CN111368828A - Multi-bill identification method and device - Google Patents
Multi-bill identification method and device Download PDFInfo
- Publication number
- CN111368828A CN111368828A CN202010124850.8A CN202010124850A CN111368828A CN 111368828 A CN111368828 A CN 111368828A CN 202010124850 A CN202010124850 A CN 202010124850A CN 111368828 A CN111368828 A CN 111368828A
- Authority
- CN
- China
- Prior art keywords
- bill
- image
- identified
- model
- images
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Character Input (AREA)
- Character Discrimination (AREA)
Abstract
The invention discloses a method for identifying multiple bills, which comprises the following steps: preprocessing a plurality of bill images to be identified; simultaneously distinguishing each bill in a plurality of bill images to be recognized by using a YOLOv3 model for the preprocessed bill images, and positioning characters in each bill image; characters in each positioned bill image are identified by using a CRNN + CTC model, and information of a plurality of bills to be identified is acquired, so that the problems that in the prior art, an identification system based on optical characters is relatively low in identification rate and high in later maintenance cost, and only a single bill can be identified at one time are solved.
Description
Technical Field
The application relates to the field of artificial intelligence, in particular to a multi-bill identification method and a multi-bill identification device.
Background
Existing document recognition techniques, including ordinary OCR optical character positioning, also have in part used mainstream deep learning techniques. However, only single bill information can be identified, the common multi-bill reimbursement scene cannot be met, and the problem of bill identification cannot be fundamentally solved. In addition, the recognition system based on the optical characters has relatively low recognition rate and higher later maintenance cost, and can not meet the existing reimbursement market.
Disclosure of Invention
The application provides a method and a device for identifying multiple bills, which solve the problems that in the prior art, an identification system based on optical characters is relatively low in identification rate and high in later maintenance cost, and only a single bill can be identified at a time.
The application provides a multi-bill identification method, which comprises the following steps:
preprocessing a plurality of bill images to be identified;
simultaneously distinguishing each bill in a plurality of bill images to be recognized by using a YOLOv3 model for the preprocessed bill images, and positioning characters in each bill image;
and identifying the characters in each positioned bill image by using a CRNN + CTC model to obtain a plurality of pieces of bill information to be identified.
Preferably, the preprocessing is performed on the image containing a plurality of bills to be identified, and comprises the following steps:
carrying out normalization processing on a bill image containing a plurality of bills to be identified;
a standardized ticket image is obtained.
Preferably, the standardized ticket image comprises:
bill images with equal size and same gray level.
Preferably, the method for distinguishing each bill in the images containing the multiple bills to be recognized simultaneously by using the YOLOv3 model comprises the following steps:
distinguishing the outline of each bill in the image containing the bills to be recognized by using a YOLOv3 model, and further acquiring the image of each bill; and
and acquiring the number of the contained bills to be identified.
Preferably, after the step of positioning the text in each bill image, the method further comprises:
cutting the character part in each bill image to obtain an image containing the character part in each bill image;
the image including the text portion is passed into a CRNN + CTC model.
Preferably, the method for recognizing the characters in each located bill image by using the CRNN + CTC model to obtain a plurality of pieces of bill information to be recognized includes:
the CRNN + CTC model receives the image containing the text part sent by the YOLOv3 model;
and the CRNN + CTC model acquires the information of each invoice in a plurality of bill images to be identified by identifying the image containing the character part.
Preferably, the information of the invoice comprises:
the type of invoice, invoice code, invoice number, invoice date, buyer and seller information, invoice amount, and invoice detail.
This application provides the recognition device of many bills simultaneously, includes:
the device comprises a preprocessing unit, a recognition unit and a recognition unit, wherein the preprocessing unit is used for preprocessing a plurality of bill images to be recognized;
the bill distinguishing and positioning unit is used for distinguishing each bill in the bill images to be recognized simultaneously by using a YOLOv3 model and positioning characters in each bill image after the preprocessing;
and the bill identification unit is used for identifying the characters in each positioned bill image by using a CRNN + CTC model and acquiring a plurality of pieces of bill information to be identified.
Preferably, the bill distinguishing and positioning unit comprises:
the single bill image acquiring subunit is used for distinguishing the outline of each bill in the multiple bill images to be identified by using a YOLOv3 model so as to acquire the image of each bill; and
and the bill quantity acquiring subunit is used for acquiring the quantity of the contained bills to be identified.
Preferably, the bill identifying unit includes:
the image receiving subunit is used for receiving the image containing the text part sent by the YOLOv3 model through the CRNN + CTC model;
and the bill information identification subunit is used for acquiring the information of each invoice in the plurality of bill images to be identified by identifying the image containing the character part through the CRNN + CTC model.
The application provides a method and a device for identifying multiple bills, which are characterized in that a YOLOv3 model is used for simultaneously distinguishing single bills and positioning characters, then a CRNN + CTC model is used for identifying information of multiple materials, and the information of multiple bills is acquired simultaneously, so that the problems that in the prior art, an identification system based on optical characters is relatively low in identification rate, high in later maintenance cost and only capable of identifying a single bill at a time are solved.
Drawings
FIG. 1 is a schematic flow chart of a multi-bill identification method provided by the present application;
fig. 2 is a schematic diagram of a multi-bill identification device provided by the application.
Detailed Description
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit of this application and is therefore not limited to the specific implementations disclosed below.
Fig. 1 is a schematic flow chart of a method for identifying multiple tickets, which is described in detail below with reference to fig. 1.
Step S101, preprocessing the bill images containing a plurality of bills to be identified.
The method provided by the application can simultaneously identify a plurality of bills, firstly, the plurality of bills are tiled, images of the plurality of bills are shot through image acquisition equipment such as a camera, each shot image comprises a plurality of bills to be identified, and the plurality of bills can contain various types of bills such as common invoices, special invoices for value-added tax, special invoices and the like.
When the image of the bill is shot, due to the reasons of angle, explosion and the like, each image interferes with the used recognition model during recognition to influence the execution efficiency, so that the bill image needs to be preprocessed before bill recognition, and the preprocessing method is to perform normalization processing on the bill image containing a plurality of to-be-recognized bills to obtain a standardized bill image. The standardized bill images can be bill images with the same size and the same gray level, wherein the preferred value of the gray level value is 0-1. The standardized bill image can reduce a large amount of calculation for processing the original image in the identification process, and the execution efficiency of the corresponding identification model is higher.
Step S102, using a YOLOv3 model to simultaneously distinguish each bill in the bill images to be recognized and locate characters in each bill image after the preprocessing.
The principle of the YOLOv3 model is to detect a target image by using feature maps of 3 different scales to obtain a feature image of a corresponding size. In the present application, the image of the bill to be recognized includes information of a plurality of bills to be recognized, and the YOLOv3 model can distinguish the outline of each bill in the image of the bill to be recognized according to the preset characteristic diagram of each bill, so as to obtain the image of each bill, and the advising also obtains the number of the bills to be recognized.
Then, the text in each bill image is located, the text part in each bill image is cut, and an image including the text part in each bill image is obtained, for example, the text part of the bill type is cut, so that an image including the text part of the bill type, such as a "common invoice for value-added tax of a certain market" or a "special invoice for value-added tax of a certain market" can be obtained. The image containing the text portion is then passed into the CRNN + CTC model.
And step S103, identifying the characters in each positioned bill image by using a CRNN + CTC model, and acquiring a plurality of pieces of bill information to be identified.
The CRNN + CTC model is used to identify text in an image, and typically includes: character detection and character recognition.
The CRNN + CTC model receives the image containing the text part sent by the YOLOv3 model, and acquires information of each invoice in a plurality of to-be-identified bill images by identifying the image containing the text part. Specifically, the CRNN + CTC model further detects the image, and then locates the range of the text. For example, the bill code is positioned, the digit of the bill code is obtained, then the positioned text area is identified, the text area is lent as character information, and therefore information of each invoice in a plurality of bill images to be identified is obtained. Information of invoices, including: the type of invoice, invoice code, invoice number, invoicing date, buyer and seller information, invoicing amount, invoice detail and the like.
Based on the same inventive concept, the present application also provides a bill recognition apparatus 200, as shown in fig. 2, including:
the preprocessing unit 210 is used for preprocessing the bill images to be identified;
the bill distinguishing and positioning unit 220 is used for distinguishing each bill in the bill images to be recognized simultaneously by using a YOLOv3 model and positioning characters in each bill image after the preprocessing;
and the bill identification unit 230 is configured to identify the characters in each located bill image by using a CRNN + CTC model, and acquire information of a plurality of bills to be identified.
Preferably, the bill distinguishing and positioning unit comprises:
the single bill image acquiring subunit is used for distinguishing the outline of each bill in the multiple bill images to be identified by using a YOLOv3 model so as to acquire the image of each bill; and
and the bill quantity acquiring subunit is used for acquiring the quantity of the contained bills to be identified.
Preferably, the bill identifying unit includes:
the image receiving subunit is used for receiving the image containing the text part sent by the YOLOv3 model through the CRNN + CTC model;
and the bill information identification subunit is used for acquiring the information of each invoice in the plurality of bill images to be identified by identifying the image containing the character part through the CRNN + CTC model.
In summary, the present application provides a method and an apparatus for identifying multiple bills, which perform single-sheet distinguishing and text positioning on multiple bills simultaneously through the YOLOv3 model, then identify information of multiple materials through the CRNN + CTC model, and obtain information of multiple bills simultaneously, thereby solving the problems that in the prior art, an identification system based on optical characters is relatively low in identification rate, high in later maintenance cost, and only a single bill can be identified at a time.
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the spirit and scope of the invention.
Claims (10)
1. A method for identifying multiple bills, comprising:
preprocessing a plurality of bill images to be identified;
simultaneously distinguishing each bill in a plurality of bill images to be recognized by using a YOLOv3 model for the preprocessed bill images, and positioning characters in each bill image;
and identifying the characters in each positioned bill image by using a CRNN + CTC model to obtain a plurality of pieces of bill information to be identified.
2. The method of claim 1, wherein preprocessing the image containing the plurality of documents to be recognized comprises:
carrying out normalization processing on a bill image containing a plurality of bills to be identified;
a standardized ticket image is obtained.
3. The method of claim 2, wherein the normalized document image comprises:
bill images with equal size and same gray level.
4. The method of claim 1, wherein distinguishing each document containing a plurality of document images to be recognized simultaneously using the YOLOv3 model comprises:
distinguishing the outline of each bill in the image containing the bills to be recognized by using a YOLOv3 model, and further acquiring the image of each bill; and
and acquiring the number of the contained bills to be identified.
5. The method of claim 1, further comprising, after the step of locating text in each document image:
cutting the character part in each bill image to obtain an image containing the character part in each bill image;
the image including the text portion is passed into a CRNN + CTC model.
6. The method of claim 1, wherein identifying the located text in each document image using a CRNN + CTC model to obtain a plurality of document information to be identified comprises:
the CRNN + CTC model receives the image containing the text part sent by the YOLOv3 model;
and the CRNN + CTC model acquires the information of each invoice in a plurality of bill images to be identified by identifying the image containing the character part.
7. The method of claim 6, wherein the information of the invoice comprises:
the type of invoice, invoice code, invoice number, invoice date, buyer and seller information, invoice amount, and invoice detail.
8. A multiple note identification device comprising:
the device comprises a preprocessing unit, a recognition unit and a recognition unit, wherein the preprocessing unit is used for preprocessing a plurality of bill images to be recognized;
the bill distinguishing and positioning unit is used for distinguishing each bill in the bill images to be recognized simultaneously by using a YOLOv3 model and positioning characters in each bill image after the preprocessing;
and the bill identification unit is used for identifying the characters in each positioned bill image by using a CRNN + CTC model and acquiring a plurality of pieces of bill information to be identified.
9. The apparatus of claim 8, wherein the ticket differentiating and locating unit comprises:
the single bill image acquiring subunit is used for distinguishing the outline of each bill in the multiple bill images to be identified by using a YOLOv3 model so as to acquire the image of each bill; and
and the bill quantity acquiring subunit is used for acquiring the quantity of the contained bills to be identified.
10. The apparatus of claim 8, wherein the bill identifying unit comprises:
the image receiving subunit is used for receiving the image containing the text part sent by the YOLOv3 model through the CRNN + CTC model;
and the bill information identification subunit is used for acquiring the information of each invoice in the plurality of bill images to be identified by identifying the image containing the character part through the CRNN + CTC model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010124850.8A CN111368828A (en) | 2020-02-27 | 2020-02-27 | Multi-bill identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010124850.8A CN111368828A (en) | 2020-02-27 | 2020-02-27 | Multi-bill identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111368828A true CN111368828A (en) | 2020-07-03 |
Family
ID=71208278
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010124850.8A Pending CN111368828A (en) | 2020-02-27 | 2020-02-27 | Multi-bill identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111368828A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112364837A (en) * | 2020-12-09 | 2021-02-12 | 四川长虹电器股份有限公司 | Bill information identification method based on target detection and text identification |
| CN112668559A (en) * | 2021-03-15 | 2021-04-16 | 冠传网络科技(南京)有限公司 | Multi-mode information fusion short video emotion judgment device and method |
| CN112749731A (en) * | 2020-12-10 | 2021-05-04 | 航天信息股份有限公司 | Bill quantity identification method and system based on deep neural network |
| CN113688834A (en) * | 2021-07-27 | 2021-11-23 | 深圳中兴网信科技有限公司 | Ticket recognition method, ticket recognition system and computer readable storage medium |
| WO2022147965A1 (en) * | 2021-01-09 | 2022-07-14 | 江苏拓邮信息智能技术研究院有限公司 | Arithmetic question marking system based on mixnet-yolov3 and convolutional recurrent neural network (crnn) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657665A (en) * | 2018-10-31 | 2019-04-19 | 广东工业大学 | A kind of invoice batch automatic recognition system based on deep learning |
| CN109766881A (en) * | 2018-11-28 | 2019-05-17 | 北京捷通华声科技股份有限公司 | A kind of character identifying method and device of vertical text image |
| CN109858414A (en) * | 2019-01-21 | 2019-06-07 | 南京邮电大学 | A kind of invoice piecemeal detection method |
| CN109977957A (en) * | 2019-03-04 | 2019-07-05 | 苏宁易购集团股份有限公司 | A kind of invoice recognition methods and system based on deep learning |
| CN110032990A (en) * | 2019-04-23 | 2019-07-19 | 杭州智趣智能信息技术有限公司 | A kind of invoice text recognition method, system and associated component |
| CN110348441A (en) * | 2019-07-10 | 2019-10-18 | 深圳市华云中盛科技有限公司 | VAT invoice recognition methods, device, computer equipment and storage medium |
| CN110348346A (en) * | 2019-06-28 | 2019-10-18 | 苏宁云计算有限公司 | A kind of bill classification recognition methods and system |
-
2020
- 2020-02-27 CN CN202010124850.8A patent/CN111368828A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657665A (en) * | 2018-10-31 | 2019-04-19 | 广东工业大学 | A kind of invoice batch automatic recognition system based on deep learning |
| CN109766881A (en) * | 2018-11-28 | 2019-05-17 | 北京捷通华声科技股份有限公司 | A kind of character identifying method and device of vertical text image |
| CN109858414A (en) * | 2019-01-21 | 2019-06-07 | 南京邮电大学 | A kind of invoice piecemeal detection method |
| CN109977957A (en) * | 2019-03-04 | 2019-07-05 | 苏宁易购集团股份有限公司 | A kind of invoice recognition methods and system based on deep learning |
| CN110032990A (en) * | 2019-04-23 | 2019-07-19 | 杭州智趣智能信息技术有限公司 | A kind of invoice text recognition method, system and associated component |
| CN110348346A (en) * | 2019-06-28 | 2019-10-18 | 苏宁云计算有限公司 | A kind of bill classification recognition methods and system |
| CN110348441A (en) * | 2019-07-10 | 2019-10-18 | 深圳市华云中盛科技有限公司 | VAT invoice recognition methods, device, computer equipment and storage medium |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112364837A (en) * | 2020-12-09 | 2021-02-12 | 四川长虹电器股份有限公司 | Bill information identification method based on target detection and text identification |
| CN112749731A (en) * | 2020-12-10 | 2021-05-04 | 航天信息股份有限公司 | Bill quantity identification method and system based on deep neural network |
| WO2022147965A1 (en) * | 2021-01-09 | 2022-07-14 | 江苏拓邮信息智能技术研究院有限公司 | Arithmetic question marking system based on mixnet-yolov3 and convolutional recurrent neural network (crnn) |
| CN112668559A (en) * | 2021-03-15 | 2021-04-16 | 冠传网络科技(南京)有限公司 | Multi-mode information fusion short video emotion judgment device and method |
| CN113688834A (en) * | 2021-07-27 | 2021-11-23 | 深圳中兴网信科技有限公司 | Ticket recognition method, ticket recognition system and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111368828A (en) | Multi-bill identification method and device | |
| CN108717545B (en) | Bill identification method and system based on mobile phone photographing | |
| CN110569878B (en) | Photograph background similarity clustering method based on convolutional neural network and computer | |
| CN104217203B (en) | Complex background card face information identifying method and system | |
| CN105528604A (en) | Bill automatic identification and processing system based on OCR | |
| US20140268250A1 (en) | Systems and methods for receipt-based mobile image capture | |
| CN111639629B (en) | Pig weight measurement method and device based on image processing and storage medium | |
| CN106874901B (en) | Driving license identification method and device | |
| CN113158895B (en) | Bill identification method and device, electronic equipment and storage medium | |
| CN104915668A (en) | Character information identification method for medical image and device thereof | |
| CN112132016B (en) | Bill information extraction method and device and electronic equipment | |
| CN109271951A (en) | A kind of method and system promoting book keeping operation review efficiency | |
| CN110569856A (en) | sample labeling method and device, and damage category identification method and device | |
| CN103729656A (en) | Seal imprint identification method and identification system | |
| CN110291527B (en) | Information processing method, system, cloud processing device and computer program product | |
| CN110688900A (en) | Withdrawal meter management method based on image recognition | |
| CN114511866A (en) | Data auditing method, device, system, processor and machine-readable storage medium | |
| CN113469005A (en) | Recognition method of bank receipt, related device and storage medium | |
| CN108921240A (en) | A kind of invoice code identification sorting system | |
| CN111462388A (en) | Bill inspection method and device, terminal equipment and storage medium | |
| CN112613367A (en) | Bill information text box acquisition method, system, equipment and storage medium | |
| CN116798061A (en) | Bill auditing and identifying method, device, terminal and storage medium | |
| US7804979B2 (en) | Feeder control system and method | |
| TWM575887U (en) | Intelligent accounting system | |
| CN115471858A (en) | Data processing method and device for bill management |
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 |