Financial bill management system and implementation method thereof

Abstract

The invention relates to a financial bill management system and an implementation method thereof. The financial bill management system comprises a data acquisition module, an AI identification module, an auditing module, a cost reimbursement module, a report generation module and a data storage and backup module. According to the financial bill management system and the implementation method thereof, by introducing the AI recognition module, automatic recognition and classification of bill types are achieved, the requirement for manual operation is reduced, the working efficiency and accuracy are improved, the AI recognition module can automatically process bill information, the whole process from image preprocessing to final key information extraction is highly automatic, and the efficiency is high. The system is not limited to bill identification and classification, but also covers a plurality of functional modules such as data acquisition, auditing, expense reimbursement processing, report generation, data storage and backup and the like, and provides a one-stop solution for users. The image preprocessing unit can effectively remove image noise, correct inclined images, cut redundant parts and enhance image contrast and brightness.

Description

Technical Field

[0001] The present invention relates to the technical field of financial bill management, and in particular to a financial bill management system and an implementation method thereof. Background Art

[0002] Financial receipts refer to various types of vouchers generated by enterprises in economic activities, which are used to record and prove the occurrence of financial transactions. These receipts are the basis of accounting and financial management, and are also an important basis for tax declaration and auditing. Receipts provide original vouchers for public institutions to conduct financial statistics and accounting. Receipt statistical management is also an important legal voucher for the income source of public institutions.

[0003] Chinese patent CN118172182A discloses a financial bill statistics management system based on data analysis, which relates to the technical field of financial bill statistics and solves the technical problems in the prior art that reasonable screening cannot be performed during the financial bill statistics process and security performance control cannot be performed on the financial bill storage. Specifically, the bill statistics screening unit screens the financial bill statistics, divides the financial bills into abnormal bills and normal bills through statistical screening, and sends them to the server; the bill storage security control unit performs security control on the storage of financial bills; the server generates a bill statistics deviation control signal and sends the bill statistics deviation control signal to the bill statistics deviation control unit, and after receiving the bill statistics deviation control signal, the bill statistics deviation control unit performs bill statistics deviation analysis and control on the financial bills.

[0004] The above application has the same problems as the prior art financial bill management system. During the financial processing process, the classification and identification of bills mainly rely on manual operations. Financial personnel need to manually check each bill, determine its type such as invoice, receipt, check, etc., and identify key information thereon such as invoice code, invoice number, amount, date, etc. Secondly, when manually entering bill information, financial personnel need to manually input the identified key information into the financial system. Due to the influence of human factors, entry errors are prone to occur. At the same time, manual processing cannot efficiently batch process a large number of bills. Each bill needs to be processed separately, and automation and batch operations cannot be achieved. Therefore, this application uses AI recognition to manage financial bills to solve the above problems. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a financial bill management system and an implementation method thereof, which has the advantages of realizing automatic identification and classification of bill types, and solves the problem of easy entry errors due to human factors.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a financial bill management system, including a data acquisition module, an AI recognition module, an audit module, an expense reimbursement module, a report generation module and a data storage and backup module;

[0007] in,

[0008] Data collection module: responsible for collecting bill information from different channels;

[0009] AI recognition module: automatically identifies the bill type and classifies the bill;

[0010] Audit module: review the compliance and authenticity of collected bills;

[0011] Expense reimbursement module: processes employee expense reimbursement requests;

[0012] Report generation module: Generate various financial reports based on bill data;

[0013] Data storage and backup module: responsible for the safe storage and regular backup of bills and related data;

[0014] The AI ​​recognition module includes an image preprocessing unit, a text recognition (OCR) unit, an image classification unit, and a key information extraction unit;

[0015] in,

[0016] Image preprocessing unit: preprocess the collected bill image;

[0017] Text recognition (OCR) unit: uses optical character recognition (OCR) technology to extract text information from images;

[0018] Image classification unit: classifies bill images and identifies the type of bill;

[0019] Key information extraction unit: extracts key information from the recognized text.

[0020] Furthermore, the AI ​​recognition module also includes a data verification unit, an exception handling unit and a user interaction unit. The data verification unit verifies the extracted ticket information to ensure the accuracy and completeness of the data; the exception handling unit handles abnormal situations that occur during the recognition process to ensure the stable operation of the system; the user interaction unit is used to provide a user interface for users to view and manage recognition results.

[0021] Further, the image preprocessing unit workflow is as follows:

[0022] S1 removes noise from the image and improves image quality;

[0023] S2 corrects the tilted image and restores it to a normal angle;

[0024] S3 cuts out the redundant parts of the image and retains the core area of ​​the bill;

[0025] S4 enhances the contrast and brightness of images, improving the clarity of text and graphics.

[0026] Furthermore, the character recognition (OCR) unit includes the following contents:

[0027] Character segmentation: separate the text in the image by lines, separate the text in each line by characters, detect the connected areas in the image, and treat each character as an independent area;

[0028] Character recognition: extract the geometric features, pixel features, and texture features of characters, match the extracted features with predefined templates, and find the most similar characters. Use machine learning algorithms to classify features and recognize characters. Use deep learning models to perform end-to-end recognition of images, extract features directly from images, and classify them.

[0029] Language processing: supports recognition of multiple languages, including but not limited to Chinese, English, Japanese and Korean;

[0030] Context understanding: Use the language model to verify and correct the recognition results, combine the context information to perform logical verification on the recognition results, use the spelling correction algorithm to perform spelling correction on the recognition results, correct possible spelling errors, and use the dictionary for matching.

[0031] Furthermore, the specific tasks of the image classification unit are:

[0032] Feature extraction: extract key features from an image, including color, shape, and texture;

[0033] Classification model: Use machine learning algorithms to classify features;

[0034] Multi-level classification: Support multi-level classification, first do rough classification, including invoices, receipts, and then do detailed classification, including special VAT invoices, ordinary invoices,

[0035] Self-learning: Supports self-learning and optimization of the model to continuously improve classification accuracy.

[0036] Further, the key information extraction unit comprises the following steps:

[0037] S1 Keyword matching: Use keyword matching technology to locate and extract key information;

[0038] S2 Regular Expression: Use regular expressions to extract information in a specific format;

[0039] S3 Natural Language Processing: Use natural language processing technology to understand the semantics of text and extract key information;

[0040] S4 Rule Engine: Design a rule engine to extract key information based on preset rules.

[0041] Further, the keyword matching technology includes the following steps:

[0042] S1 Predefined keyword library: collects keywords related to target information, including but not limited to invoice code, invoice number, amount and date;

[0043] S2 Text preprocessing: remove noise from the text, including but not limited to punctuation and special characters, convert the text into a unified format, and segment the text into words or phrases;

[0044] S3 keyword matching: Use string matching methods to find out whether predefined keywords exist in the text, use regular expressions to match specific patterns, extract qualified information, and use multiple keywords to match and extract information from multiple fields.

[0045] S4 result verification: Check whether the extracted information meets the preset format requirements, check whether the extracted information is logical, compare the extracted information with the existing data, and verify its authenticity and consistency.

[0046] Furthermore, the specific steps and implementation of character recognition using a support vector machine (SVM) in the image classification unit are as follows. The goal of the support vector machine is to find an optimal hyperplane in the feature space so that the intervals between sample points of different categories are maximized. This hyperplane can be expressed by the following formula:

[0047] w·x+b=0,

[0048] Among them, w is the weight vector, x is the input feature vector, and b is the bias term. In order to find the optimal hyperplane, SVM is implemented by minimizing the following objective function:

[0049]

[0050] At the same time, the following constraints are met:

[0051] y i (w·x i +b)≥1, i=1, 2,..., n,

[0052] Among them, yi is the label of the i-th sample (+1 or -1), xi is the feature vector of the i-th sample, and n is the total number of samples.

[0053] Furthermore, the data collection module may collect data in ways including but not limited to paper bill scanning, electronic invoice import, bank statement connection and third-party platform connection.

[0054] The present application further proposes an implementation method of a financial bill management system, including the financial bill management system described above, comprising the following steps:

[0055] 1) The data collection module collects bill information from different channels;

[0056] 2) The image preprocessing unit preprocesses the collected bill image;

[0057] 3) The text recognition (OCR) unit uses optical character recognition (OCR) technology to extract text information from the image;

[0058] 4) The image classification unit classifies the bill image and identifies the type of bill;

[0059] 5) The key information extraction unit extracts key information from the recognized text.

[0060] Compared with the prior art, the technical solution of this application has the following beneficial effects:

[0061] The financial bill management system and its implementation method, by introducing an AI recognition module, realize automatic recognition and classification of bill types, reduce the need for manual operation, and improve work efficiency and accuracy. The AI ​​recognition module can automatically process bill information, from image preprocessing to the final extraction of key information, and the entire process is highly automated. The system is not limited to the recognition and classification of bills, but also covers multiple functional modules such as data collection, review, expense reimbursement processing, report generation, and data storage and backup, providing users with a one-stop solution. The image preprocessing unit can effectively remove image noise, correct tilted images, crop redundant parts, and enhance image contrast and brightness. These measures significantly improve the accuracy of subsequent text recognition. It adopts a combination of OCR technology and deep learning models to support multi-language recognition, and can perform logical verification and spelling correction in combination with context information, thereby improving the accuracy and robustness of text recognition. The system's built-in data verification unit and exception handling unit can automatically check the accuracy and completeness of the data, and handle abnormal situations that may occur during the recognition process to ensure the stable operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 The figure is a schematic diagram of the implementation flow of the system of the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] See also Figure 1 , a financial bill management system in this embodiment includes a data acquisition module, an AI recognition module, an audit module, an expense reimbursement module, a report generation module and a data storage and backup module;

[0065] in,

[0066] Data collection module: responsible for collecting bill information from different channels;

[0067] AI recognition module: automatically identifies the bill type and classifies the bill;

[0068] Audit module: review the compliance and authenticity of collected bills;

[0069] Expense reimbursement module: processes employee expense reimbursement requests;

[0070] Report generation module: Generate various financial reports based on bill data;

[0071] Data storage and backup module: responsible for the safe storage and regular backup of bills and related data;

[0072] The AI ​​recognition module includes an image preprocessing unit, a text recognition (OCR) unit, an image classification unit, and a key information extraction unit;

[0073] in,

[0074] Image preprocessing unit: preprocess the collected bill image;

[0075] Text recognition (OCR) unit: uses optical character recognition (OCR) technology to extract text information from images;

[0076] Image classification unit: classifies bill images and identifies the type of bill;

[0077] Key information extraction unit: extracts key information from the recognized text.

[0078] Among them, the AI ​​recognition module also includes a data verification unit, an exception handling unit and a user interaction unit. The data verification unit verifies the extracted ticket information to ensure the accuracy and completeness of the data; the exception handling unit handles abnormal situations that occur during the recognition process to ensure the stable operation of the system; the user interaction unit is used to provide a user interface for users to view and manage recognition results.

[0079] Among them, the workflow of the image preprocessing unit is as follows:

[0080] S1 removes noise from the image and improves image quality;

[0081] S2 corrects the tilted image and restores it to a normal angle;

[0082] S3 cuts out the redundant parts of the image and retains the core area of ​​the bill;

[0083] S4 enhances the contrast and brightness of images, improving the clarity of text and graphics.

[0084] It should be noted that the image quality of bill images can be significantly improved through these four steps of image preprocessing. The specific advantages include:

[0085] 1. Reduce noise interference and improve image cleanliness.

[0086] 2. Correct the tilt to ensure the text is in the correct direction.

[0087] 3. Cut out unnecessary parts, reduce the amount of calculation, and focus on key information.

[0088] 4. Enhance contrast and brightness, and improve the clarity of text and patterns.

[0089] Among them, the character recognition (OCR) unit includes the following contents:

[0090] Character segmentation: separate the text in the image by lines, separate the text in each line by characters, detect the connected areas in the image, and treat each character as an independent area;

[0091] Character recognition: extract the geometric features, pixel features, and texture features of characters, match the extracted features with predefined templates, and find the most similar characters. Use machine learning algorithms to classify features and recognize characters. Use deep learning models to perform end-to-end recognition of images, extract features directly from images, and classify them.

[0092] Language processing: supports recognition of multiple languages, including but not limited to Chinese, English, Japanese and Korean;

[0093] Context understanding: Use the language model to verify and correct the recognition results, combine the context information to perform logical verification on the recognition results, use the spelling correction algorithm to perform spelling correction on the recognition results, correct possible spelling errors, and use the dictionary for matching.

[0094] Among them, the specific tasks of the image classification unit are:

[0095] Feature extraction: extract key features from an image, including color, shape, and texture;

[0096] Classification model: Use machine learning algorithms to classify features;

[0097] Multi-level classification: Support multi-level classification, first do rough classification, including invoices, receipts, and then do detailed classification, including special VAT invoices, ordinary invoices,

[0098] Self-learning: Supports self-learning and optimization of the model to continuously improve classification accuracy.

[0099] It should be noted that through the above four tasks, the image classification unit plays an important role in the financial bill management system. Its main advantages include:

[0100] 1. Comprehensiveness: Extract multiple features to improve classification accuracy.

[0101] 2. High Accuracy: Use advanced machine learning algorithms to ensure high accuracy of classification.

[0102] 3. Hierarchical classification: Multi-level classification method improves the efficiency and accuracy of classification.

[0103] 4. Continuous optimization: The self-learning mechanism enables the system to continuously optimize, adapt to changes, and maintain high performance.

[0104] The key information extraction unit includes the following steps:

[0105] S1 Keyword matching: Use keyword matching technology to locate and extract key information;

[0106] S2 Regular Expression: Use regular expressions to extract information in a specific format;

[0107] S3 Natural Language Processing: Use natural language processing technology to understand the semantics of text and extract key information;

[0108] S4 Rule Engine: Design a rule engine to extract key information based on preset rules.

[0109] It should be noted that, by combining the advantages of these steps, the key information extraction unit can accurately and efficiently extract key information from complex bill texts, while maintaining the flexibility and adaptability of the system to meet the information extraction needs in different business scenarios.

[0110] Among them, the keyword matching technology includes the following steps:

[0111] S1 Predefined keyword library: collects keywords related to target information, including but not limited to invoice code, invoice number, amount and date;

[0112] S2 Text preprocessing: remove noise from the text, including but not limited to punctuation and special characters, convert the text into a unified format, and segment the text into words or phrases;

[0113] S3 keyword matching: Use string matching methods to find out whether predefined keywords exist in the text, use regular expressions to match specific patterns, extract qualified information, and use multiple keywords to match and extract information from multiple fields.

[0114] S4 result verification: Check whether the extracted information meets the preset format requirements, check whether the extracted information is logical, compare the extracted information with the existing data, and verify its authenticity and consistency.

[0115] In summary, keyword matching technology can efficiently and accurately extract key information from large amounts of text through a series of carefully designed steps, while ensuring the accuracy and reliability of the information through result verification. This is especially important for application scenarios that require processing large amounts of text data, such as financial invoice management systems.

[0116] Among them, the specific steps and implementation of using support vector machine (SVM) for character recognition in the image classification unit. The goal of the support vector machine is to find an optimal hyperplane in the feature space to maximize the interval between sample points of different categories. This hyperplane can be expressed by the following formula:

[0117] w·x+b=0,

[0118] Among them, w is the weight vector, x is the input feature vector, and b is the bias term. In order to find the optimal hyperplane, SVM is implemented by minimizing the following objective function:

[0119]

[0120] At the same time, the following constraints are met:

[0121] y i (w·x i +b)≥1, i=1, 2,..., n,

[0122] Among them, yi is the label of the i-th sample (+1 or -1), xi is the feature vector of the i-th sample, and n is the total number of samples.

[0123] Among them, the data collection module's collection methods include but are not limited to paper bill scanning, electronic invoice import, bank statement docking and third-party platform docking.

[0124] The present application further proposes an implementation method of a financial bill management system, including a financial bill management system, comprising the following steps:

[0125] 1) The data collection module collects bill information from different channels;

[0126] 2) The image preprocessing unit preprocesses the collected bill image;

[0127] 3) The text recognition (OCR) unit uses optical character recognition (OCR) technology to extract text information from the image;

[0128] 4) The image classification unit classifies the bill image and identifies the type of bill;

[0129] 5) The key information extraction unit extracts key information from the recognized text.

[0130] In summary, through the above five steps, the financial bill management system realizes the full process automation management from data collection to key information extraction. Its main advantages include:

[0131] Diversity: Supports multiple data collection methods to adapt to different scenarios and needs.

[0132] High-quality images: Improve image quality and recognition accuracy through image preprocessing.

[0133] High Accuracy: Utilizes advanced OCR technology and image classification technology to achieve high-precision text recognition and classification.

[0134] Automation: The entire process is automated to reduce manual operations and improve efficiency.

[0135] Self-learning capability: supports self-learning and optimization of the model to continuously improve the performance and accuracy of the system.

[0136] These advantages not only improve the efficiency and accuracy of financial bill processing, but also enhance the robustness and adaptability of the system, ensuring the efficient operation of the system.

[0137] The beneficial effects of the above embodiments are as follows: by introducing the AI ​​recognition module, automatic recognition and classification of bill types are realized, the need for manual operation is reduced, and work efficiency and accuracy are improved. The AI ​​recognition module can automatically process bill information, and the entire process from image preprocessing to final key information extraction is highly automated. The system is not limited to bill recognition and classification, but also covers multiple functional modules such as data collection, review, expense reimbursement processing, report generation, and data storage and backup, providing users with a one-stop solution. The image preprocessing unit can effectively remove image noise, correct tilted images, crop redundant parts, and enhance image contrast and brightness. These measures significantly improve the accuracy of subsequent text recognition. The combination of OCR technology and deep learning models supports multi-language recognition, and can perform logic verification and spelling correction in combination with context information, which improves the accuracy and robustness of text recognition. The system's built-in data verification unit and exception handling unit can automatically check the accuracy and completeness of the data, and handle abnormal situations that may occur during the recognition process to ensure the stable operation of the system. .

[0138] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0139] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A financial bill management system, characterized in that: It includes data collection module, AI recognition module, audit module, expense reimbursement module, report generation module and data storage and backup module; in, Data collection module: responsible for collecting bill information from different channels; AI recognition module: automatically identifies the bill type and classifies the bill; Audit module: review the compliance and authenticity of collected bills; Expense reimbursement module: processes employee expense reimbursement requests; Report generation module: Generate various financial reports based on bill data; Data storage and backup module: responsible for the safe storage and regular backup of bills and related data; The AI ​​recognition module includes an image preprocessing unit, a text recognition (OCR) unit, an image classification unit, and a key information extraction unit; in, Image preprocessing unit: preprocess the collected bill image; Text recognition (OCR) unit: uses optical character recognition (OCR) technology to extract text information from images; Image classification unit: classifies bill images and identifies the type of bill; Key information extraction unit: extracts key information from the recognized text.

2. A financial bill management system according to claim 1, characterized in that: The AI ​​recognition module also includes a data verification unit, an exception handling unit and a user interaction unit. The data verification unit verifies the extracted ticket information to ensure the accuracy and completeness of the data; the exception handling unit handles abnormal situations that occur during the recognition process to ensure the stable operation of the system; the user interaction unit is used to provide a user interface for users to view and manage recognition results.

3. A financial bill management system according to claim 1, characterized in that: The image preprocessing unit workflow is as follows: S1 removes noise from the image and improves image quality; S2 corrects the tilted image and restores it to a normal angle; S3 cuts out the redundant parts of the image and retains the core area of ​​the bill; S4 enhances the contrast and brightness of images, improving the clarity of text and graphics.

4. A financial bill management system according to claim 1, characterized in that: The character recognition (OCR) unit includes the following contents: Character segmentation: separate the text in the image by lines, separate the text in each line by characters, detect the connected areas in the image, and treat each character as an independent area; Character recognition: extract the geometric features, pixel features, and texture features of characters, match the extracted features with predefined templates, and find the most similar characters. Use machine learning algorithms to classify features and recognize characters. Use deep learning models to perform end-to-end recognition of images, extract features directly from images, and classify them. Language processing: supports recognition of multiple languages, including but not limited to Chinese, English, Japanese and Korean; Context understanding: Use the language model to verify and correct the recognition results, combine the context information to perform logical verification on the recognition results, use the spelling correction algorithm to perform spelling correction on the recognition results, correct possible spelling errors, and use the dictionary for matching.

5. A financial bill management system according to claim 1, characterized in that: The specific tasks of the image classification unit are: Feature extraction: extract key features from an image, including color, shape, and texture; Classification model: Use machine learning algorithms to classify features; Multi-level classification: Support multi-level classification, first do rough classification, including invoices, receipts, and then do detailed classification, including special VAT invoices, ordinary invoices, Self-learning: Supports self-learning and optimization of the model to continuously improve classification accuracy.

6. A financial bill management system according to claim 1, characterized in that: The key information extraction unit comprises the following steps: S1 Keyword matching: Use keyword matching technology to locate and extract key information; S2 Regular Expression: Use regular expressions to extract information in a specific format; S3 Natural Language Processing: Use natural language processing technology to understand the semantics of text and extract key information; S4 Rule Engine: Design a rule engine to extract key information based on preset rules.

7. A financial bill management system according to claim 6, characterized in that: The keyword matching technique comprises the following steps: S1 Predefined keyword library: collects keywords related to target information, including but not limited to invoice code, invoice number, amount and date; S2 Text preprocessing: remove noise from the text, including but not limited to punctuation and special characters, convert the text into a unified format, and segment the text into words or phrases; S3 keyword matching: Use string matching methods to find out whether predefined keywords exist in the text, use regular expressions to match specific patterns, extract qualified information, and use multiple keywords to match and extract information from multiple fields. S4 result verification: Check whether the extracted information meets the preset format requirements, check whether the extracted information is logical, compare the extracted information with the existing data, and verify its authenticity and consistency.

8. A financial bill management system according to claim 1, characterized in that: The specific steps and implementation of character recognition using a support vector machine (SVM) in the image classification unit. The goal of the support vector machine is to find an optimal hyperplane in the feature space so that the intervals between sample points of different categories are maximized. This hyperplane can be expressed by the following formula: w·x+b=0, Among them, w is the weight vector, x is the input feature vector, and b is the bias term. In order to find the optimal hyperplane, SVM is implemented by minimizing the following objective function: At the same time, the following constraints are met: y i (w·x i +b)≥1,i=1,2,...,n, Among them, yi is the label of the i-th sample (+1 or -1), xi is the feature vector of the i-th sample, and n is the total number of samples.

9. A financial bill management system according to claim 1, characterized in that: The data collection module collection methods include but are not limited to paper bill scanning, electronic invoice import, bank statement connection and third-party platform connection.

10. A method for implementing a financial bill management system comprises a financial bill management system as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: 1) The data collection module collects bill information from different channels; 2) The image preprocessing unit preprocesses the collected bill image; 3) The text recognition (OCR) unit uses optical character recognition (OCR) technology to extract text information from the image; 4) The image classification unit classifies the bill image and identifies the type of bill; 5) The key information extraction unit extracts key information from the recognized text.

Images