector contains information about the motion blur degree, direction and the like of the image, namely p= [ p ₁,p₂,⋯,p_m ], wherein m is the dimension of the vector, and each element represents motion blur information of different aspects.

In order to be able to apply motion blur parameters to the feature map adjustment of the spectral channel estimation branch, the estimated output needs to be encoded to generate a weight matrix. Let the coding function be g, which maps the motion blur parameter vector p to a weight matrix Z. The size and shape of this weight matrix is matched to the first feature map of the spectral channel estimation branch for subsequent weighted fusion operations.

For example, the first feature map of the spectral channel estimation branch is d×l×c (where D is height, L is width, and C is the number of channels), then the weight matrix Z may be c×c, and the encoding process may be expressed as z=g (p).

The spectrum channel estimation branch also receives an original vehicle image as input, and a series of convolution, pooling and other operations are performed to obtain a first feature map, wherein the feature map comprises feature information of the image on different spectrum channels.

Then the generated weight matrix Z is weighted and fused with the first feature map of the spectrum channel estimation branch, namely, for each position (i, j) in the first feature map, the feature vector in the channel dimension of the first feature mapAnd performing matrix multiplication operation on the weight matrix Z to obtain an adjusted characteristic vector f _i,j'=Z×f_i,j.

The adjusted feature vectors F _i,j' for all positions are combined to obtain the adjusted first feature map F ₁. The adjusted first feature map F ₁ integrates motion blur parameters and spectrum channel weight information, can reflect the features of the image more accurately, and provides a better foundation for the follow-up image deblurring and spectrum optimization processing.

In an embodiment, based on the foregoing embodiment, the artificial intelligence model performs dot product operation on the estimated output of the spectral channel estimation branch and the second feature map of the motion blur estimation branch to obtain a tensor with the same shape as the second feature map;

Carrying out normalization processing on tensors to obtain attention weights;

In this embodiment, in the artificial intelligence model based on the dual-branch network, the motion blur estimation branch is responsible for estimating motion blur parameters of an image, and the spectral channel estimation branch is responsible for estimating weights of different spectral channels of the image.

The spectrum channel estimation branch receives an original vehicle image, and outputs estimation results of different spectrum channel weights of the image after a series of convolution, pooling and other neural network operations. If the estimated output is a tensor S with a shape of C (channel number), it reflects the importance weight of each spectrum channel.

And the motion blur estimation branch can obtain a series of characteristic diagrams as a second characteristic diagram in the process of processing the original image.

And then carrying out dot product operation on the estimated output S of the spectrum channel estimation branch and a second feature map of the motion blur estimation branch, wherein the purpose is to integrate the weight information of the spectrum channel into the feature map of the motion blur estimation branch, so that the feature map can better reflect the importance of different spectrum channels.

For channel vectors at each spatial position (i, j) in the second feature mapOutput of the same with an estimate of the spectral channel estimation branchDot product operation is performed. That is, for each channel K (1≤K≤C), the calculation is performed. Combining the calculation results of all the positions, a tensor T with the same shape as the second characteristic diagram is obtained.

The resulting tensor T is normalized to map its element values into the [0,1] interval, so that it can be used as an attention weight.

Normalization of attention weights using Softmax function, i.e. channel vectors at each spatial position (i, j) in tensor TCalculating an attention weight vectorWherein:

;

the attention weight vectors a _i,j for all locations are then combined to obtain the attention weight tensor a. And multiplying the attention weight tensor A with a second feature map of the motion blur estimation branch element by element in the space dimension, enhancing the expression of important features and inhibiting the influence of unimportant features.

For the channel vector at each spatial position (i, j) in the second feature map and the attention weight tensor aAndElement-by-element multiplication to obtain an adjusted channel vectorWherein. The adjusted channel vectors q _i,j for all positions are combined to obtain the adjusted second feature map F ₂.

Therefore, the artificial intelligent model realizes the adjustment of the spectral channel estimation branch to the motion blur estimation branch feature map, enhances the expression of important features in the feature map by using a attention mechanism, improves the capturing and processing capacity of the model to image features, and is beneficial to more accurately identifying license plates.

In addition, the embodiment of the application also provides a vehicle-related payment system, and the internal architecture of the vehicle-related payment system can be shown in fig. 3, and the vehicle-related payment system comprises a processor, a memory, a communication interface and an input interface which are connected through a system bus. Wherein the processor is configured to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database is used for storing data called by the computer program. The communication interface is used for carrying out data communication with an external terminal. The input interface is used for receiving signals input by external equipment. The computer program is executed by a processor to implement a method of vehicle-related payment for road traffic as described in the above embodiments.

It will be appreciated by those skilled in the art that the structure shown in fig. 3 is merely a block diagram of a portion of the structure associated with the present inventive arrangements and is not intended to limit the wading payment system to which the present inventive arrangements are applied. For example, in some alternative embodiments, the wading payment system may further include an output interface (not shown), and the output interface is also coupled to the system bus and configured to output corresponding signals to the peripheral devices.

Furthermore, the present application proposes a computer-readable storage medium comprising a computer program which, when executed by a processor, implements the steps of the method for vehicle-related payment for traffic on highways as described in the above embodiments. It is understood that the computer readable storage medium in this embodiment may be a volatile readable storage medium or a nonvolatile readable storage medium.

In summary, in order to provide the vehicle-related payment method, the vehicle-related payment system and the computer-readable storage medium for passing on the expressway in the embodiment of the application, a special double-branch network pre-training artificial intelligent model is adopted, even if the image motion is fuzzy due to the high-speed running of a vehicle or the illumination condition on the expressway is complex, a clear and accurate fusion image can be generated for license plate number identification, so that the accurate identification requirement under the complex scene of the expressway is met, and once the identification is successful, the vehicle-related payment system can rapidly respond, automatically settle and pass the fee according to the pre-signed payment mode of the vehicle, no manual intervention is needed, the waiting time of the vehicle is reduced, the high-efficiency and convenient high-speed passing payment experience is provided for the vehicle owner, and the whole vehicle-related payment and passing process does not need to install ETC and other equipment for high-speed passing charging in advance, so that the cost of the vehicle passing on the expressway is reduced.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium provided by the present application and used in embodiments may include non-volatile and/or volatile memory. The nonvolatile Memory can include Read-Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash Memory. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (SYNCHLINK) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, apparatus, article, or method that comprises the element.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the scope of the application, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application or direct or indirect application in other related technical fields are included in the scope of the present application.

Claims

1. A vehicle-related payment method for highway travel, comprising:

Based on the cameras deployed at the highway entrance, vehicle images of vehicles about to pass are collected;

Utilize a pre-trained artificial intelligence model to identify the license plate number corresponding to a vehicle image; wherein the artificial intelligence model employs a dual-branch network, wherein a motion blur estimation branch is used to estimate the motion blur parameters of the image, and a spectral channel estimation branch is used to estimate the weights of different spectral channels of the image. An interaction mechanism is introduced between the outputs of the two branches, so that the estimated output of each branch acts on the feature map adjustment of the other branches; and, in each branch, the estimated output and the adjusted feature map of each branch are used to process the original image, and the images processed by each branch are fused to generate a deblurred and spectrally optimized fused image for license plate number recognition;

When it is detected that the identified license plate number exists in the vehicle-related payment system, the vehicle-related payment system notifies the barrier device at the highway entrance to lift the barrier to allow passage and record the corresponding passage information;

When the camera deployed at the highway exit captures the image of a vehicle about to pass and uses the artificial intelligence model to identify the corresponding license plate number, the vehicle-related payment system calls the billing model provided by the highway management system to calculate the highway toll based on the traffic information of the currently identified license plate number at the highway entrance and the traffic information at the current highway exit. Among them, if the camera deployed at the highway interval toll point captures the corresponding vehicle image and uses the artificial intelligence model to identify the corresponding license plate number, the traffic information of the highway interval toll point associated with the license plate number will be used as one of the calculation factors of the highway toll;

Automatically settle highway tolls based on the payment method pre-signed by the vehicle associated with the license plate number in the vehicle-related payment system;

After the toll payment is successfully settled, the vehicle payment system notifies the barrier device at the highway exit to lift the bar and let the vehicle pass;

The artificial intelligence model encodes the estimated output of the motion blur estimation branch to generate a weight matrix, and performs weighted fusion of the weight matrix with the first feature map of the spectral channel estimation branch to obtain an adjusted first feature map; the estimated output of the spectral channel estimation branch and the first feature map are used to process the original image to obtain a spectral optimization map after removing the influence of motion blur;

The artificial intelligence model performs a dot product operation on the estimated output of the spectral channel estimation branch and the second feature map of the motion blur estimation branch to obtain a tensor with the same shape as the second feature map; normalizes the tensor to obtain attention weights; multiplies the attention weights by the second feature map element-by-element in the spatial dimension to obtain an adjusted second feature map; and deblurs the original image based on the adjusted second feature map in combination with the previously estimated motion blur parameters to obtain a deblurred clear image;

The motion blur estimation branch processes the deblurred clear image, and the spectral channel estimation branch processes the spectral optimization image, which are fused to generate a deblurred and spectrally optimized fused image.

2. The vehicle-related payment method for highway tolls according to claim 1, wherein the step of automatically settling highway tolls based on the payment method pre-signed by the vehicle associated with the license plate number in the vehicle-related payment system comprises:

If the vehicle associated with the license plate number has pre-signed an online payment method in the vehicle-related payment system, the corresponding online payment method will be used to automatically settle the highway toll.

3. The vehicle-related payment method for highway tolls as claimed in claim 1, wherein the step of automatically settling highway tolls based on the payment method pre-signed by the vehicle associated with the license plate number in the vehicle-related payment system comprises:

If the vehicle associated with the license plate number has pre-signed a payment method of prepaid in the vehicle-related payment system, the corresponding highway toll will be deducted from the prepaid amount;

If the prepaid amount is fully deducted, the expressway toll settlement is deemed successful; and if there is any excess in the prepaid amount, the excess amount will be refunded to the original route;

If the prepaid amount cannot be fully deducted, a prompt message will be output for the car owner to manually pay the remaining amount.

4. The vehicle-related payment method for highway travel according to claim 3, characterized in that if the vehicle associated with the license plate number has pre-signed a payment method of prepaid in the vehicle-related payment system, before the step of deducting the corresponding highway toll from the prepaid amount, the method further comprises:

When the vehicle-related payment system receives the scheduled toll information sent by the associated device of the vehicle whose payment method is prepaid, it calls the billing model to calculate the corresponding estimated toll fee;

Based on the estimated toll, prepayment information is sent to the associated device of the corresponding vehicle to prompt the owner to pay the corresponding amount in advance.

5. The vehicle-related payment method for highway travel according to claim 4, wherein the predetermined travel information includes any one of the following:

The selected information of the highway entrance and exit sent by the associated device before the vehicle enters the highway;

The selected information of the highway exit sent by the associated device after the vehicle enters the highway, wherein the traffic information of the vehicle at the highway entrance is also used when calculating the estimated toll;

Notification information sent by the associated device upon a vehicle's arrival at a highway payment plaza. This notification information also includes the vehicle's highway entrance information and highway exit information associated with the highway payment plaza when calculating the estimated toll.

The associated device sends the navigation planning route before the vehicle enters the highway. When the vehicle-related payment system receives the navigation planning route, it extracts the corresponding highway passage trajectory and calculates the estimated toll.

6. The vehicle-related payment method for highway travel according to claim 1, characterized in that after the step of detecting that the identified license plate number exists in the vehicle-related payment system and notifying the barrier device at the highway entrance to lift the barrier and release the vehicle and recording the corresponding passage information, the method further comprises:

The vehicle-related payment system is based on a satellite positioning module, which receives positioning signals generated by the vehicle's associated equipment on the highway and generates vehicle tracks;

Among them, when calculating highway tolls, vehicle trajectory is used as one of the calculation factors.

7. A vehicle-related payment system, characterized in that the vehicle-related payment system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the vehicle-related payment method for traveling on a highway as described in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the vehicle-related payment method for passing through a highway as described in any one of claims 1 to 6 are implemented.