License plate verification method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of electric bicycle license plate verification, and relates to a license plate verification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining an account verification request, carrying out the identity recognition of hardware equipment according to the account verification request, obtaining an identity verification result, and carrying out the verification of a license plate according to the identity verification result. The method comprises the steps of generating license plate image recognition information, obtaining a license plate recognition result from the license plate image recognition information, obtaining a license plate inspection and recognition standard, recognizing the license plate recognition result according to the license plate inspection and recognition standard to obtain a license plate verification result, and marking a vehicle according to the license plate verification result to obtain a vehicle validity result. The method and the device have the effect of reducing the risk that the vehicle is penalized.

Description

Technical Field

[0001] The present invention relates to the technical field of electric bicycle license plate verification, and in particular to a license plate verification method, device, computer equipment and storage medium. Background Art

[0002] Currently, in the electric bicycle industry, the issue of license plate validity has been plaguing various electric bicycle-related companies, involving issues such as whether the vehicle can be driven and rented normally. Often, they can only know whether the license plate is normal through feedback from other platforms and when problems occur afterwards, but they cannot know the validity of the license plate in advance, which leads to the vehicle being fined.

[0003] Regarding the above-mentioned related technologies, how to identify and verify the license plate of electric bicycles and reduce the risk of vehicles being fined is a technical problem that urgently needs to be solved. Summary of the invention

[0004] In order to reduce the risk of vehicles being fined, the present application provides a license plate verification method, device, computer equipment and storage medium.

[0005] The above-mentioned invention objective of the present application is achieved through the following technical solutions: A license plate verification method, the license plate verification method comprising: Obtain an account verification request, perform identity recognition on a hardware device according to the account verification request, and obtain an identity verification result; generate license plate image recognition information according to the identity verification result, and obtain a license plate recognition result from the license plate image recognition information; A license plate inspection and recognition standard is obtained, the license plate inspection and recognition standard recognizes the license plate recognition result to obtain a license plate verification result, and a vehicle is identified according to the license plate verification result to obtain a vehicle validity result.

[0006] By adopting the above technical solution, through the management of hardware equipment and account allocation binding, the detection results of different accounts logged in on the mobile terminal are combined with the license plate image information submitted by the current user for identification and matching verification. By associating the license plate recognition of the image with the vehicle, the hardware information and the vehicle are matched, so as to achieve the vehicle verification result. According to the matching result, the vehicle is identified and the validity of the vehicle is known.

[0007] In a preferred example, the present application may be further configured as follows: the obtaining of the account verification request, performing identity recognition on the hardware device according to the account verification request, and obtaining the identity verification result specifically includes: Get GPS coordinates and identify the location of the electric bicycle; The user initiates the account verification request through the mobile terminal, matches the hardware device ID, and generates the identity verification result after the match is successful.

[0008] By adopting the above technical solution, the vehicle location can be obtained in real time, which is convenient and easy. At the same time, the account number and the hardware device ID are matched one by one, which increases the security of license plate recognition.

[0009] In a preferred example, the present application may be further configured as follows: generating license plate image recognition information according to the identity authentication result, and obtaining the license plate recognition result from the license plate image recognition information, specifically including: Constructing a high-resolution convolutional neural network model, training the high-resolution convolutional neural network model by inputting a low-resolution electric bicycle image and a high-resolution electric bicycle image, and converting the low-resolution electric bicycle image into the high-resolution electric bicycle image; When the identity authentication result is passed, the high-resolution electric bicycle image is acquired, and the license plate recognition result is obtained by using OCR image recognition technology.

[0010] By adopting the above technical solutions, the resolution and clarity of the image are improved through resolution technology, making the license plate characters clearer, thereby significantly improving the accuracy and reliability of OCR recognition. Clear license plate images help reduce blur and noise, reducing the probability of OCR systems misrecognizing characters. Super-resolution neural networks can effectively process low-quality images, improve the robustness of the system under different shooting conditions, and ensure high-quality recognition results in various environments.

[0011] In a preferred example, the present application can be further configured as follows: when the identity authentication result is passed, the high-resolution electric bicycle image is obtained, and the license plate recognition result is obtained by using the OCR image recognition technology, specifically including: using the OCR image recognition technology, establishing Tesseract OCR recognition information and CRNN model, wherein the CRNN model is constructed by a convolutional neural network model and a recursive neural network model; Recognize the high-resolution electric bicycle image through the Tesseract OCR recognition information to obtain a Tesseract recognition result, wherein the Tesseract recognition result includes a recognition character and a first confidence score; Inputting the high-resolution electric bicycle image into a CRNN model to obtain a CRNN recognition result, wherein the CRNN recognition result includes a character sequence and a second confidence score; A weighted vote is performed based on the first confidence score and the second confidence score to generate the license plate recognition result.

[0012] By adopting the above technical solution and combining the recognition results of the two models, the possible errors of a single model can be reduced, and the accuracy and credibility of recognition can be improved. Combining the recognition results of multiple models can enhance the robustness of the system and make the system more robust. By using weighted voting, the possible misrecognition rate of a single model can be effectively suppressed, and the reliability of the recognition results can be improved.

[0013] In a preferred example, the present application can be further configured as follows: the OCR image recognition technology is used to establish Tesseract OCR recognition information and a CRNN model, wherein the CRNN model is constructed by a convolutional neural network model and a recursive neural network model, specifically including: Extracting image features using the convolutional neural network model, and then serializing the features using the recursive neural network model to recognize a character sequence; The image features extracted by the convolutional neural network model are flattened to form a time step sequence, which is input into the recursive neural network model for processing to obtain a CTC loss function, which is used to calculate the loss between the predicted character sequence and the true label sequence to obtain the CRNN model.

[0014] By adopting the above technical solution, combining Tesseract OCR and CRNN models for license plate recognition, and using confidence scores for weighted voting to generate the final recognition results, the recognition accuracy and system robustness can be significantly improved. Tesseract and CRNN complement each other's respective advantages, improve the adaptability to different image quality and shooting conditions, and reduce recognition errors.

[0015] In a preferred example, the present application can be further configured as follows: performing weighted voting according to the first confidence score and the second confidence score to generate the license plate recognition result, specifically including: Normalize the confidence scores of Tesseract and CRNN so that they are in the range of 0-1; The first confidence score and the second confidence score are weighted to obtain a weighted score, the weighted scores of Tesseract and CRNN are compared, and the one with a higher score is selected as the license plate recognition result.

[0016] By adopting the above technical solution, the confidence scores of Tesseract and CRNN are standardized to the range of 0-1, and a weighted calculation is performed to obtain a comprehensive score, which can improve the consistency and reliability of the results and reduce errors.

[0017] In a preferred example, the present application may be further configured as follows: the obtaining of the license plate inspection and recognition standard, the license plate inspection and recognition standard identifying the license plate recognition result, obtaining the license plate verification result, identifying the vehicle according to the license plate verification result, and obtaining the vehicle validity result, specifically including: According to the license plate inspection and recognition standard, character format features, character quantity features and character legitimacy features are obtained, the license plate recognition result is recognized, and a license plate verification result is obtained; According to the license plate verification result, the vehicle is identified to obtain a vehicle validity result.

[0018] By adopting the above technical solution and through multiple recognition of the license plate, the error of license plate recognition is reduced.

[0019] The second object of the invention is achieved by the following technical solutions: A license plate verification device, the license plate verification device comprising: An identity verification module is used to obtain an account verification request, identify the hardware device according to the account verification request, and obtain an identity verification result; A license plate result module, used to generate license plate image recognition information according to the identity verification result, and obtain a license plate recognition result from the license plate image recognition information; The license plate recognition module is used to obtain a license plate inspection and recognition standard, the license plate inspection and recognition standard recognizes the license plate recognition result to obtain a license plate verification result, and identifies the vehicle according to the license plate verification result to obtain a vehicle validity result.

[0020] The third objective of the present application is achieved through the following technical solutions: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned license plate verification method when executing the computer program.

[0021] The fourth objective of the present application is achieved through the following technical solutions: A computer-readable storage medium stores a computer program, which implements the steps of the above-mentioned license plate verification method when executed by a processor.

[0022] In summary, the present application includes at least one of the following beneficial technical effects: 1. Through the management of hardware equipment inspection and account allocation binding, the detection results of different accounts logged in on the mobile terminal are combined with the license plate image information submitted by the current user for identification and matching verification. By associating the license plate recognition of the image with the vehicle, the hardware information and the vehicle are matched to achieve the vehicle verification result. The vehicle is identified based on the matching results to know the validity of the vehicle; 2. Improve the resolution and clarity of the image through resolution technology, making the license plate characters clearer, thereby significantly improving the accuracy and reliability of OCR recognition. Clear license plate images help reduce blur and noise, and reduce the probability of OCR systems misrecognizing characters. Super-resolution neural networks can effectively process low-quality images, improve the robustness of the system under different shooting conditions, and ensure high-quality recognition results in various environments; 3. By combining Tesseract OCR and CRNN models for license plate recognition and using confidence scores for weighted voting to generate the final recognition results, the recognition accuracy and system robustness can be significantly improved. The advantages of Tesseract and CRNN complement each other, improving the adaptability to different image quality and shooting conditions and reducing recognition errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flow chart of a license plate verification method in one embodiment of the present application; Figure 2 is a flowchart of the implementation of step S10 of the license plate verification method in one embodiment of the present application; Figure 3 is a flowchart for implementing step S20 of the license plate verification method in one embodiment of the present application; Figure 4 is a flowchart for implementing step S22 of the license plate verification method in one embodiment of the present application; Figure 5 is a flowchart of the implementation of step S221 of the license plate verification method in one embodiment of the present application; Figure 6 is a flowchart for implementing step S224 of the license plate verification method in one embodiment of the present application; Figure 7 is a flowchart for implementing step S30 of the license plate verification method in one embodiment of the present application; Figure 8 This is a principle block diagram of a license plate verification method in one embodiment of the present application; Fig. 9 It is a schematic diagram of a device in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The present application is further described in detail below in conjunction with the accompanying drawings.

[0025] In one embodiment, if Figure 1 As shown, the present application discloses a license plate verification method, which specifically includes the following steps: S10: obtaining an account verification request, performing identity identification on a hardware device according to the account verification request, and obtaining an identity verification result.

[0026] In this embodiment, the account verification request refers to the account password verification request sent through the mobile terminal. The hardware device refers to the device used by the user to identify the vehicle. The identity verification result refers to the result of whether the hardware device is associated with the account.

[0027] Specifically, by managing the hardware devices and matching them with the customer's accounts one by one, each hardware device corresponds to an account, and the account is verified and requested to log in to identify whether the identity matches the hardware device. If the match is successful, the mobile terminal can use the license plate verification function. If the match is unsuccessful, the license plate verification function cannot be used.

[0028] S20: Generate license plate image recognition information according to the identity authentication result, and obtain the license plate recognition result from the license plate image recognition information.

[0029] In this embodiment, the license plate image recognition information refers to the image features extracted and processed from the license plate image for identifying and verifying the license plate number. The license plate recognition result refers to the license plate number and related confidence information extracted from the license plate image through image processing and recognition technology.

[0030] Specifically, when the identity authentication is met, the user checks the electric bicycle license plate through the mobile terminal to obtain the electric bicycle image, processes the electric bicycle image in high definition, and trains the image through the Tesseract OCR recognition model and the CRNN model to obtain the license plate recognition result.

[0031] S30: Acquire a license plate inspection and recognition standard, identify the license plate recognition result by the license plate inspection and recognition standard, obtain a license plate verification result, identify the vehicle according to the license plate verification result, and obtain a vehicle validity result.

[0032] In this embodiment, the license plate verification standard refers to the criteria used to evaluate the accuracy and effectiveness of the license plate recognition result. The license plate verification result refers to whether the recognized license plate number is valid.

[0033] Specifically, the user first verifies the license plate by placing the hardware device close to the license plate, and then uses the license plate recognition result to identify whether the license plate number is a system license plate. If the license plate cannot be recognized, the license plate is recognized and confirmed again. After the license plate is identified as a system license plate, the hardware device is queried whether there is any return information. When the hardware device returns information within a specific time, the hardware device return information is associated with the license plate verification result. If the hardware device information is normal, the associated license plate number is normal. If the hardware device does not return information within a specific time, the associated license plate number is abnormal. If the hardware device information is abnormal, the associated license plate number is abnormal.

[0034] In this embodiment, the information returned by the hardware device refers to the hardware code, status and power level.

[0035] In one embodiment, if Figure 2 As shown, in step S10, an account verification request is obtained, and the hardware device is identified according to the account verification request to obtain an identity verification result, which specifically includes: S11: Obtain GPS coordinates and identify the location of the electric bicycle.

[0036] In this embodiment, the GPS coordinates refer to the location of the electric bicycle.

[0037] Specifically, when the user recognizes the license plate, the vehicle location is queried through GPS coordinates.

[0038] S12: The user initiates an account verification request through the mobile terminal and matches the hardware device ID. After the match is successful, an identity verification result is generated.

[0039] Specifically, the user initiates an account verification request through the mobile application. After receiving the request, the system obtains the unique hardware device ID of the device. The received device ID is matched with the device ID stored in the hardware device storage database. If the match is successful, the authentication result is generated as successful authentication, otherwise it is a failed authentication. The generated authentication result is returned to the mobile application, and the user interface is updated to display the authentication status.

[0040] It can be understood that the hardware device storage database refers to a database established for centralized management of hardware.

[0041] In one embodiment, if Figure 3 As shown, in step S20, the license plate image recognition information is generated according to the identity authentication result, and the license plate recognition result is obtained from the license plate image recognition information, which specifically includes: S21: construct a resolution convolutional neural network model, train the resolution convolutional neural network model by inputting low-resolution electric bicycle images and high-resolution electric bicycle images, and convert the low-resolution electric bicycle images into high-resolution electric bicycle images.

[0042] In this embodiment, the high-resolution convolutional neural network model refers to a model established to convert an image into a high-resolution image.

[0043] Specifically, low-resolution and high-resolution electric bicycle license plate image datasets are used as inputs to the resolution convolutional neural network model, the datasets are divided, training parameters are set, and the training set is used for model training and verification. The MSE loss function is used to identify the difference between the generated image and the real high-resolution image, and the model is optimized to improve the super-resolution effect. In order to achieve the conversion of low-resolution electric bicycle images into high-resolution images. The calculation formula of the MSE loss function is as follows: Where N represents the total number of pixels in the image. high (i) represents the gray value of the i-th pixel in the real high-resolution image. Represents the grayscale value of the i-th pixel in the generated image. By calculating the square difference of all pixels and taking the average value, when MSE is close to 0, the generated image is close to the real high-resolution image.

[0044] S22: When the identity authentication result is passed, a high-resolution image of the electric bicycle is obtained, and the license plate recognition result is obtained by using OCR image recognition technology.

[0045] In this embodiment, the OCR image recognition technology refers to recognizing the text content in the license plate image and converting it into text through image processing recognition technology.

[0046] Specifically, Tesseract OCR and a trained CRNN model are constructed through OCR image recognition technology, and Tesseract OCR recognition information and CRNN model are established through OCR image recognition technology, wherein the CRNN model is composed of a convolutional neural network (CNN) and a recurrent neural network (RNN). First, the input high-resolution electric bicycle license plate image is preprocessed, including grayscale, denoising, binarization and tilt correction, and then Tesseract OCR is used for preliminary recognition to obtain preliminary license plate recognition results. Then, the image is subjected to character recognition through the trained CRNN model to obtain more refined recognition results. Finally, the license plate recognition result is obtained by weighted voting on the Tesseract OCR recognition result and the CRNN model recognition result.

[0047] In this embodiment, Tesseract OCR refers to recognizing text in an image and converting it into editable text.

[0048] In one embodiment, if Figure 4As shown, in step S22, when the identity authentication result is passed, a high-resolution electric bicycle image is obtained, and the license plate recognition result is obtained by using OCR image recognition technology, which specifically includes: S221: Through OCR image recognition technology, Tesseract OCR recognition information and CRNN model are established, where the CRNN model is constructed through convolutional neural network model and recursive neural network model.

[0049] Specifically, a Tesseract OCR model and a CRNN model are established. The preprocessed electric bicycle license plate image is preliminarily recognized through Tesseract OCR to obtain preliminary recognition results. Then, a CRNN model is built, combining convolutional neural network (CNN) and recurrent neural network (RNN) to perform character recognition on the image.

[0050] S222: Recognize the high-resolution electric bicycle image through Tesseract OCR recognition information to obtain a Tesseract recognition result, which includes a recognized character and a first confidence score.

[0051] Specifically, Tesseract OCR is called for recognition, a high-resolution electric bicycle image is inputted into it, and the recognized characters and the first confidence score are obtained. Finally, the output of Tesseract is parsed to extract the recognition results, including the recognized characters and the first confidence score.

[0052] S223: Input the high-resolution electric bicycle image into the CRNN model to obtain a CRNN recognition result, which includes a character sequence and a second confidence score.

[0053] Specifically, the preprocessed image is input into the CRNN model for recognition, and the recognized character sequence and the corresponding second confidence score are output. Finally, the output of the CRNN model is parsed to extract the recognized character sequence and the second confidence score as the recognition result of the CRNN.

[0054] S224: Perform weighted voting based on the first confidence score and the second confidence score to generate a license plate recognition result.

[0055] Specifically, the first confidence score and the second confidence score are normalized to a range between 0 and 1 for weighted voting, and the first confidence score and the second confidence score are weighted to obtain a weighted confidence, and the weighted confidence = ω1×first confidence+ω2×second confidence, where ω1 and ω2 are the weights of the first confidence score and the second confidence score, respectively. The recognition result with higher confidence is selected as the license plate recognition result.

[0056] In one embodiment, if Figure 5 As shown, in step S221, Tesseract OCR recognition information and CRNN model are established by OCR image recognition technology, wherein the CRNN model is constructed by convolutional neural network model and recursive neural network model, specifically including: S2211: Extracting images using a convolutional neural network model In one embodiment, Figure 5 As shown, in step S221, the features are then serialized through a recursive neural network model to recognize character sequences.

[0057] Specifically, the image data of the electric bicycle license plate is obtained. A CRNN model is constructed, in which the CNN part extracts the local features of the image, and the RNN part processes these feature sequences, captures the temporal relationship between characters, and solves the character alignment problem through the CTC loss function. Then, the model is trained and evaluated through the training set and the validation set. During the recognition process, the image is preprocessed and feature extracted, and the feature sequence is processed by the RNN to output the character sequence, which is decoded by CTC to accurately recognize the license plate characters.

[0058] S2212: Flatten the image features extracted by the convolutional neural network model to form a time step sequence, input it into the recursive neural network model for processing, and obtain the CTC loss function. The CTC loss function is used to calculate the loss between the predicted character sequence and the true label sequence to obtain the CRNN model.

[0059] Specifically, a CRNN model is constructed, and the local features of the image are extracted using the CNN part. These features are then flattened to form a time step sequence and input into the RNN layer for processing to capture the temporal relationship between characters. The output of the RNN is mapped to a character set through a fully connected layer, and the CTC loss function is used to handle the alignment problem between the character sequence and the label sequence. During the training process, the input electric bicycle license plate image is preprocessed to extract features and form a time step sequence. After processing through the RNN, the character sequence is output and finally decoded using CTC.

[0060] In one embodiment, if Figure 5 As shown, in step S224, a weighted vote is performed based on the first confidence score and the second confidence score to generate a license plate recognition result, which specifically includes S2241: Normalize the confidence scores of Tesseract and CRNN so that the confidence scores are in the range of 0-1.

[0061] Specifically, get the original confidence scores of both. For Tesseract, divide its score by 100 for normalization; for CRNN, if its original score range is not 0-1, it needs to be divided by its maximum score value for normalization. The normalized confidence score of Tesseract is, The normalized confidence score of CRNN is, This normalization process ensures that all confidence scores are in the range of 0 to 1.

[0062] S2242: Perform weighted calculation on the first confidence score and the second confidence score to obtain a weighted score, compare the weighted scores of Tesseract and CRNN, and select the one with a higher score as the license plate recognition result.

[0063] Specifically, get the confidence score scoreTesseract of Tesseract and standardize it. Get the confidence score scoreCRNN of CRNN and standardize standardized_score CRNN =score CRNN . Determine the weighting coefficients of Tesseract and CRNN, recorded as ω Tesseract and ω CRNN , where ω Tesseract +ω CRNN = 1. Use the weighted coefficient to calculate the weighted score of Tesseract and CRNN, weighted_score Tesseract =ωTesseract×standardized_score Tesseract , weighted_score CRNN =ωCRNN×standardized_score CRNN .

[0064] Compare the weighted scores and select the result with the higher score as the license plate recognition result. final_result=Tesseract_result if weighted_score Tesseract >weighted_score CRNN , CRNN_result otherwise.

[0065] In one embodiment, if Figure 7As shown, in step S30, the license plate inspection and recognition standard is obtained, the license plate inspection and recognition standard recognizes the license plate recognition result, obtains the license plate verification result, identifies the vehicle according to the license plate verification result, and obtains the vehicle validity result, which specifically includes: S31: According to the license plate inspection and recognition standard, character format features, character quantity features and character legality features are obtained, and the license plate recognition result is recognized to obtain the license plate verification result.

[0066] Specifically, according to the license plate inspection and recognition standard, when the electric bicycle is a system license plate, the hardware device is queried to return information, and the information returned by the hardware device is matched and recognized with the license plate number to obtain the license plate verification result. S32: According to the license plate verification result, the vehicle is identified to obtain the vehicle validity result.

[0067] Specifically, according to the license plate verification result, the license plate is marked to identify whether the license plate is valid.

[0068] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0069] In one embodiment, a license plate verification device is provided, which corresponds one-to-one to the license plate verification method in the above embodiment. Figure 8 As shown, the license plate verification device includes an identity authentication module, a license plate result module and a license plate recognition module. The detailed description of each functional module is as follows: The identity verification module is used to obtain an account verification request, identify the hardware device according to the account verification request, and obtain an identity verification result; The license plate result module is used to generate license plate image recognition information according to the identity authentication result, and obtain the license plate recognition result from the license plate image recognition information; The license plate recognition module is used to obtain the license plate inspection and recognition standard, the license plate inspection and recognition standard identifies the license plate recognition result, obtains the license plate verification result, identifies the vehicle according to the license plate verification result, and obtains the vehicle validity result.

[0070] Optionally, authentication modules include: The positioning submodule is used to obtain GPS coordinates and identify the location of the electric bicycle; The identity authentication submodule is used by users to initiate account verification requests through mobile terminals, match hardware device IDs, and generate identity authentication results after successful matching.

[0071] Optionally, the license plate result module includes: A resolution convolutional neural network model submodule is used to construct a resolution convolutional neural network model, and train the resolution convolutional neural network model by inputting low-resolution electric bicycle images and high-resolution electric bicycle images to convert the low-resolution electric bicycle images into high-resolution electric bicycle images; The license plate recognition submodule is used to obtain a high-resolution image of the electric bicycle when the identity authentication result is passed, and obtain the license plate recognition result by using OCR image recognition technology.

[0072] Optionally, the license plate recognition submodule includes: The model component unit is used to establish Tesseract OCR recognition information and CRNN model through OCR image recognition technology, wherein the CRNN model is constructed through convolutional neural network model and recursive neural network model; The Tesseract recognition result unit is used to recognize the high-resolution electric bicycle image through Tesseract OCR recognition information to obtain Tesseract recognition results. The Tesseract recognition results include recognition characters and a first confidence score. A CRNN recognition result unit, used for inputting a high-resolution electric bicycle image into a CRNN model to obtain a CRNN recognition result, wherein the CRNN recognition result includes a character sequence and a second confidence score; The weighted voting unit is used to perform weighted voting according to the first confidence score and the second confidence score to generate a license plate recognition result.

[0073] Optionally, the model components include: A character sequence recognition subunit is used to extract image features using a convolutional neural network model, and then serialize the features through a recurrent neural network model to recognize character sequences; The CRNN model construction subunit is used to flatten the image features extracted by the convolutional neural network model to form a time step sequence, input it into the recursive neural network model for processing, and obtain the CTC loss function. The CTC loss function is used to calculate the loss between the predicted character sequence and the true label sequence to obtain the CRNN model.

[0074] Optionally, weighted voting units include: The range setting subunit is used to standardize the confidence scores of Tesseract and CRNN so that the confidence scores are in the range of 0-1; The weighted comparison subunit is used to perform weighted calculation on the first confidence score and the second confidence score to obtain a weighted score, compare the weighted scores of Tesseract and CRNN, and select the one with a higher score as the license plate recognition result.

[0075] Optionally, the license plate recognition module includes: The license plate verification result submodule is used to obtain the character format characteristics, character quantity characteristics and character legality characteristics according to the license plate inspection and recognition standards, identify the license plate recognition results, and obtain the license plate verification results; The vehicle validity result submodule is used to identify the vehicle according to the license plate verification result and obtain the vehicle validity result.

[0076] For the specific definition of the license plate verification device, please refer to the definition of the license plate verification method above, which will not be repeated here. Each module in the above license plate verification device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0077] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig. 9 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for the hardware device to store the database. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a license plate verification method is implemented.

[0078] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: Obtain an account verification request, identify the hardware device according to the account verification request, and obtain the identity verification result; Generate license plate image recognition information according to the identity verification result, and obtain the license plate recognition result from the license plate image recognition information; The license plate inspection and recognition standard is obtained, the license plate inspection and recognition standard recognizes the license plate recognition result, and the license plate verification result is obtained. According to the license plate verification result, the vehicle is identified to obtain the vehicle validity result.

[0079] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Obtain an account verification request, identify the hardware device according to the account verification request, and obtain the identity verification result; Based on the identity authentication result, the license plate image recognition information is generated, and the license plate recognition result is obtained from the license plate image recognition information; the license plate inspection and recognition standard is obtained, the license plate inspection and recognition standard identifies the license plate recognition result, and the license plate verification result is obtained. According to the license plate verification result, the vehicle is identified to obtain a vehicle validity result.

[0080] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile 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 (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), 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).

[0081] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0082] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A license plate verification method, characterized in that: The license plate verification method comprises: Obtain an account verification request, perform identity recognition on the hardware device according to the account verification request, and obtain an identity verification result; Generate license plate image recognition information according to the identity verification result, and obtain a license plate recognition result from the license plate image recognition information; A license plate inspection and recognition standard is obtained, the license plate inspection and recognition standard recognizes the license plate recognition result to obtain a license plate verification result, and a vehicle is identified according to the license plate verification result to obtain a vehicle validity result.

2. The license plate verification method according to claim 1, characterized in that: The obtaining of the account verification request and the identification of the hardware device according to the account verification request to obtain the identity verification result specifically include: Get GPS coordinates and identify the location of the electric bicycle; The user initiates the account verification request through the mobile terminal, matches the hardware device ID, and generates the identity verification result after the match is successful.

3. The license plate verification method according to claim 1, characterized in that: The step of generating license plate image recognition information according to the identity authentication result and obtaining the license plate recognition result from the license plate image recognition information specifically includes: Constructing a high-resolution convolutional neural network model, training the high-resolution convolutional neural network model by inputting a low-resolution electric bicycle image and a high-resolution electric bicycle image, and converting the low-resolution electric bicycle image into the high-resolution electric bicycle image; When the identity authentication result is passed, the high-resolution electric bicycle image is acquired, and the license plate recognition result is obtained by using OCR image recognition technology.

4. The license plate verification method according to claim 3, characterized in that: When the identity authentication result is passed, the high-resolution electric bicycle image is obtained, and the license plate recognition result is obtained by using OCR image recognition technology, which specifically includes: By using the OCR image recognition technology, Tesseract OCR recognition information and a CRNN model are established, wherein the CRNN model is constructed by a convolutional neural network model and a recursive neural network model; Recognize the high-resolution electric bicycle image through the Tesseract OCR recognition information to obtain a Tesseract recognition result, wherein the Tesseract recognition result includes a recognition character and a first confidence score; Inputting the high-resolution electric bicycle image into a CRNN model to obtain a CRNN recognition result, wherein the CRNN recognition result includes a character sequence and a second confidence score; A weighted vote is performed based on the first confidence score and the second confidence score to generate the license plate recognition result.

5. The license plate verification method according to claim 4, characterized in that: The OCR image recognition technology is used to establish Tesseract OCR recognition information and a CRNN model, wherein the CRNN model is constructed by a convolutional neural network model and a recursive neural network model, specifically including: Extracting image features using the convolutional neural network model, and then serializing the features using the recursive neural network model to recognize a character sequence; The image features extracted by the convolutional neural network model are flattened to form a time step sequence, which is input into the recursive neural network model for processing to obtain a CTC loss function, which is used to calculate the loss between the predicted character sequence and the true label sequence to obtain the CRNN model.

6. The license plate verification method according to claim 4, characterized in that: The step of performing weighted voting according to the first confidence score and the second confidence score to generate the license plate recognition result specifically includes: Normalize the confidence scores of Tesseract and CRNN so that they are in the range of 0-1; The first confidence score and the second confidence score are weighted to obtain a weighted score, the weighted scores of Tesseract and CRNN are compared, and the one with a higher score is selected as the license plate recognition result.

7. The license plate verification method according to claim 1, characterized in that: The obtaining of the license plate inspection and recognition standard, the license plate inspection and recognition standard identifying the license plate recognition result to obtain the license plate verification result, and the vehicle is identified according to the license plate verification result to obtain the vehicle validity result, specifically including: According to the license plate inspection and recognition standard, character format features, character quantity features and character legitimacy features are obtained, the license plate recognition result is recognized, and a license plate verification result is obtained; According to the license plate verification result, the vehicle is identified to obtain a vehicle validity result.

8. A license plate verification device, characterized in that: The license plate verification device comprises: An identity verification module, used to obtain an account verification request, identify the hardware device according to the account verification request, and obtain an identity verification result; A license plate result module, used to generate license plate image recognition information according to the identity verification result, and obtain a license plate recognition result from the license plate image recognition information; The license plate recognition module is used to obtain a license plate inspection and recognition standard, the license plate inspection and recognition standard recognizes the license plate recognition result to obtain a license plate verification result, and identifies the vehicle according to the license plate verification result to obtain a vehicle validity result.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the license plate verification method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the license plate verification method according to any one of claims 1 to 7 are implemented.

Images