CN113822269A - Paper deposit receipt automatic input method based on image recognition - Google Patents
Paper deposit receipt automatic input method based on image recognition Download PDFInfo
- Publication number
- CN113822269A CN113822269A CN202111390231.4A CN202111390231A CN113822269A CN 113822269 A CN113822269 A CN 113822269A CN 202111390231 A CN202111390231 A CN 202111390231A CN 113822269 A CN113822269 A CN 113822269A
- Authority
- CN
- China
- Prior art keywords
- deposit receipt
- image
- receipt image
- deposit
- correction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The invention discloses an automatic paper deposit receipt input method based on image recognition. The invention comprises an acquisition module, a storage module, a deposit receipt image recognition model and a re-correction module. The system server is provided with a task triggering program, when a storage module of the server receives a new deposit receipt image obtained by scanning of the acquisition module, the task triggering program firstly transmits the new deposit receipt image to the deposit receipt image identification model, acquires an identification result through the deposit receipt image identification model, and fills the identification result in a corresponding position in a service system or a form. The method mainly solves the problem that the daily management of enterprises regularly stores the bills, realizes automatic picture identification and automatic information input into a business system by applying an artificial intelligence algorithm and simultaneously realizing the electronic storage of the paper bills in a model training mode. Thereby greatly improving the working efficiency of customers.
Description
Technical Field
The invention particularly relates to an automatic paper deposit receipt input method based on image recognition. In particular to a technology for automatic input after automatic scanning and identification of paper regular bank notes of enterprises.
Background
In the daily operation process of an enterprise, the enterprise inventory funds are stored in bank regular financial products, a bank issues paper regular deposit slips, the enterprise needs to manually input account information according to the deposit slip information, and the input information comprises account numbers, account names, amounts, interest rates, time limits and the like.
Manually entering ledger information has the following disadvantages, including.
The working efficiency is low, and if a large amount of stored sheets are encountered, the stored sheets need to be manually input one by one, so that the time is long.
Secondly, under the condition of storing a large number of bills, manual entry of the ledger information is slow, mistakes are easy to make, and electronic storage cannot be realized.
Disclosure of Invention
The invention aims to provide an automatic entry method of paper deposit receipt based on image recognition, which realizes intelligent recognition and entry of paper deposit receipt of enterprises.
The technical scheme adopted by the invention for solving the technical problem is as follows.
Step 1, image acquisition, namely acquiring a paper deposit receipt image.
And 2, opening a file sharing storage service, and dividing all stored deposit receipt images into a training set and a test set.
And 3, constructing a deposit receipt image recognition model.
And 4, step 4: and training the deposit receipt image recognition model by using the deposit receipt image in the training set.
And 5: and testing the deposit receipt image recognition model by using the deposit receipt image in the test set. Carrying out manual marking and manual correction on the deposit receipt image with the problem identified; and then re-identifying and checking the deposit receipt image.
Further, the image acquisition in step 1 specifically realizes that: a high-definition scanner is deployed in an enterprise and supports maximum A4 paper scanning, the scanner is connected with a user computer service, and a user can check scanned files through the user computer after scanning.
Further, the deposit receipt image recognition model in step 3 includes an image processing module, an image correction module and an effective information extraction module, and is specifically implemented as follows.
An image processing module: firstly, extracting characteristic information of an input deposit receipt image, wherein the extracted characteristic information is mainly a color value; generating a oscillogram according to all color values extracted from the deposit receipt image; and then extracting a required waveform in the waveform image according to a specified threshold value, and generating a new image from the extracted waveform, wherein the generated new image is a black-and-white image.
An image correction module: performing noise reduction on the new image by adopting median filtering to remove interference noise points in the image; scanning the profile of the new image; acquiring a circumscribed rectangle with the minimum outline so as to obtain the coordinates of the central point, the length and the width of the circumscribed rectangle and the inclination angle relative to the horizontal line; and then, the circumscribed rectangle is horizontally corrected by combining the coordinates of the central point, the length, the width and the inclination angle with a trigonometric function.
The effective information extraction module: and performing effective area division on the post-corrected deposit receipt image, and concretely realizing the following steps: calculating the position and size of a rectangular area where each effective information in the input deposit receipt image is located by taking the minimum circumscribed rectangle in the corrected deposit receipt image as a reference object, and then cutting out the rectangular area where the effective information is located; and performing text matching through a tesseract engine to obtain character data corresponding to the rectangular area.
Further, the correction is specifically realized as follows.
If the height of the right angle of the center point of the minimum circumscribed rectangle is p, the bottom of the right angle is o, and the inclination angle with the horizontal line is a, the width difference w = of the coordinate axes between the center of the memory list and the center of the circumscribed rectanglep * cos a + o * sin aHeight difference h =p * sin a - o * cos aAnd after obtaining the coordinate of the central point of the deposit receipt, correcting the input deposit receipt image by taking the central point as a correction central point and the inclination angle a as a correction amplitude to obtain the deposit receipt image in the horizontal direction and the deposit receipt boundary area.
Further, the correction of step 5 is implemented as follows.
Correcting the coordinates of the text frame in the deposit receipt image according to the actual situation according to the out-of-range deviation or error of the corresponding data in the business system and the deposit receipt image; and then re-identifying and checking the deposit receipt image.
The invention has the following beneficial effects.
The invention provides an intelligent identification product for enterprises, and the intelligent identification product has the image identification capability meeting the special requirements of the enterprises by adopting a pyrrch algorithm library and through sample sampling and algorithm training, and can accurately identify the regular deposit receipt information of banks. The user can automatically scan and generate pictures through the scanner after taking the paper receipt, the system automatically investigates OCR program to identify receipt information and automatically fills the receipt information into the business system, the automatic input technology of the paper receipt is realized, and no manual work is needed in the whole process.
The method mainly solves the problem that the daily management of enterprises regularly stores the bills, realizes automatic picture identification and automatic information input into a business system by applying an artificial intelligence algorithm and simultaneously realizing the electronic storage of the paper bills in a model training mode. Thereby greatly improving the working efficiency of customers.
Detailed Description
The present invention will be further described with reference to the following examples.
Step 1: a high-definition scanner is deployed in an enterprise and supports maximum A4 paper scanning, the scanner is connected with a user computer service, and a user can check scanned files through the user computer after scanning.
Step 2: and opening a file sharing storage service, storing the files scanned by the user in a centralized storage space of the server, and dividing all deposit receipt images stored in the centralized storage space into a training set and a test set.
And step 3: and constructing a deposit receipt image recognition model, which comprises an image processing module, an image correction module and an effective information extraction module. Each module is specifically implemented as follows.
An image processing module: firstly, extracting characteristic information of an input deposit receipt image, wherein the extracted characteristic information is mainly a color value; generating a oscillogram according to all color values extracted from the deposit receipt image; and then extracting a required waveform in the waveform image according to a specified threshold value, and generating a new image from the extracted waveform, wherein the generated new image is a black-and-white image.
For example, if the deposit receipt image is an agricultural deposit receipt, the specific operation is as follows: performing feature extraction on the agricultural bank deposit receipt image, and extracting a unique graphic mark of the agricultural bank on the deposit receipt through a oscillogram; the unique marks on the deposit receipt images of different banks are different, and the unique graphic marks are determined by the oscillogram extracted by the appointed threshold value of each bank.
An image correction module: performing noise reduction on the new image by adopting median filtering to remove interference noise points in the image; scanning the new image for contours (i.e., edges of black and white boundary regions); acquiring a circumscribed rectangle with the minimum outline so as to obtain the coordinates of the central point, the length and the width of the circumscribed rectangle and the inclination angle relative to the horizontal line; and then, the circumscribed rectangle is horizontally corrected by combining the coordinates of the central point, the length, the width and the inclination angle with a trigonometric function.
The specific correction is realized as follows: if the height of the right angle of the center point of the minimum circumscribed rectangle is p, the bottom of the right angle is o, and the inclination angle with the horizontal line is a, the width difference w = of the coordinate axes between the center of the memory list and the center of the circumscribed rectanglep * cos a + o * sin aHeight difference h =p * sin a - o * cos aAnd after obtaining the coordinate of the central point of the deposit receipt, correcting the input deposit receipt image by taking the central point as a correction central point and the inclination angle a as a correction amplitude to obtain the deposit receipt image in the horizontal direction and the deposit receipt boundary area.
The effective information extraction module: the effective area division is carried out on the corrected deposit receipt image, and the specific implementation is as follows: calculating the position and size of a rectangular area where each effective information in the input deposit receipt image is located by taking the minimum circumscribed rectangle in the corrected deposit receipt image as a reference object, and then cutting out the rectangular area where the effective information is located; and performing text matching through a tesseract engine to obtain character data corresponding to the rectangular area.
The tesseract engine can improve the recognition success rate through normalized sample training for text matching.
And 4, step 4: and training the deposit receipt image recognition model by using the deposit receipt image in the training set.
And 5: testing the deposit receipt image recognition model by using the deposit receipt image in the test set; and checking the test result, identifying the problematic deposit receipt image, manually marking the deposit receipt image, and correcting the model.
The specific correction is as follows: and correcting the coordinates of the text frame in the deposit receipt image according to the actual situation when the corresponding data in the deposit receipt image and the service system have deviation or error. And then re-identifying and checking the deposit receipt image.
And 5: and when the centralized storage space of the server receives a scanned new deposit receipt image, the task triggering event firstly transmits the new deposit receipt image to a deposit receipt image recognition model, acquires a recognition result through the deposit receipt image recognition model and fills the recognition result in a corresponding position in a service system or a form.
And meanwhile, a task triggering program and an OCR recognition service network are opened, so that normal information communication is ensured.
Step 6: and deploying an automatic entry program, wherein the automatic entry program can access the service system, and simultaneously open and trigger a program network with a task, and the automatic entry program can automatically enter the regular deposit receipt information acquired by the deposit receipt image identification model, including account, account name, amount, time limit, interest and abstract, so that full-automatic entry of the information is realized.
The invention also provides an automatic paper deposit receipt input system based on image recognition, which comprises an acquisition module, a storage module, a deposit receipt image recognition model and a re-correction module. The system server is provided with a task triggering program, when a storage module of the server receives a new deposit receipt image obtained by scanning of the acquisition module, the task triggering program firstly transmits the new deposit receipt image to the deposit receipt image identification model, acquires an identification result through the deposit receipt image identification model, and fills the identification result in a corresponding position in a service system or a form.
The acquisition module is used for acquiring a scanning image of the paper deposit receipt; a high-definition scanner is deployed in an enterprise and supports maximum A4 paper scanning, the scanner is connected with a user computer service, and a user can check a scanned paper deposit receipt file through the user computer after scanning.
The storage module is used for storing the paper deposit document scanned by the user; and opening a file sharing storage service, storing the files scanned by the user in a centralized storage space of the server, and dividing all the deposit receipt images stored in the centralized storage space into a training set and a test set.
The deposit receipt image identification model comprises an image processing unit, an image correction unit and an effective information extraction unit, and each unit is realized as follows.
An image processing unit: firstly, extracting characteristic information of an input deposit receipt image by the existing method, wherein the extracted characteristic information is mainly a color value; generating a waveform image corresponding to the deposit receipt image according to all color values extracted from the deposit receipt image; and then extracting a required waveform in the waveform image according to a specified threshold value, and generating a new mark image by using the extracted waveform, wherein the generated new mark image is a black-and-white image.
For example, if the deposit receipt image is an agricultural deposit receipt, the specific operation is as follows: performing feature extraction on the agricultural bank deposit receipt image, and extracting a unique graphic mark of the agricultural bank on the deposit receipt through a oscillogram; the unique marks on the deposit receipt images of different banks are different, and the unique graphic marks are determined by the oscillogram extracted by the appointed threshold value of each bank.
An image correction unit: carrying out noise reduction on the marked image by adopting median filtering, and removing interference noise points in the marked image; scanning the outline of the image of the mark (i.e. the edge of the black-white boundary region); acquiring a circumscribed rectangle with the minimum outline so as to obtain the coordinates of the central point, the length and the width of the circumscribed rectangle and the inclination angle relative to the horizontal line; and then, the circumscribed rectangle is horizontally corrected by combining the coordinates of the central point, the length, the width and the inclination angle with a trigonometric function.
The specific correction is realized as follows: if the height of the right angle of the center point of the minimum circumscribed rectangle is p, the bottom of the right angle is o, and the inclination angle with the horizontal line is a, the width difference w = of the coordinate axes between the center of the memory list and the center of the circumscribed rectanglep * cos a + o * sin aHeight difference h =p * sin a - o * cos aAfter obtaining the coordinate of the central point of the deposit receipt, the central point is taken as the correction central point, the inclination angle a is taken as the correction amplitude, and the input deposit receipt image is corrected to obtain the deposit receipt image and the deposit receipt edge in the horizontal directionA border area.
An effective information extraction unit: the effective area division is carried out on the corrected deposit receipt image, and the specific implementation is as follows: calculating the position and size of a rectangular area where each effective information in the input deposit receipt image is located by taking the minimum circumscribed rectangle in the corrected deposit receipt image as a reference object, and then cutting out the rectangular area where the effective information is located; and performing text matching through a tesseract engine to obtain character data corresponding to the rectangular area.
The re-correcting module is used for correcting and identifying the deposit receipt image with the problem; if the deposit list image in the test set is used for testing the deposit list image recognition model; and checking the test result, identifying the deposit receipt image with obvious problems in the test result, manually marking, and then correcting the model again.
The specific correction is as follows: and correcting the coordinates of the text frame in the deposit receipt image according to the actual situation if the corresponding data in the deposit receipt image and the service system have large deviation or errors. And then re-identifying and checking the deposit receipt image.
The system deploys an automatic entry program, the automatic entry program can access the service system, and meanwhile, a task trigger program is opened, and the automatic entry program can automatically enter the regular deposit receipt information acquired by the deposit receipt image identification model, including account, account name, amount of money, time limit, interest and abstract, so that full-automatic entry of the information is realized.
Claims (7)
1. An automatic paper deposit receipt input method based on image recognition is characterized by comprising the following steps:
step 1, image acquisition, namely acquiring a paper deposit receipt image;
step 2, opening a file sharing storage service, and dividing all stored deposit receipt images into a training set and a test set;
step 3, constructing a deposit receipt image recognition model;
and 4, step 4: training the deposit receipt image recognition model by using the deposit receipt image in the training set;
and 5: testing the deposit receipt image recognition model by using the deposit receipt image in the test set; carrying out manual marking and manual correction on the deposit receipt image with the problem identified; and then re-identifying and checking the deposit receipt image.
2. The automatic paper deposit receipt entering method based on image recognition as claimed in claim 1, wherein the image acquisition in step 1 is realized by: a high-definition scanner is deployed in an enterprise and supports maximum A4 paper scanning, the scanner is connected with a user computer service, and a user can check scanned files through the user computer after scanning.
3. The automatic entry method of paper deposit receipt based on image recognition as claimed in claim 1, wherein the deposit receipt image recognition model in step 3 comprises an image processing module, an image correction module and an effective information extraction module, and is implemented as follows:
an image processing module: firstly, extracting characteristic information of an input deposit receipt image, wherein the extracted characteristic information is mainly a color value; generating a oscillogram according to all color values extracted from the deposit receipt image; then extracting a required waveform in the waveform diagram according to a specified threshold value, and generating a new image from the extracted waveform, wherein the generated new image is a black-and-white image;
an image correction module: performing noise reduction on the new image by adopting median filtering to remove interference noise points in the image; scanning the profile of the new image; acquiring a circumscribed rectangle with the minimum outline so as to obtain the coordinates of the central point, the length and the width of the circumscribed rectangle and the inclination angle relative to the horizontal line; then, horizontally correcting the circumscribed rectangle by combining the coordinates of the central point, the length, the width and the inclination angle with a trigonometric function;
the effective information extraction module: the effective area division is carried out on the corrected deposit receipt image, and the specific implementation is as follows: calculating the position and size of a rectangular area where each effective information in the input deposit receipt image is located by taking the minimum circumscribed rectangle in the corrected deposit receipt image as a reference object, and then cutting out the rectangular area where the effective information is located; and performing text matching through a tesseract engine to obtain character data corresponding to the rectangular area.
4. The automatic paper deposit receipt entering method based on image recognition as claimed in claim 3, characterized in that the correction is implemented as follows:
if the height of the right angle of the center point of the minimum circumscribed rectangle is p, the bottom of the right angle is o, and the inclination angle with the horizontal line is a, the width difference w = of the coordinate axes between the center of the memory list and the center of the circumscribed rectanglep * cos a + o * sin aHeight difference h =p * sin a - o * cos aAnd after obtaining the coordinate of the central point of the deposit receipt, correcting the input deposit receipt image by taking the central point as a correction central point and the inclination angle a as a correction amplitude to obtain the deposit receipt image in the horizontal direction and the deposit receipt boundary area.
5. The automatic entry method of paper deposit receipt based on image recognition according to claim 3 or 4, characterized in that the correction of step 5 is implemented as follows:
correcting the coordinates of the text frame in the deposit receipt image according to the actual situation according to the out-of-range deviation or error of the corresponding data in the business system and the deposit receipt image; and then re-identifying and checking the deposit receipt image.
6. The automatic paper deposit receipt entering method based on the image recognition is characterized in that a system for realizing the method comprises an acquisition module, a storage module, a deposit receipt image recognition model and a re-correction module; deploying a task triggering program in a system server, when a storage module of the server receives a new deposit receipt image scanned by an acquisition module, transmitting the new deposit receipt image to a deposit receipt image recognition model by the task triggering program, acquiring a recognition result through the deposit receipt image recognition model, and filling the recognition result in a corresponding position in a service system or a form; and if errors exist, the result is corrected through the re-correcting module.
7. The automatic paper deposit receipt entering method based on image recognition as claimed in claim 6, wherein the method is characterized in that
The acquisition module is used for acquiring a scanning image of the paper deposit receipt; a high-definition scanner is deployed in an enterprise and supports maximum A4 paper scanning, the scanner is connected with a user computer service, and a user can check a scanned paper deposit receipt file through the user computer after scanning;
the storage module is used for storing the paper deposit document scanned by the user; opening a file sharing storage service, storing the files scanned by a user in a centralized storage space of a server, and dividing all deposit receipt images stored in the centralized storage space into a training set and a test set;
the deposit receipt image recognition model comprises an image processing unit, an image correction unit and an effective information extraction unit, and each unit is realized as follows:
an image processing unit: firstly, extracting characteristic information of an input deposit receipt image by the existing method, wherein the extracted characteristic information is mainly a color value; generating a waveform image corresponding to the deposit receipt image according to all color values extracted from the deposit receipt image; then extracting a required waveform from the waveform image according to a specified threshold value, and generating a new marking image from the extracted waveform, wherein the generated new marking image is a black-and-white image;
an image correction unit: carrying out noise reduction on the marked image by adopting median filtering, and removing interference noise points in the marked image; scanning the outline of the marker image; acquiring a circumscribed rectangle with the minimum outline so as to obtain the coordinates of the central point, the length and the width of the circumscribed rectangle and the inclination angle relative to the horizontal line; then, horizontally correcting the circumscribed rectangle by combining the coordinates of the central point, the length, the width and the inclination angle with a trigonometric function;
the specific correction is realized as follows: if the height of the right angle of the center point of the minimum circumscribed rectangle is p, the bottom of the right angle is o, and the inclination angle with the horizontal line is a, the width difference w = of the coordinate axes between the center of the memory list and the center of the circumscribed rectanglep * cos a + o * sin aHeight difference h =p * sin a - o * cos aAfter obtaining the coordinate of the central point of the deposit receipt, correcting the input deposit receipt image by taking the central point as a correction central point and the inclination angle a as a correction amplitude to obtain a deposit receipt image in the horizontal direction and a deposit receipt boundary area;
an effective information extraction unit: the effective area division is carried out on the corrected deposit receipt image, and the specific implementation is as follows: calculating the position and size of a rectangular area where each effective information in the input deposit receipt image is located by taking the minimum circumscribed rectangle in the corrected deposit receipt image as a reference object, and then cutting out the rectangular area where the effective information is located; performing text matching through a tesseract engine to obtain character data corresponding to the rectangular area;
the re-correcting module is used for correcting the deposit receipt image with the problem of the identification result; testing the deposit receipt image recognition model by using the deposit receipt image in the test set; checking the test result, identifying the deposit receipt image with obvious problems in the test result, carrying out artificial marking, and then carrying out artificial correction; the specific correction is as follows: correcting the coordinates of the text frame in the deposit receipt image according to the actual situation if the corresponding data in the deposit receipt image and the service system have large deviation or errors; and then re-identifying and checking the deposit receipt image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111390231.4A CN113822269B (en) | 2021-11-23 | 2021-11-23 | Paper deposit receipt automatic input method based on image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111390231.4A CN113822269B (en) | 2021-11-23 | 2021-11-23 | Paper deposit receipt automatic input method based on image recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113822269A true CN113822269A (en) | 2021-12-21 |
| CN113822269B CN113822269B (en) | 2022-03-29 |
Family
ID=78919807
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111390231.4A Active CN113822269B (en) | 2021-11-23 | 2021-11-23 | Paper deposit receipt automatic input method based on image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113822269B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116302260A (en) * | 2023-02-27 | 2023-06-23 | 浙江同花顺智能科技有限公司 | Method and system for guiding user to conduct stock account opening online by digital virtual person |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180018774A1 (en) * | 2016-07-15 | 2018-01-18 | Abbyy Development Llc | Method and system for preparing text images for optical-character recognition |
| CN112966537A (en) * | 2021-02-10 | 2021-06-15 | 北京邮电大学 | Form identification method and system based on two-dimensional code positioning |
-
2021
- 2021-11-23 CN CN202111390231.4A patent/CN113822269B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180018774A1 (en) * | 2016-07-15 | 2018-01-18 | Abbyy Development Llc | Method and system for preparing text images for optical-character recognition |
| CN112966537A (en) * | 2021-02-10 | 2021-06-15 | 北京邮电大学 | Form identification method and system based on two-dimensional code positioning |
Non-Patent Citations (1)
| Title |
|---|
| 贾子轩: "一种基于图像处理的化验单识别算法研究", 《中国优秀博硕士学位论文全文数据库（硕士）医药卫生科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116302260A (en) * | 2023-02-27 | 2023-06-23 | 浙江同花顺智能科技有限公司 | Method and system for guiding user to conduct stock account opening online by digital virtual person |
| CN116302260B (en) * | 2023-02-27 | 2024-02-13 | 浙江同花顺智能科技有限公司 | Method and system for guiding user to conduct stock account opening online by digital virtual person |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113822269B (en) | 2022-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10019740B2 (en) | System and methods of an expense management system based upon business document analysis | |
| US5140650A (en) | Computer-implemented method for automatic extraction of data from printed forms | |
| US7590275B2 (en) | Method and system for recognizing a candidate character in a captured image | |
| US6950553B1 (en) | Method and system for searching form features for form identification | |
| CA2589947C (en) | Machine character recognition verification | |
| US8155425B1 (en) | Automated check detection and image cropping | |
| CN109657665A (en) | A kind of invoice batch automatic recognition system based on deep learning | |
| CN111476109A (en) | Bill processing method, bill processing apparatus, and computer-readable storage medium | |
| CN110597806A (en) | Wrong question set generation and answer statistics system and method based on reading and amending identification | |
| CN107016363A (en) | Bill images managing device, bill images management system and method | |
| CN112669515B (en) | Bill image recognition method and device, electronic equipment and storage medium | |
| CN113822269B (en) | Paper deposit receipt automatic input method based on image recognition | |
| US20230368391A1 (en) | Image Evaluation and Dynamic Cropping System | |
| JP2002092682A (en) | Method and system of detecting watermark | |
| CN109741273A (en) | A kind of mobile phone photograph low-quality images automatically process and methods of marking | |
| CN116092231A (en) | Ticket identification method, ticket identification device, terminal equipment and storage medium | |
| CN115116068A (en) | Archive intelligent filing system based on OCR | |
| US8249364B2 (en) | Method for resolving contradicting output data from an optical character recognition (OCR) system, wherein the output data comprises more than one recognition alternative for an image of a character | |
| CN111462388A (en) | Bill inspection method and device, terminal equipment and storage medium | |
| CN112508000A (en) | Method and equipment for generating OCR image recognition model training data | |
| CN110619060A (en) | Cigarette carton image database construction method and cigarette carton anti-counterfeiting query method | |
| CN114912417A (en) | Service data processing method, device, equipment and storage medium | |
| CN114708600A (en) | Document auditing method and device, computer equipment and storage medium | |
| CN111046860A (en) | Intelligent claim settlement system based on mobile carrier | |
| CN113762244A (en) | Document information extraction method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |