Bill information identification method, apparatus and device, and storage medium

Abstract

The invention discloses a bill information identification method and device, equipment and a storage medium, and relates to the technical field of bill identification, and the method comprises the steps: inputting to-be-identified bill image data into different identification models, and obtaining the identification data information of a plurality of identification models; the identification data information of the multiple identification models is compared, and the editing distance of identification results of different identification models is calculated; and if the editing distance does not meet a preset similarity threshold value and the identification frequency is lower than a preset frequency threshold value, repeatedly identifying the to-be-identified bill image data to determine bill information. According to the method, the to-be-recognized bill image data is input into different recognition models, the comprehensiveness and accuracy of recognition can be remarkably improved in the face of complex and changeable bill information, differences of recognition results of the models are quantified through the editing distance, the differences of recognition of different models on different types of bill information can be determined, and the recognition accuracy of the different types of bill information is improved. And the bill information is identified through the model, so that the identification efficiency of the bill information is improved.

Description

Technical Field

[0001] The present application relates to the technical field of bill identification, and more specifically, to a bill information identification method, device, equipment and storage medium. Background Art

[0002] Bill recognition technology plays a vital role in modern financial management and auditing. Bills are not only the vouchers of transactions, but also an important basis for corporate financial records and audits. However, the types and formats of bills have gradually become diversified, including but not limited to invoices, receipts, bank acceptance bills, etc. Due to the diverse formats of bills, traditional bill recognition methods face many challenges.

[0003] In the prior art, a special recognition model is usually established for bills of a specific format, and different bills are input into the corresponding recognition model according to their bill formats to identify relevant bill information. In the case where the text on the bill may be difficult to recognize due to factors such as printing quality, handwriting, watermarks, seals or blurred images, manual recognition is generally required.

[0004] However, it may take a long time to classify the bill types and then use a special recognition model to identify them, resulting in low bill recognition efficiency and failure to adapt to changes in bill formats or new types of bills. At the same time, when faced with a large number of bills, manual bill recognition is also inefficient and increases the labor cost of bill recognition. Summary of the invention

[0005] In response to at least one defect or improvement need in the prior art, the present invention provides a method, device, equipment and storage medium for identifying bill information, which are used to solve the problem that when identifying bills through specific models in the prior art, it takes a long time to classify and cannot adapt to changes in bill formats or new types of bills. When manual identification is faced with a large number of bills, the recognition efficiency is also low and the labor cost of bill identification is increased.

[0006] To achieve the above object, according to a first aspect of the present invention, a method for identifying bill information is provided, comprising:

[0007] Input the image data of the bill to be identified into different recognition models to obtain recognition data information of multiple recognition models;

[0008] Compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models;

[0009] If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the to-be-recognized bill image data is repeatedly recognized to determine the bill information.

[0010] In a possible implementation, the recognition data information of multiple recognition models are compared, and the edit distances of the recognition results of different recognition models are calculated, which also includes:

[0011] Assign weights to each field in the identified data information through machine learning algorithms;

[0012] Convert each field in the recognition data information into a vector and calculate the cosine similarity between the two vectors;

[0013] The edit distance of the recognition results of different recognition models is calculated based on cosine similarity and weight ratio.

[0014] In a possible implementation, the edit distances of the recognition results of different recognition models are calculated based on the cosine similarity and the weight ratio, including:

[0015] Multiply the cosine similarity of each field by the corresponding weight ratio to get the weighted similarity;

[0016] Calculate the overall similarity of the recognition results of different recognition models based on weighted similarity and weight ratio;

[0017] The edit distance of the recognition results of different recognition models is determined based on the machine learning algorithm and the overall similarity.

[0018] In a possible implementation, the edit distances of the recognition results of different recognition models are determined according to the machine learning algorithm and the overall similarity, including:

[0019] Establish and train the edit distance prediction model based on the machine learning algorithm;

[0020] The overall similarity is input into a well-trained edit distance prediction model to obtain the edit distance of the recognition results of different recognition models.

[0021] In a possible implementation, the image data of the bill to be recognized is input into different recognition models to obtain recognition data information of multiple recognition models, including:

[0022] Determine the identification data type based on historical bill data;

[0023] Identify data information from the bill image data to be identified through different recognition models according to the type of identification data.

[0024] In a possible implementation, if the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the bill image data to be recognized is repeatedly recognized to determine the bill information, including:

[0025] Count the number of times the image data of the bill to be recognized is input into different recognition models;

[0026] If the number of recognitions reaches the preset threshold, the recognition of the bill image data to be recognized will be stopped and a warning signal will be issued.

[0027] In a possible implementation, if the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the to-be-recognized bill image data is repeatedly recognized to determine the bill information, further comprising:

[0028] If the edit distance exceeds a preset similarity threshold, the data information is identified as confirmed ticket information, and the identified data information is stored in a preset database.

[0029] According to a second aspect of the present invention, there is also provided a device for identifying bill information, comprising:

[0030] A data recognition module, which is configured to input the bill image data to be recognized into different recognition models to obtain recognition data information of multiple recognition models;

[0031] A comparison and calculation module, which is configured to compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models;

[0032] The judgment loop module is configured to repeatedly identify the bill image data to be identified to determine the bill information if the edit distance does not meet the preset similarity threshold and the number of identification times is lower than the preset number threshold.

[0033] According to the third aspect of the present invention, there is also provided a ticket information identification device, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of any one of the above-mentioned ticket information identification methods.

[0034] According to a fourth aspect of the present invention, there is also provided a storage medium storing a computer program executable by an access authentication device, wherein when the computer program runs on the access authentication device, the access authentication device executes the steps of any of the above-mentioned ticket information identification methods.

[0035] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0036] The present invention provides a method for identifying bill information. By inputting the bill image data to be identified into different recognition models, the advantages of each model in a specific recognition task can be fully utilized. Different models may be constructed based on different algorithms or training data. Therefore, when facing complex and changeable bill information, the use of multivariate models can significantly improve the comprehensiveness and accuracy of recognition. After obtaining the recognition data information of multiple recognition models, the method quantifies the differences in the recognition results of each model by comparing the results of different models and calculating the edit distance between them. When the edit distance of the recognition result does not meet the preset similarity threshold, it means that there is a large difference in the recognition results between different models. Through the repeated recognition mechanism, the bill image data is recognized again, and multiple attempts are made to reduce the impact of accidental errors and model limitations on the recognition results, thereby improving the stability and reliability of recognition. Therefore, the method effectively improves the accuracy and efficiency of bill information recognition by comprehensively using multivariate recognition models, comparing recognition results and setting a repeated recognition mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A schematic diagram of a flow chart of an embodiment of a method for identifying bill information provided by the present invention;

[0039] Figure 2 The present invention provides Figure 1 A schematic flow chart of an embodiment of step S102;

[0040] Figure 3 The present invention provides Figure 2 A schematic diagram of a flow chart of an embodiment of step S203;

[0041] Figure 4 A schematic diagram of the structure of an embodiment of a device for identifying bill information provided by the present invention;

[0042] Figure 5 A schematic diagram of the structure of a bill information recognition device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0044] The terms "first", "second", "third", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0045] The present invention provides a method, device, equipment and storage medium for identifying bill information, which are described below respectively.

[0046] See also Figure 1 , Figure 1 A flowchart of an embodiment of a method for identifying bill information provided by the present invention is provided. In a specific embodiment of the present invention, a method for identifying bill information is disclosed, including:

[0047] S101, inputting the bill image data to be recognized into different recognition models to obtain recognition data information of multiple recognition models;

[0048] S102, comparing the recognition data information of multiple recognition models, and calculating the edit distances of the recognition results of different recognition models;

[0049] S103: If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, repeatedly recognize the bill image data to be recognized to determine the bill information.

[0050] In the above embodiment, the bill image data to be identified is first input into a series of recognition models. These models may be based on different algorithm architectures, such as deep learning, machine learning or traditional image processing technology, aiming to use their respective advantages to fully capture the information on the bill. By processing the recognition tasks of multiple models in parallel, multiple recognition results can be obtained, providing rich materials for subsequent data comparison and verification. As a preferred embodiment, the models used in the present invention include but are not limited to multi-modal large models such as InternVL2-8B and InternLM-XComposer2.5.

[0051] Next, the recognition data obtained from multiple models are carefully compared, and the similarities and differences between them are quantified by calculating the edit distance between the recognition results of different models. Edit distance, as an indicator to measure the difference between strings, can intuitively reflect the degree of consistency between recognition results. Through edit distance, potential recognition errors or inconsistencies can be quickly identified, providing a basis for subsequent decision-making.

[0052] After obtaining the edit distance, if the calculated edit distance does not meet the preset similarity threshold, it means that there are large differences in the recognition results between different models, which may be caused by poor image quality, model performance limitations, or the complexity of the bill itself. At this time, it is necessary to check whether the current number of recognitions has reached the preset number threshold. If the number of recognitions is still below the threshold, the system will trigger a repeated recognition mechanism to process the bill image data again in order to obtain more accurate and consistent recognition results.

[0053] It should be noted that both the preset similarity threshold and the preset number threshold can be specifically set according to actual needs, and the present invention does not make any further limitations on this.

[0054] This repeated recognition process not only helps reduce the impact of accidental errors, but also overcomes the limitations of the model itself through multiple attempts, thereby further improving the accuracy and reliability of recognition. At the same time, through the preset similarity threshold and number threshold, the system can avoid unnecessary repeated calculations and waste of resources while ensuring recognition quality.

[0055] Compared with the prior art, the present embodiment provides a method for identifying bill information. By inputting the bill image data to be identified into different recognition models, the advantages of each model in a specific recognition task can be fully utilized. Different models may be constructed based on different algorithms or training data. Therefore, when facing complex and changeable bill information, the use of multivariate models can significantly improve the comprehensiveness and accuracy of recognition. After obtaining the recognition data information of multiple recognition models, the method quantifies the differences in the recognition results of each model by comparing the results of different models and calculating the edit distance between them. When the edit distance of the recognition result does not meet the preset similarity threshold, it means that there is a large difference in the recognition results between different models. Through the repeated recognition mechanism, the bill image data is recognized again, and multiple attempts are made to reduce the impact of accidental errors and model limitations on the recognition results, thereby improving the stability and reliability of recognition. Therefore, the present method effectively improves the accuracy and efficiency of bill information recognition by comprehensively using multivariate recognition models, comparing recognition results and setting a repeated recognition mechanism.

[0056] See also Figure 2 , Figure 2The present invention provides Figure 1 In some embodiments of the present invention, the recognition data information of multiple recognition models are compared, and the edit distances of the recognition results of different recognition models are calculated, which also includes:

[0057] S201, assigning a weight ratio to each field in the identification data information through a machine learning algorithm;

[0058] S202, converting each field in the recognition data information into a vector, and calculating the cosine similarity between two vectors;

[0059] S203, calculating the edit distance of the recognition results of different recognition models according to the cosine similarity and the weight ratio.

[0060] In the above embodiments, the importance of each field in the recognition data information is evaluated through a machine learning algorithm (such as a decision tree, random forest or neural network, etc.), and a weight ratio is assigned to each field. The weight ratio reflects the relative importance of each field in the overall recognition result, which helps to more accurately evaluate the differences between different recognition models in subsequent calculations.

[0061] Next, each field in the identification data information is converted into a vector representation, and the field content (such as text, numbers, etc.) is converted into a vector in a high-dimensional space through a certain encoding method (such as word embedding, TF-IDF, One-Hot encoding, etc.), and the vectorized field is convenient for mathematical operations and similarity comparison. It can be understood that converting the field into a vector representation can be achieved by existing technologies, and the present invention does not impose further restrictions on this.

[0062] After the field is vectorized, the cosine similarity between the two vectors is calculated. Cosine similarity is an indicator that measures the degree of similarity between two vectors in direction. The value range is between -1 and 1. When the directions of the two vectors are exactly the same, the cosine similarity is 1; when the directions are completely opposite, the cosine similarity is -1; when the two vectors are perpendicular, the cosine similarity is 0. By calculating the cosine similarity between the corresponding field vectors in the recognition results of different recognition models, the degree of similarity between them can be quantified.

[0063] Finally, based on the calculated cosine similarity and the previously assigned field weight ratio, the edit distance of the recognition results of different recognition models is calculated comprehensively, by multiplying the cosine similarity of each field by the corresponding weight ratio, and then taking a weighted sum of all fields. The result of the weighted sum can be used as a comprehensive indicator to measure the difference in recognition results of different recognition models, that is, a substitute or supplement for the edit distance. It should be noted that the "edit distance" here is no longer just the number of character replacement, insertion and deletion operations in the traditional sense, but a comprehensive score that combines field similarity and weight.

[0064] See also Figure 3 , Figure 3 The present invention provides Figure 2 In some embodiments of the present invention, the edit distance of the recognition results of different recognition models is calculated according to the cosine similarity and the weight ratio, including:

[0065] S301, multiplying the cosine similarity of each field by the corresponding weight ratio to obtain weighted similarity;

[0066] S302, calculating the overall similarity of recognition results of different recognition models based on weighted similarity and weight proportion;

[0067] S303: Determine the edit distance of the recognition results of different recognition models according to the machine learning algorithm and the overall similarity.

[0068] In the above embodiment, the cosine similarity of each field is multiplied by its corresponding weight ratio to obtain the weighted similarity of each field. The introduction of weight ratio can more accurately reflect the relative importance of different fields in the overall recognition result, and the weighted similarity not only considers the similarity between the field contents, but also incorporates the importance information of the field, providing more comprehensive data support for the subsequent overall similarity calculation.

[0069] After obtaining the weighted similarity of each field, it is necessary to calculate the overall similarity of the recognition results of different recognition models based on these weighted similarities and the corresponding weight ratios, and obtain a comprehensive indicator that can reflect the similarity of the overall recognition results by summing or weighted averaging the weighted similarities. The calculation process of the overall similarity needs to fully consider the weight ratio of each field to ensure the accuracy and reliability of the evaluation results.

[0070] The machine learning algorithm and overall similarity are used to determine the edit distance of the recognition results of different recognition models. The edit distance is a quantitative indicator that can comprehensively reflect the degree of difference in recognition results. Machine learning algorithms (such as regression models, decision trees, etc.) can predict or calculate the edit distance between the recognition results of different recognition models based on the overall similarity and known training data. By selecting a suitable machine learning algorithm and fully training and optimizing it, the accuracy and stability of the prediction results can be ensured.

[0071] It should be noted that in actual applications, the specific calculation method of the edit distance may vary depending on the application scenario and requirements. For example, in some scenarios, it may be more preferable to use a traditional edit distance that can directly reflect the number of character operations; while in other scenarios, it may be more preferable to use a quantitative indicator that can comprehensively reflect the degree of difference in recognition results as a substitute or supplement to the edit distance. Therefore, when determining the specific calculation method of the edit distance, it is necessary to fully consider the application scenario and requirements and choose an appropriate calculation method.

[0072] In some embodiments of the present invention, determining the edit distance of the recognition results of different recognition models according to the machine learning algorithm and the overall similarity includes:

[0073] Establish and train the edit distance prediction model based on the machine learning algorithm;

[0074] The overall similarity is input into a well-trained edit distance prediction model to obtain the edit distance of the recognition results of different recognition models.

[0075] In the above embodiment, an edit distance prediction model is established according to a machine learning algorithm (such as a support vector machine, a random forest, a neural network, etc.), and the historical data (including known overall similarity and corresponding edit distance) is used to train the edit distance prediction model so that it can learn the mapping relationship between overall similarity and edit distance. During the training process, the parameters of the model can be continuously adjusted until the prediction performance of the model reaches a satisfactory level.

[0076] After the model training is completed, the calculated overall similarity is input into the fully trained edit distance prediction model. Based on the input overall similarity and the learned mapping relationship, the edit distance between the recognition results of different recognition models is predicted. The predicted edit distance will serve as an important basis for subsequent decision-making and optimization.

[0077] In some embodiments of the present invention, the image data of the bill to be identified is input into different recognition models to obtain recognition data information of multiple recognition models, including:

[0078] Determine the identification data type based on historical bill data;

[0079] Identify data information from the bill image data to be identified through different recognition models according to the type of identification data.

[0080] In the above embodiment, based on the historical bill data, the identification data types that may be contained in the bill to be identified are analyzed and determined (such as 'machine number', 'invoice code', 'invoice number', 'invoice date', 'verification code', 'purchaser's name', 'purchaser's taxpayer identification number', 'purchaser's address and telephone number', 'opening bank and account number', 'goods or taxable services, service name', 'specification model', 'unit', 'quantity', 'unit price', 'amount', 'tax rate', 'tax amount', 'total price and tax (in uppercase)', 'total price and tax (in lowercase)', 'seller's name', 'seller's taxpayer identification number', 'seller's address and telephone number', 'seller's address and telephone number', 'opening bank and account number', 'remarks', 'payee', 'review', 'invoice issuer', etc.).

[0081] Next, according to the determined recognition data type, corresponding features are extracted from the image data of the bill to be recognized, and then these features are recognized from the bill to be recognized using different recognition models, thereby obtaining recognition data information of multiple recognition models.

[0082] In some embodiments of the present invention, if the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the process of repeatedly recognizing the to-be-recognized bill image data to determine the bill information includes:

[0083] Count the number of times the image data of the bill to be recognized is input into different recognition models;

[0084] If the number of recognitions reaches the preset threshold, the recognition of the bill image data to be recognized will be stopped and a warning signal will be issued.

[0085] In the above embodiment, during each recognition process, the number of times the image data of the bill to be recognized is input into different recognition models is counted. This statistical information helps to track the progress of the recognition process and determine whether it is necessary to continue to repeat the recognition. It should be noted that the number of times is not counted once when it is input into a model, but once each time the image data of the bill to be recognized is input into different recognition models, and the number of models input each time is the same.

[0086] If the number of recognitions reaches the preset threshold, the current recognition process is considered sufficient and no further recognition is required. At this point, the system will stop recognizing the image data of the bill to be recognized and issue a warning signal, prompting the user to check the bill quality or adjust the recognition parameters. Generally, recognition can be performed manually after the warning signal is issued.

[0087] In some embodiments of the present invention, if the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the process of repeatedly recognizing the to-be-recognized bill image data to determine the bill information further includes:

[0088] If the edit distance exceeds a preset similarity threshold, the data information is identified as confirmed ticket information, and the identified data information is stored in a preset database.

[0089] In the above embodiment, if during one recognition process, the calculated edit distance exceeds the preset similarity threshold, but the number of recognitions is still lower than the preset number threshold, then it is considered that although the current recognition result is different, there is still room for improvement. In this case, the recognition data information will not be immediately stored as the confirmed bill information, but will continue to be repeatedly recognized until the stop condition is met (such as the number of recognitions reaches the threshold or the edit distance meets the threshold).

[0090] Only when certain conditions are met (such as the edit distance meets the preset similarity threshold and the number of recognition times does not exceed the threshold, or the edit distance reaches a stable state after repeated recognition), the recognition data information will be stored in the preset database as the final ticket information, thereby ensuring the accuracy and reliability of the stored ticket information.

[0091] In order to better implement the method for identifying bill information in the embodiment of the present invention, based on the method for identifying bill information, correspondingly, please refer to Figure 4 , Figure 4 This is a schematic diagram of the structure of an embodiment of a device for identifying bill information provided by the present invention. The embodiment of the present invention provides a device for identifying bill information 400, including:

[0092] A data recognition module 410 is configured to input the to-be-recognized bill image data into different recognition models to obtain recognition data information of multiple recognition models;

[0093] A comparison and calculation module 420 is configured to compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models;

[0094] The judgment loop module 430 is configured to repeatedly identify the bill image data to be identified to determine the bill information if the edit distance does not meet the preset similarity threshold and the number of identification times is lower than the preset number threshold.

[0095] It should be noted here that: the device 400 provided in the above embodiment can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above method embodiments, which will not be repeated here.

[0096] See also Figure 5 , Figure 5 A schematic diagram of the structure of a bill information recognition device provided by an embodiment of the present invention. Based on the above-mentioned bill information recognition method, the present invention also provides a bill information recognition device, which can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palm computer, and a server. The bill information recognition device 500 includes a processor 510, a memory 520, and a display 530. Figure 5 Only some components of the ticket information recognition device are shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components may be implemented instead.

[0097] In some embodiments, the memory 520 may be an internal storage unit of the ticket information recognition device 500, such as a hard disk or memory of the ticket information recognition device 500. In other embodiments, the memory 520 may also be an external storage device of the ticket information recognition device 500, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the ticket information recognition device 500. Further, the memory 520 may also include both an internal storage unit and an external storage device of the ticket information recognition device 500. The memory 520 is used to store application software and various data installed in the ticket information recognition device 500, such as program codes for installing the ticket information recognition device 500. The memory 520 may also be used to temporarily store data that has been output or is to be output. In one embodiment, the memory 520 stores a ticket information recognition program 540, which can be executed by the processor 510, thereby realizing the ticket information recognition method of each embodiment of the present application.

[0098] In some embodiments, the processor 510 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes stored in the memory 520 or process data, such as executing a method for identifying bill information.

[0099] In some embodiments, the display 530 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 530 is used to display information on the ticket information recognition device 500 and to display a visual user interface. The components 510-530 of the ticket information recognition device 500 communicate with each other via a system bus.

[0100] In one embodiment, when the processor 510 executes the ticket information recognition program 540 in the memory 520 , the steps in the above ticket information recognition method are implemented.

[0101] This embodiment further provides a computer-readable storage medium on which a bill information recognition program is stored. When the bill information recognition program is executed by a processor, the following steps are implemented:

[0102] Input the image data of the bill to be identified into different recognition models to obtain recognition data information of multiple recognition models;

[0103] Compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models;

[0104] If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the to-be-recognized bill image data is repeatedly recognized to determine the bill information.

[0105] In summary, the present invention provides a method for identifying bill information. By inputting the bill image data to be identified into different recognition models, the advantages of each model in a specific recognition task can be fully utilized. Different models may be constructed based on different algorithms or training data. Therefore, when facing complex and changeable bill information, the use of multivariate models can significantly improve the comprehensiveness and accuracy of recognition. After obtaining the recognition data information of multiple recognition models, the method quantifies the differences in the recognition results of each model by comparing the results of different models and calculating the edit distance between them. When the edit distance of the recognition result does not meet the preset similarity threshold, it means that there is a large difference in the recognition results between different models. Through the repeated recognition mechanism, the bill image data is recognized again, and multiple attempts are made to reduce the impact of accidental errors and model limitations on the recognition results, thereby improving the stability and reliability of recognition. Therefore, this method effectively improves the accuracy and efficiency of bill information recognition by comprehensively using multivariate recognition models, comparing recognition results, and setting a repeated recognition mechanism.

[0106] The present application also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0107] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0108] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0109] In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0112] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk and other media that can store program code.

[0113] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0114] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0115] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0116] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying bill information, characterized in that: include: Input the image data of the bill to be identified into different recognition models to obtain recognition data information of multiple recognition models; Compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models; If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, the to-be-recognized bill image data is repeatedly recognized to determine the bill information.

2. The method for identifying bill information according to claim 1, characterized in that: The step of comparing the recognition data information of the multiple recognition models and calculating the edit distances of the recognition results of the different recognition models further includes: Assigning a weight ratio to each field in the identification data information through a machine learning algorithm; Convert each field in the identification data information into a vector, and calculate the cosine similarity between the two vectors; The edit distances of the recognition results of different recognition models are calculated according to the cosine similarity and the weight ratio.

3. The method for identifying bill information according to claim 2, characterized in that: The calculating the edit distance of the recognition results of different recognition models according to the cosine similarity and the weight ratio includes: Multiply the cosine similarity of each field by the corresponding weight ratio to get the weighted similarity; Calculating the overall similarity of recognition results of different recognition models based on the weighted similarity and the weight proportion; The edit distances of the recognition results of different recognition models are determined according to the machine learning algorithm and the overall similarity.

4. The method for identifying bill information according to claim 3, characterized in that: The determining the edit distance of the recognition results of different recognition models according to the machine learning algorithm and the overall similarity includes: Establishing and training an edit distance prediction model according to the machine learning algorithm; The overall similarity is input into a well-trained edit distance prediction model to obtain the edit distances of recognition results of different recognition models.

5. The method for identifying bill information according to claim 1, characterized in that: The step of inputting the to-be-recognized bill image data into different recognition models to obtain recognition data information of multiple recognition models includes: Determine the identification data type based on historical bill data; The identification data information is identified from the to-be-identified bill image data according to the identification data type through different identification models.

6. The method for identifying bill information according to claim 1, characterized in that: If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number threshold, repeatedly recognizing the to-be-recognized bill image data to determine the bill information includes: Counting the number of times the to-be-recognized bill image data is input into different recognition models; If the number of recognitions reaches the preset number threshold, the recognition of the to-be-recognized bill image data is stopped and a warning signal is issued.

7. The method for identifying bill information according to claim 1, characterized in that: If the edit distance does not meet the preset similarity threshold and the number of recognitions is lower than the preset number of thresholds, the to-be-recognized bill image data is repeatedly recognized to determine the bill information, further comprising: If the edit distance exceeds the preset similarity threshold, the identification data information is confirmed ticket information, and the identification data information is stored in a preset database.

8. A device for identifying bill information, characterized in that: include: A data recognition module, which is configured to input the bill image data to be recognized into different recognition models to obtain recognition data information of multiple recognition models; A comparison and calculation module, which is configured to compare the recognition data information of multiple recognition models and calculate the edit distance of the recognition results of different recognition models; The judgment loop module is configured to repeatedly identify the to-be-identified bill image data to determine the bill information if the edit distance does not meet a preset similarity threshold and the number of identification times is lower than a preset number threshold.

9. A bill information recognition device, characterized in that: It comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of the ticket information identification method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: It stores a computer program executable by an access authentication device. When the computer program runs on the access authentication device, the access authentication device executes the steps of the ticket information recognition method according to any one of claims 1 to 7.

Images