Stereo garage offline payment method and system based on multi-scene recognition

Abstract

The present application relates to the field of offline payment, and discloses a stereo garage offline payment method and system based on multi-scene recognition, comprising the following steps: based on multi-source recognition technology, analyzing the license plate number, parking time and position of the vehicle, and combining various algorithms to accurately calculate the parking fee of the vehicle. Vehicle information is preset in advance, and when the vehicle leaves, the vehicle information is retrieved in the database, offline payment is automatically performed, offline payment threat analysis and security optimization are performed during the offline payment transaction, and finally the offline payment transaction record is synchronized to the server for checking and correcting by the vehicle owner. The present application can calculate the fee of the vehicle parked in the stereo garage through multi-scene vehicle recognition, and perform offline payment and transaction monitoring and processing on the vehicle, thereby improving the use convenience of the stereo garage, ensuring the smooth payment in the case of network interruption, and improving the security of the transaction.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of offline payment, in particular to a stereo garage offline payment method and system based on multi-scene recognition. BACKGROUND

[0002] Offline payment refers to a way of completing payment transactions without real-time network connection, which is usually used in unstable network connection environments where direct communication with payment gateways or bank servers is impossible. In the context of offline payment, there are higher risks, such as the possibility of payment information being stolen by malicious software or hackers during the payment process, resulting in the leakage of user's bank card information or other sensitive information; payment data may be tampered with during transmission, resulting in tampering with payment amounts or recipient information, causing losses to users and merchants; due to unstable network or data transmission delay, transaction data may be inaccurate, resulting in payment disputes, etc. Offline payment for vehicle retrieval in a stereo garage will also encounter the above-mentioned various problems, but offline payment will improve the convenience of vehicle retrieval, improve travel efficiency, and facilitate the public, so a multi-scene recognition algorithm is needed to optimize and update the offline payment processing of the stereo garage, to ensure the security and convenience of offline payment. SUMMARY

[0003] The present application overcomes the shortcomings of the prior art and provides a stereo garage offline payment method and system based on multi-scene recognition.

[0004] To achieve the above-mentioned purposes, the technical solution adopted by the present application is as follows:

[0005] The present application provides a stereo garage offline payment method based on multi-scene recognition, comprising the following steps:

[0006] Introducing blind deconvolution algorithm and deep neural network, performing license plate recognition analysis on the target vehicle in the stereo garage to obtain the license plate number of the target vehicle;

[0007] Obtaining the parking position and parking time of the target vehicle in the stereo garage, and combining the license plate number of the target vehicle, calculating the real-time parking fee of the target vehicle;

[0008] Controlling the stereo garage server to implement offline payment processing for vehicles leaving the stereo garage, and performing payment threat analysis and payment security evaluation optimization on the stereo garage server;

[0009] If the stereo garage server still cannot collect the first type of parking fee to the offline payment account of the target vehicle within the offline payment time of the parking fee, the stereo garage server will remind the target vehicle to pay the parking fee.

[0010] Further, in a preferred embodiment of the present application, the blind deconvolution algorithm and the deep neural network are introduced to recognize and analyze the license plate number of the target vehicle in the stereo garage, specifically:

[0011] The vehicle that needs to be calculated for parking fee is calibrated as the target vehicle, and the gate of the stereo garage entrance is obtained, which is calibrated as the entrance gate;

[0012] When the target vehicle enters the entrance gate, the driving image of the target vehicle is obtained in real time through the camera in the stereo garage, which is calibrated as the target vehicle driving image;

[0013] The target vehicle driving image is subjected to image grayscale processing and image noise reduction processing to obtain a preprocessed target vehicle driving image;

[0014] The blind deconvolution algorithm is introduced, the initial blur kernel is determined based on the blind deconvolution algorithm, the preprocessed target vehicle driving image is subjected to convolution iterative calculation through the initial blur kernel, and a first iteration number is preset. When the number of convolution iterative calculation of the initial blur kernel on the preprocessed target vehicle driving image reaches the first iteration number, the convolution iterative calculation is stopped, and a motion blur elimination class target vehicle driving image is output;

[0015] The motion blur elimination class target vehicle driving image is subjected to feature extraction to obtain a class image feature data, a license plate number prediction number is preset, the class image feature data is imported into a deep neural network, and the target vehicle is subjected to license plate number prediction based on the license plate number prediction number through the deep neural network to obtain a predicted license plate number of the target vehicle;

[0016] All predicted license plate numbers of the target vehicle are analyzed, and the predicted license plate number with the highest prediction frequency is calibrated as the license plate number to be analyzed. The proportion of the license plate number to be analyzed in all predicted license plate numbers is determined, and a minimum proportion is preset;

[0017] If the proportion of the license plate number to be analyzed in all predicted license plate numbers is greater than the minimum proportion, the license plate number to be analyzed is the license plate number of the target vehicle, and the license plate number to be analyzed is calibrated as the target vehicle license plate number;

[0018] If the proportion of the license plate number to be analyzed in all predicted license plate numbers is less than the minimum proportion, the gradient descent method is introduced to update the initial blur kernel, the motion blur elimination class target vehicle driving image is subjected to convolution iterative second update through the updated initial blur kernel to obtain a motion blur elimination class target vehicle driving image, and the target vehicle is subjected to license plate number prediction based on the motion blur elimination class target vehicle driving image to obtain the target vehicle license plate number.

[0019] Further, in a preferred embodiment of the present application, the parking position and parking time of the target vehicle in the stereo garage are obtained, and the real-time parking fee of the target vehicle is calculated in combination with the target vehicle license plate number, specifically:

[0020] The stereo garage server is obtained, the target vehicle license plate number is imported into the stereo garage server memory, and the pre-registered license plate number that does not need to pay for parking is obtained in the stereo garage server, and is calibrated as a type of license plate number;

[0021] When the target vehicle enters the stereo garage, the data of the target vehicle license plate number is compared in the stereo garage server, if the target vehicle license plate number is not a type of license plate number, the path tracking real-time analysis of the target vehicle license plate number is performed through the camera in the stereo garage, and the parking position of the target vehicle is located.

[0022] In the stereo garage server, the parking time of the target vehicle is calculated in real time, and the vehicle parking charging rule is obtained, based on the vehicle parking charging rule and the parking time of the target vehicle, the parking fee of the target vehicle is calculated in real time in the stereo garage server, and the parking fee of the target vehicle is calibrated as a type of parking fee.

[0023] Further, in a preferred embodiment of the present application, the stereo garage server is controlled to implement offline payment processing for vehicles leaving the stereo garage, and payment threat analysis and payment security evaluation optimization are performed on the stereo garage server, specifically:

[0024] In the stereo garage server, the offline payment account of the target vehicle is input, the offline payment account of the target vehicle is bound with the target vehicle license plate number, and in combination with the type of parking fee in the stereo garage server, the vehicle information of the target vehicle is obtained;

[0025] The gate of the stereo garage exit is obtained, which is calibrated as an exit gate, the exit gate is connected with the stereo garage server, and the camera on the exit gate is obtained, which is calibrated as an exit gate camera, and the release recognition distance is preset;

[0026] If the distance between the vehicle and the exit gate is less than the release recognition distance, the corresponding vehicle is calibrated as a vehicle to be driven out, and the license plate number of the vehicle to be driven out is obtained through the exit gate camera;

[0027] The license plate number of the vehicle to be driven out is analyzed in the stereo garage server, if the license plate number of the vehicle to be driven out is a type of license plate number, the exit gate is controlled to directly release the vehicle to be driven out;

[0028] If the license plate number of the vehicle to be driven out is the license plate number of the target vehicle, it is proved that the vehicle to be driven out is the target vehicle, the exit gate is controlled to release the target vehicle, and the vehicle information of the target vehicle is retrieved based on the license plate number of the target vehicle in the stereo garage server;

[0029] When the target vehicle leaves the stereo garage, a preset parking fee offline payment time is set, the vehicle information of the target vehicle is analyzed by the stereo garage server, a type of parking fee is obtained, and it is judged whether the stereo garage server can collect the type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time;

[0030] If not, an offline payment attack tree model is constructed, payment threat analysis is performed on the stereo garage server based on the offline payment attack tree model, and payment security evaluation optimization is performed on the stereo garage server based on the payment threat analysis result.

[0031] Further, in a preferred embodiment of the present application, the offline payment attack tree model is constructed, the payment threat analysis is performed on the stereo garage server based on the offline payment attack tree model, and the payment security evaluation optimization is performed on the stereo garage server based on the payment threat analysis result, specifically:

[0032] The attack tree algorithm is introduced, and threat modeling is performed based on the attack tree algorithm to obtain an attack tree basic model;

[0033] The historical data network is obtained, in which all dangerous attack paths existing when the stereo garage server performs parking fee offline payment processing on the offline payment account of the target vehicle are retrieved and labeled as offline payment attack paths;

[0034] The offline payment attack paths are imported into the attack tree basic model for model convolution training to obtain an offline payment attack tree model;

[0035] Based on the offline payment attack tree model, the occurrence probability of different offline payment attack paths when the stereo garage server performs parking fee offline payment processing on the offline payment account of the target vehicle is evaluated and labeled as offline payment attack path occurrence probability;

[0036] A preset offline payment attack path influence danger probability is set, the offline payment attack path occurrence probability is analyzed, and the corresponding offline payment attack path with the offline payment attack path occurrence probability greater than the offline payment attack path influence danger probability is labeled as a dangerous attack path;

[0037] The dangerous attack path occurrence probability is sorted in descending order to obtain a dangerous attack path occurrence probability sorting table, and in the historical data network, based on the dangerous attack path occurrence probability sorting table, the path security patch of the dangerous attack path is searched and output in sequence, so that the stereo garage server can collect the first type of parking fee to the offline payment account of the target vehicle within the offline payment time of the parking fee.

[0038] Further, in a preferred embodiment of the present application, if the stereo garage server still cannot collect the first type of parking fee to the offline payment account of the target vehicle within the offline payment time of the parking fee, the stereo garage server reminds the target vehicle of vehicle fee payment, specifically:

[0039] If the path security patch of the dangerous attack path is output, and the stereo garage server still cannot collect the first type of parking fee to the offline payment account of the target vehicle within the offline payment time of the parking fee, a target vehicle payment list is generated in the stereo garage server;

[0040] The target vehicle payment list is sent to the offline payment account of the target vehicle by the stereo garage server, and a payment time is preset, and it is judged whether the offline payment account of the target vehicle completes the payment within the payment time after the target vehicle payment list is sent to the offline payment account of the target vehicle by the stereo garage server;

[0041] If not, the stereo garage blacklist is set, and the license plate number of the target vehicle is included in the stereo garage blacklist in the stereo garage server;

[0042] The camera of the entrance gate is acquired, and when the camera of the entrance gate recognizes the license plate number of the target vehicle in the stereo garage blacklist, the entrance gate is controlled to perform the prohibition of driving into operation on the target vehicle corresponding to the license plate number of the target vehicle in the stereo garage blacklist.

[0043] The second aspect of the present application also provides a stereo garage offline payment system based on multi-scene recognition, which comprises a memory and a processor, and the memory stores a stereo garage offline payment method, and the stereo garage offline payment method is executed by the processor to realize the following steps:

[0044] The blind deconvolution algorithm and the deep neural network are introduced to recognize and analyze the license plate number of the target vehicle in the stereo garage to obtain the license plate number of the target vehicle.

[0045] The parking position and parking time of the target vehicle are acquired in the stereo garage, and the real-time parking fee of the target vehicle is calculated in combination with the license plate number of the target vehicle.

[0046] The control stereo garage server carries out offline payment processing on the vehicle leaving the stereo garage, and carries out payment threat analysis and payment security assessment optimization on the stereo garage server.

[0047] If the stereo garage server still cannot collect the first type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time, the stereo garage server reminds the target vehicle of vehicle payment through the offline payment account of the target vehicle.

[0048] The technical defects in the background art are solved, and the present application has the following advantages: based on multi-source identification technology, the license plate number, parking time and position of the vehicle are analyzed, and a variety of algorithms are combined to accurately calculate the parking fee of the vehicle. Vehicle information is preset in advance, and vehicle information is retrieved in the database when the vehicle leaves, and offline payment is automatically performed, and offline payment threat analysis and security optimization are performed during offline payment transaction, and finally the offline payment transaction record is synchronized to the server for vehicle owner checking and correction. The present application can calculate the parking fee of the vehicle in the stereo garage through multi-scene vehicle identification, and perform offline payment and transaction monitoring on the vehicle, improve the use convenience of the stereo garage, and ensure the smooth payment in the case of network interruption, and improve the security of the transaction. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings of embodiments according to these drawings without creative labor.

[0050] Figure 1 A flowchart of a stereo garage offline payment method based on multi-scene identification is shown;

[0051] Figure 2 A method flowchart for implementing offline payment processing on a vehicle leaving a stereo garage, and performing payment threat analysis and payment security assessment optimization during offline payment is shown;

[0052] Figure 3 A program view of a stereo garage offline payment system based on multi-scene identification is shown. DETAILED DESCRIPTION

[0053] In order to more clearly illustrate the above-mentioned purposes, features and advantages of the present application, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0054] In the following description, a lot of specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can be practiced in other manners different from those described herein, therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0055] Figure 1 A flowchart of a stereo garage offline payment method based on multi-scene recognition is shown, including the following steps:

[0056] S102: Introduce a blind deconvolution algorithm and a deep neural network, and perform license plate number recognition analysis on the target vehicle in the stereo garage to obtain the license plate number of the target vehicle;

[0057] S104: Obtain the parking position and parking time of the target vehicle in the stereo garage, and calculate the real-time parking fee of the target vehicle in combination with the license plate number of the target vehicle;

[0058] S106: Control the stereo garage server to implement offline payment processing for the vehicle leaving the stereo garage, and perform payment threat analysis and payment security evaluation optimization on the stereo garage server;

[0059] S108: If the stereo garage server still cannot collect the first type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time, the stereo garage server reminds the target vehicle to pay the vehicle fee.

[0060] Further, in a preferred embodiment of the present application, the blind deconvolution algorithm and the deep neural network are introduced to perform license plate number recognition analysis on the target vehicle in the stereo garage to obtain the license plate number of the target vehicle, specifically:

[0061] The vehicle that needs to perform parking fee calculation is calibrated as the target vehicle, and the gate of the stereo garage entrance is obtained, which is calibrated as the entrance gate;

[0062] When the target vehicle enters the entrance gate, the driving image of the target vehicle is obtained in real time through the camera in the stereo garage, which is calibrated as the target vehicle driving image;

[0063] The target vehicle driving image is subjected to image grayscale processing and image noise reduction processing to obtain a preprocessed target vehicle driving image;

[0064] A blind deconvolution algorithm is introduced, an initial blur convolution kernel is determined based on the blind deconvolution algorithm, the preprocessed target vehicle driving image is subjected to convolution iterative calculation through the initial blur convolution kernel, and a first iteration number is preset, when the number of convolution iterative calculation of the initial blur convolution kernel on the preprocessed target vehicle driving image reaches the first iteration number, the convolution iterative calculation is stopped, and a motion blur elimination first type of target vehicle driving image is output.

[0065] extracting features of the one-class vehicle driving image subjected to the motion blur elimination, obtaining one-class image feature data, presetting a license plate number prediction number, inputting the one-class image feature data into a deep neural network, and predicting the license plate number of the target vehicle through the deep neural network based on the license plate number prediction number, to obtain a predicted license plate number of the target vehicle;

[0066] analyzing all the predicted license plate numbers of the target vehicle, and marking the predicted license plate number with the most prediction times as a license plate number to be analyzed, judging the proportion of the license plate number to be analyzed in all the predicted license plate numbers, and presetting a minimum proportion;

[0067] if the proportion of the license plate number to be analyzed in all the predicted license plate numbers is greater than the minimum proportion, the license plate number to be analyzed is the license plate number of the target vehicle, and the license plate number to be analyzed is marked as the license plate number of the target vehicle;

[0068] if the proportion of the license plate number to be analyzed in all the predicted license plate numbers is less than the minimum proportion, a gradient descent method is introduced to update the initial fuzzy convolution kernel, the one-class vehicle driving image subjected to the motion blur elimination is subjected to convolution iteration twice through the updated initial fuzzy convolution kernel, to obtain a two-class vehicle driving image subjected to the motion blur elimination, and the license plate number of the target vehicle is predicted based on the two-class vehicle driving image subjected to the motion blur elimination, to obtain the license plate number of the target vehicle.

[0069] It should be noted that the parking time, parking position and other vehicle information of the vehicle in the stereo garage are determined by the license plate of the vehicle, so the license plate number of the vehicle needs to be obtained. There are many parking spaces in the garage, and the charging standards of different parking spaces may be different, so the moving track of the vehicle needs to be obtained to determine the parking position of the vehicle. When the vehicle enters the gate, the motion image of the vehicle needs to be captured in real time by the camera to generate the moving track of the vehicle, and the number of the license plate needs to be accurately determined, which can be determined by image recognition. During the driving of the target vehicle, the picture of the target vehicle captured by the camera may be blurred due to high speed or dirt, resulting in motion blur of the recognized image, so that the license plate number of the target vehicle in the motion process cannot be accurately recognized. Therefore, the image with motion blur needs to be processed to eliminate the motion blur, so as to accurately recognize the license plate number of the target vehicle in the motion process. The blind deconvolution algorithm is introduced, which is an algorithm that can restore the original clear image from the blurred or real image. The image is convolved by the blur kernel to realize the processing of eliminating the image motion blur. After obtaining the vehicle image with motion blur eliminated, the image is imported into the deep neural network to predict the license plate number. After multiple predictions, multiple prediction results are obtained, one prediction result corresponds to one license plate number, and if the proportion of the license plate number with the highest frequency in all license plate numbers is greater than a preset value, it proves that the prediction effect is accurate, and the license plate number with the highest frequency is directly output to obtain the target vehicle license plate number. If the proportion of the license plate number with the highest frequency in all license plate numbers is less than a preset value, it proves that the effect of eliminating the image motion blur may not be ideal, resulting in inaccurate prediction of the license plate number, so the blur kernel needs to be optimized by using the gradient descent method. The gradient descent method can minimize the error between the motion blurred image and the real image to improve the elimination effect of the motion blur. After re-eliminating the motion blur, the target vehicle license plate number is obtained.

[0070] Further, in a preferred embodiment of the present application, the parking position and parking time of the target vehicle in the stereo garage are obtained, and the real-time parking fee of the target vehicle is calculated by combining the target vehicle license plate number, specifically:

[0071] The stereo garage server is obtained, the target vehicle license plate number is imported into the stereo garage server memory, and the pre-registered license plate number that does not need to pay for parking is obtained in the stereo garage server, which is calibrated as a type of license plate number;

[0072] When the target vehicle enters the stereo garage, the data of the target vehicle license plate number is compared in the stereo garage server, if the target vehicle license plate number is not a type of license plate number, the path tracking of the target vehicle license plate number is analyzed in real time by the camera in the stereo garage, and the parking position of the target vehicle is located;

[0073] In the stereoscopic garage server, the parking time of the target vehicle is calculated in real time, and the vehicle parking charging rule is obtained, and based on the vehicle parking charging rule and the parking time of the target vehicle, the parking fee of the target vehicle is calculated in real time in the stereoscopic garage server, and the parking fee of the target vehicle is marked as a type of parking fee.

[0074] It should be noted that after the license plate number of the target vehicle is identified during driving, that is, after the target vehicle license plate number is obtained, since there are multiple cameras in the garage, multiple cameras simultaneously perform image recognition, the driving trajectory of the target vehicle can be generated, and the parking position thereof can be determined. In combination with the parking time and the vehicle parking charging rule, the parking fee of the target vehicle can be calculated in real time, and is marked as a type of parking fee. It should be noted that if the target vehicle license plate number is a license plate number that has been registered in the stereoscopic garage server, for example, the target vehicle is the vehicle of the owner, and the owner has purchased a parking space, then when the target vehicle enters the stereoscopic garage, no charging processing is required.

[0075] Further, in a preferred embodiment of the present application, if the stereoscopic garage server still cannot collect the type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time, the stereoscopic garage server reminds the target vehicle to pay the vehicle fee, specifically:

[0076] If the path security patch of the dangerous attack path is output, and the stereoscopic garage server still cannot collect the type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time, a target vehicle payment list is generated in the stereoscopic garage server;

[0077] The stereoscopic garage server sends the target vehicle payment list to the offline payment account of the target vehicle, and presets a payment time, and judges whether the offline payment account of the target vehicle completes the payment within the payment time after the stereoscopic garage server sends the target vehicle payment list to the offline payment account of the target vehicle;

[0078] If not, the stereoscopic garage black list is set, and the target vehicle license plate number is included in the stereoscopic garage black list in the stereoscopic garage server;

[0079] The camera of the entrance gate is obtained, and when the camera of the entrance gate recognizes the target vehicle license plate number in the stereoscopic garage black list, the entrance gate performs a prohibition driving-in operation on the target vehicle corresponding to the target vehicle license plate number in the stereoscopic garage black list.

[0080] It should be noted that if the path security patch of the dangerous attack path is output, and the stereo garage server still cannot collect the first type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time, it proves that the reason why the stereo garage server cannot collect the parking fee from the offline payment account of the target vehicle may be that the user end has a problem, such as that the balance of the bank card bound in the offline payment account of the target vehicle is insufficient, or that the offline payment account of the target vehicle is set to require payment confirmation. Therefore, it is necessary to generate a target vehicle payment list in the stereo garage server and send it to the offline payment account of the target vehicle to remind the user to pay. If the payment is still not made within the predetermined time, the license plate number corresponding to the target vehicle will be put into a blacklist, and the target vehicle will be prohibited from entering the stereo garage.

[0081] Figure 2 A method flow chart for implementing offline payment processing for vehicles leaving the stereo garage and performing payment threat analysis and payment security evaluation optimization during offline payment is shown, including the following steps:

[0082] S202: obtaining vehicle information of the target vehicle according to the offline payment account of the target vehicle, the license plate number of the target vehicle and the first type of parking fee in the stereo garage server;

[0083] S204: controlling the stereo garage server to implement offline payment processing for vehicles leaving the stereo garage based on the vehicle information of the target vehicle;

[0084] S206: constructing an offline payment attack tree model, performing payment threat analysis on the stereo garage server based on the offline payment attack tree model, and performing payment security evaluation optimization on the stereo garage server based on the payment threat analysis result.

[0085] Further, in a preferred embodiment of the present application, the control of the stereo garage server to implement offline payment processing for vehicles leaving the stereo garage and the payment threat analysis and payment security evaluation optimization of the stereo garage server are specifically:

[0086] An exit gate of the stereo garage is obtained, which is designated as an exit gate, the exit gate is connected with the stereo garage server, a camera is obtained on the exit gate, which is designated as an exit gate camera, and a release recognition distance is preset;

[0087] If the distance between a vehicle and the exit gate is less than the release recognition distance, the corresponding vehicle is designated as a vehicle to be driven out, and the license plate number of the vehicle to be driven out is obtained through license plate number recognition of the vehicle to be driven out by the exit gate camera;

[0088] The license plate number of the vehicle to be driven out is analyzed in the stereo garage server, if the license plate number of the vehicle to be driven out is a type of license plate number, the exit gate is directly controlled to release the vehicle to be driven out;

[0089] If the license plate number of the vehicle to be driven out is the target vehicle license plate number, it is proved that the vehicle to be driven out is the target vehicle, the exit gate is controlled to release the target vehicle, and the vehicle information of the target vehicle is retrieved based on the target vehicle license plate number in the stereo garage server;

[0090] When the target vehicle leaves the stereo garage, a preset parking fee offline payment time is set, the vehicle information of the target vehicle is analyzed by the stereo garage server to obtain a type of parking fee, and it is judged whether the stereo garage server can collect the type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time.

[0091] It should be noted that after obtaining the vehicle information of the target vehicle, when the target vehicle needs to drive out of the stereo garage, the corresponding charging process needs to be performed in real time according to the vehicle information of the target vehicle. When the target vehicle approaches the exit gate, that is, the distance of the vehicle is less than the release identification distance, the license plate number of the vehicle to be driven out is identified by the garage exit camera, and the vehicle information of the vehicle is retrieved according to the license plate number. If the license plate number of the vehicle is a type of license plate number, the corresponding vehicle can be directly released. If the license plate number of the vehicle is not a type of license plate number, the vehicle is also directly released, but after the release, the vehicle information is obtained according to the license plate number, and the parking fee is collected from the vehicle after the vehicle leaves the garage. The parking fee collection method is offline payment. The advantage of the present application is that the travel time of the vehicle can be saved, and the travel efficiency of the vehicle can be improved.

[0092] Further, in a preferred embodiment of the present application, the offline payment attack tree model is constructed, the stereo garage server is analyzed based on the offline payment attack tree model, and the payment security evaluation optimization of the stereo garage server is based on the payment threat analysis result, specifically:

[0093] The attack tree algorithm is introduced, and threat modeling is performed based on the attack tree algorithm to obtain an attack tree basic model;

[0094] The historical data network is obtained, in the historical data network, all dangerous attack paths existing when the stereo garage server performs parking fee offline payment processing to the offline payment account of the target vehicle are retrieved, and are calibrated as offline payment attack paths;

[0095] The offline payment attack path is imported into the attack tree basic model for model convolution training to obtain an offline payment attack tree model;

[0096] Based on the offline payment attack tree model, the occurrence probability of different offline payment attack paths when the stereo garage server performs parking fee offline payment processing on the offline payment account of the target vehicle is evaluated, and is labeled as an offline payment attack path occurrence probability;

[0097] A preset offline payment attack path impact risk probability is analyzed based on the offline payment attack path occurrence probability, and the corresponding offline payment attack path with an offline payment attack path occurrence probability greater than the offline payment attack path impact risk probability is labeled as a dangerous attack path.

[0098] The offline payment attack path occurrence probabilities of the dangerous attack paths are sorted in descending order to obtain a dangerous attack path occurrence probability sorting table. In the historical data network, the path security patches of the dangerous attack paths are searched and output based on the dangerous attack path occurrence probability sorting table, so that the stereo garage server can collect a type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time.

[0099] It should be noted that when the stereo garage server collects parking fees from the offline payment account of the target vehicle, if there is a vulnerability in the stereo garage server, it is easy to cause attacks during offline payment, and thus user payment information, transaction data, etc. Therefore, payment security needs to be evaluated to ensure the security of offline payment. Threat modeling can be used to determine possible security threats and attack paths, and thus to prevent processing. Threat modeling can be achieved by constructing an attack tree, which can be used to analyze potential threats and attack paths in the system and determine the generation probability of different attack paths. A basic model of the attack tree is constructed, and the attack tree basic model is trained. The training method is to preset multiple possible attack paths on the attack tree basic model to obtain an offline payment attack tree model, and to determine the occurrence probability of different attack paths when the stereo garage server collects parking fees from the offline payment account of the target vehicle, i.e. the offline payment attack path occurrence probability. If the offline payment attack path occurrence probability is high, it means that the corresponding offline payment attack path is likely to have an impact during offline payment, and needs to be prevented. Therefore, a dangerous attack path occurrence probability sorting table is constructed, and the path security patches of the dangerous attack paths are searched and output according to the order of different attack paths in the dangerous attack path occurrence probability sorting table in the historical data network. The path security patches of the dangerous attack paths can effectively repair the vulnerabilities in the stereo garage server, prevent the stereo garage server from being attacked by dangerous attack paths, and ensure that the stereo garage server can collect a type of parking fee from the offline payment account of the target vehicle within the parking fee offline payment time.

[0100] Further, the multi-scene recognition-based stereo garage offline payment method further comprises the following steps.

[0101] When the target vehicle license plate number is a first type of license plate number, the target vehicle is determined as a first type of vehicle, and a parking position of the first type of vehicle is determined in a stereo garage server; when the first type of vehicle enters the stereo garage, a real-time parking position of the first type of vehicle is determined through a camera of the stereo garage.

[0102] If the real-time parking position of the first type of vehicle does not match the parking position of the first type of vehicle, an offline payment account of the first type of vehicle is obtained, and parking position change information is sent to the offline payment account of the first type of vehicle through the stereo garage server.

[0103] A preset parking position change limited time is set, if the real-time parking position of the first type of vehicle still does not match the parking position of the first type of vehicle after the parking position change limited time, a parking time of the first type of vehicle is obtained, and a parking fee of the first type of vehicle is calculated according to a vehicle parking charging rule.

[0104] It should be noted that the first type of vehicle is a vehicle with a parking space, i.e. an owner's vehicle, etc. After the owner purchases a parking space, the vehicle should be parked in the owner's parking space or a parking space with the same charging as the owner's parking space, and cannot be parked in a parking space with different charging, because the charging standards of different parking spaces may be different. The parking space with the same charging as the owner's parking space is the parking position, and the camera of the stereo garage can obtain the vehicle moving track according to the license plate number, determine the parking position, and if the parking position of the vehicle is different from the parking position, the vehicle owner needs to be reminded to move the vehicle through the stereo garage server, and if the vehicle is not moved within the limited time, the first type of vehicle parking fee needs to be charged to the vehicle owner according to the charging standard.

[0105] As shown in FIG. Figure 3 The second aspect of the present application further provides a multi-scene recognition-based stereo garage offline payment system, which comprises a memory 31 and a processor 32, the memory 31 stores a stereo garage offline payment method, and the stereo garage offline payment method is executed by the processor 32 to realize the following steps:

[0106] The blind deconvolution algorithm and the deep neural network are introduced to identify and analyze the license plate number of the target vehicle in the stereo garage to obtain the target vehicle license plate number.

[0107] The parking position and the parking time of the target vehicle are obtained in the stereo garage, and the real-time parking fee of the target vehicle is calculated according to the target vehicle license plate number.

[0108] The control stereo garage server carries out offline payment processing on the vehicle leaving the stereo garage, and carries out payment threat analysis and payment security evaluation optimization on the stereo garage server;

[0109] If the stereo garage server still cannot collect the first type of parking fee to the offline payment account of the target vehicle within the parking fee offline payment time, the stereo garage server reminds the target vehicle to pay the vehicle fee.

[0110] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-scene recognition-based offline payment method for automated parking garages, characterized in that: Includes the following steps: By introducing blind deconvolution algorithm and deep neural network, the license plate number of the target vehicle is identified and analyzed in the three-dimensional parking garage to obtain the license plate number of the target vehicle; The parking location and parking time of the target vehicle in the multi-level parking garage are obtained, and the real-time parking fee of the target vehicle is calculated by combining the target vehicle's license plate number. The system controls the automated parking garage server to process offline payments for vehicles leaving the garage, and performs payment threat analysis and payment security assessment and optimization on the server. Specifically, an offline payment attack tree model is constructed during the offline payment processing. Based on this model, a payment threat analysis is performed on the automated parking garage server. Finally, based on the payment threat analysis results, a payment security assessment and optimization of the automated parking garage server is conducted. An attack tree algorithm is introduced, and threat modeling is performed based on the attack tree algorithm to obtain a basic attack tree model; Obtain historical data network, and in the historical data network, retrieve all dangerous attack paths that exist when the three-dimensional parking garage server performs offline payment processing of parking fees to the offline payment account of the target vehicle, and mark them as offline payment attack paths; The offline payment attack path is imported into the attack tree base model for model convolution training to obtain the offline payment attack tree model. Based on the offline payment attack tree model, the probability of different offline payment attack paths occurring when the automated parking garage server performs offline payment processing for parking fees to the offline payment account of the target vehicle is evaluated and denoted as the probability of occurrence of the offline payment attack path. The probability of offline payment attack paths being affected is preset. The occurrence probability of the offline payment attack paths is analyzed, and the offline payment attack paths whose occurrence probability is greater than the probability of being affected by the offline payment attack paths are marked as dangerous attack paths. The probability of occurrence of offline payment attack paths of dangerous attack paths is sorted in descending order to obtain a dangerous attack path occurrence probability sorting table. In the historical data network, based on the dangerous attack path occurrence probability sorting table, path security patches of dangerous attack paths are retrieved and output in sequence, so that the three-dimensional parking garage server can collect a type of parking fee from the offline payment account of the target vehicle during the offline payment period of parking fees. If the automated parking garage server is unable to collect the first-class parking fee from the offline payment account of the target vehicle within the offline payment period, it will send a vehicle payment reminder to the offline payment account of the target vehicle through the automated parking garage server.

2. The offline payment method for multi-scene recognition parking garages according to claim 1, characterized in that, The method involves introducing a blind deconvolution algorithm and a deep neural network to perform license plate recognition and analysis on target vehicles within a multi-level parking garage, thereby obtaining the target vehicle's license plate number. Specifically: The vehicle for which parking fees need to be calculated is identified as the target vehicle, and the gate at the entrance of the multi-level parking garage is obtained and identified as the entrance gate. When the target vehicle enters the entrance gate, the camera in the automated parking garage will capture the driving image of the target vehicle in real time and mark it as the driving image of the target vehicle. The target vehicle driving image is subjected to image grayscale processing and image noise reduction processing to obtain a preprocessed target vehicle driving image; A blind deconvolution algorithm is introduced. An initial blurred convolution kernel is determined based on the blind deconvolution algorithm. The preprocessed target vehicle driving image is subjected to convolution iteration calculation through the initial blurred convolution kernel. A first iteration number is preset. When the number of convolution iteration calculations of the preprocessed target vehicle driving image by the initial blurred convolution kernel reaches the first iteration number, the convolution iteration calculation is stopped, and a motion blur-reduced target vehicle driving image is output. The motion blur removal of the target vehicle driving image is used to extract features to obtain a type of image feature data. The number of license plate number predictions is preset. The type of image feature data is imported into a deep neural network. Based on the number of license plate number predictions, the deep neural network is used to predict the license plate number of the target vehicle to obtain the predicted license plate number of the target vehicle. Analyze all predicted license plate numbers of the target vehicle, and mark the predicted license plate number that appears most frequently as the license plate number to be analyzed. Determine the proportion of the license plate number to be analyzed among all predicted license plate numbers, and preset the minimum proportion. If the proportion of the license plate number to be analyzed in all predicted license plate numbers is greater than the minimum proportion, then the license plate number to be analyzed is the license plate number of the target vehicle, and the license plate number to be analyzed is marked as the license plate number of the target vehicle. If the proportion of the license plate number to be analyzed among all predicted license plate numbers is less than the minimum proportion, then the gradient descent method is introduced to update the initial blur convolution kernel. The updated initial blur convolution kernel is used to perform a second convolution iteration update on the motion blur removal of the first type of target vehicle driving image to obtain the motion blur removal of the second type of target vehicle driving image. Based on the motion blur removal of the second type of target vehicle driving image, the license plate number of the target vehicle is predicted to obtain the license plate number of the target vehicle.

3. The offline payment method for multi-scene recognition parking garages according to claim 1, characterized in that, The process of obtaining the parking location and parking time of the target vehicle within the automated parking garage, and calculating the real-time parking fee for the target vehicle based on its license plate number, specifically involves: Obtain the automated parking garage server, import the target vehicle's license plate number into the automated parking garage server for storage, and retrieve the pre-registered license plate number that does not require parking payment from the automated parking garage server, and mark it as a Class I license plate number; When a target vehicle enters the automated parking garage, the vehicle's license plate number is compared with the data in the automated parking garage server. If the vehicle's license plate number is not of the same type, the target vehicle's license plate number is tracked and analyzed in real time by the cameras in the automated parking garage to locate the vehicle's parking position. Within the automated parking garage server, the parking time of the target vehicle is calculated in real time, and the vehicle parking fee rules are obtained. Based on the vehicle parking fee rules and the parking time of the target vehicle, the parking fee of the target vehicle is calculated in real time within the automated parking garage server, and the parking fee of the target vehicle is marked as a type of parking fee.

4. The offline payment method for multi-scene recognition parking garages according to claim 1, characterized in that, The control server for the automated parking garage performs offline payment processing for vehicles leaving the garage, and conducts payment threat analysis and payment security assessment and optimization of the server, specifically: In the automated parking garage server, the offline payment account of the target vehicle is entered, and the offline payment account of the target vehicle is bound to the license plate number of the target vehicle. Combined with a type of parking fee in the automated parking garage server, the vehicle information of the target vehicle is obtained. The exit gate of the automated parking garage is obtained and labeled as the exit gate. The exit gate is connected to the automated parking garage server, and the camera on the exit gate is obtained and labeled as the exit gate camera. At the same time, the passage recognition distance is preset. If the distance between a vehicle and the exit gate is less than the release recognition distance, the corresponding vehicle is marked as a vehicle to be driven out. The license plate number of the vehicle to be driven out is obtained by the license plate number recognition of the vehicle to be driven out through the exit gate camera. The license plate number of the vehicle to be driven out is analyzed in the automated parking garage server. If the license plate number of the vehicle to be driven out is a Class I license plate number, the exit gate is controlled to allow the vehicle to drive out directly. If the license plate number of the vehicle to be driven out is the same as that of the target vehicle, it proves that the vehicle to be driven out is the target vehicle. The exit gate is controlled to release the target vehicle, and the vehicle information of the target vehicle is retrieved in the automated parking garage server based on the license plate number of the target vehicle. When the target vehicle leaves the automated parking garage, a preset offline payment time for parking fees is established. The automated parking garage server analyzes the vehicle information of the target vehicle to obtain the first type of parking fee. At the same time, it determines whether the automated parking garage server can collect the first type of parking fee from the target vehicle's offline payment account during the offline payment time. If not, an offline payment attack tree model is constructed, and payment threat analysis is performed on the automated parking garage server based on the offline payment attack tree model. Based on the payment threat analysis results, payment security assessment and optimization are performed on the automated parking garage server.

5. The offline payment method for multi-scene recognition parking garages according to claim 1, characterized in that, If the automated parking garage server is unable to collect the first type of parking fee from the target vehicle's offline payment account within the offline payment period, the server will send a payment reminder to the target vehicle's offline payment account. Specifically: If, after the path security patch for the dangerous attack path is output, the automated parking garage server is still unable to collect the first type of parking fee from the offline payment account of the target vehicle within the offline payment period, then a payment list for the target vehicle will be generated in the automated parking garage server. The system sends a payment list of the target vehicle to the offline payment account of the target vehicle through the automated parking garage server, and presets a payment time. It then determines whether the offline payment account of the target vehicle has completed the payment within the payment time after the automated parking garage server sends the payment list to the offline payment account of the target vehicle. If not, then set up a blacklist for the automated parking system and add the target vehicle's license plate number to the blacklist in the automated parking system server; The system acquires information from the entrance gate's camera. When the camera at the entrance gate identifies the license plate number of a target vehicle in the blacklist of the automated parking garage, it controls the entrance gate to prohibit the target vehicle corresponding to that license plate number from entering the automated parking garage.

6. An offline payment system for a multi-scene parking garage based on multi-scene recognition, characterized in that: The automated parking garage offline payment system includes a memory and a processor. The memory stores an automated parking garage offline payment method program. When the automated parking garage offline payment method program is executed by the processor, it implements the steps of the automated parking garage offline payment method as described in any one of claims 1-5.

Images