CN108470164A - A kind of digital recognition system and method for financial statement - Google Patents
A kind of digital recognition system and method for financial statement Download PDFInfo
- Publication number
- CN108470164A CN108470164A CN201810228641.0A CN201810228641A CN108470164A CN 108470164 A CN108470164 A CN 108470164A CN 201810228641 A CN201810228641 A CN 201810228641A CN 108470164 A CN108470164 A CN 108470164A
- Authority
- CN
- China
- Prior art keywords
- image
- financial statement
- unit
- line
- 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
Classifications
-
- 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
- G06V30/40—Document-oriented image-based pattern recognition
- G06V30/41—Analysis of document content
- G06V30/413—Classification of content, e.g. text, photographs or tables
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/242—Aligning, centring, orientation detection or correction of the image by image rotation, e.g. by 90 degrees
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of digital recognition systems and method for financial statement, including table detection module and digital identification module;The table detection module includes image correction unit and table detection unit;Described image corrects unit, for the financial statement image to be corrected, keeps its table change horizontal;The table detection unit, for detecting the table line in the financial statement image after correcting, extraction unit table images;The number identification module, for identification content in the unit table images.Present invention is mainly applied to the number identifications in financial statement, realize the whole-process automatic identification of review process.Both manpower has been saved, has in turn ensured the just, openly of verifying work.
Description
Technical field
The present invention relates to the artificial intelligence judgment technology fields of financial statement, more particularly to a kind of number for financial statement
Word recognition system and method.
Background technology
Constantly improve with living standards of the people with the continuous social and economic development, bank is for identification financial statement
Demand is increasing.Mainly by manual identified, this method cost of labor is higher for number identification in traditional financial statement,
It is less efficient, and repeated verification operation easy tos produce fatigue for a long time, the defective modes such as carelessness influence to verify accuracy rate.
How accurately and rapidly the number in financial statement to be identified, while avoiding manual identified of high cost, it is easily tired
Labor, the easily drawbacks such as carelessness are the technical issues of being badly in need of solving.
Invention content
The technical problem to be solved by the present invention is to:It is proposed a kind of digital recognition system and method for financial statement,
Number that can be in automatic identification financial statement, to meet the needs of nowadays identifying efficiency, accuracy rate in work to financial statement.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of digital recognition system for financial statement, including table detection module and digital identification module;The table
Lattice detection module includes image correction unit and table detection unit;
Described image corrects unit, for the financial statement image to be corrected, keeps its table change horizontal;
The table detection unit, for detecting the table line in the financial statement image after correcting, extraction unit trrellis diagram
Picture;
The number identification module, for identification content in the unit table images.
A kind of digit recognition method for financial statement includes the following steps:
S1, financial statement image is downloaded from server;
S2, image rectification is carried out to the financial statement image using based on Hough straight-line method, keeps table change horizontal;
S3, table line, and extraction unit table images are detected using morphological image;
S4, the number in the unit table images is identified using the digital identification model based on deep learning network successively,
The content of whole table of final output.
Further, the method for described image correction is as follows:
S21,50 straight lines random in the financial statement image are obtained using Hough straight-line method, is taken if less than 50
All straight lines;
S22, the slope according to 50 straight lines of acquisition, calculate G-bar;
S23, image is corrected according to obtained G-bar, keeps table change horizontal.
Further, the method for the extraction unit table images is as follows:
S31, binary conversion treatment is carried out to the financial statement image;
S32, the horizontal, vertical two kinds of strips of construction corrode core, are filtered to bianry image, lateral encroaching core handles to obtain table
Horizontal line, the vertical core that corrodes handles to obtain the vertical line of table;
The horizontal line and vertical line that S33, basis obtain, determine the position of each cell in table area.
Further, the obtaining step of the digital identification model is as follows:
Unit table images under S41, acquisition different location, different noises;
S42, the position that each number in the unit table images is marked using rectangle frame, and record respective classes mark
Label;
S43, number identification deep neural network model is trained using above-mentioned digital data sets, obtain digital identification model.
The beneficial effects of the invention are as follows:Present invention is mainly applied to the number identifications in financial statement, realize audit
The whole-process automatic identification of process.Both manpower has been saved, has in turn ensured the just, openly of verifying work.
Description of the drawings
Fig. 1 is the structural schematic diagram of the digital recognition system of the present invention.
Fig. 2 is the flow chart of the digit recognition method of the present invention.
Specific implementation mode
Below in conjunction with attached drawing.The present invention will be further described.
The digital recognition system structure of the present invention is as shown in Figure 1, including table detection module and digital identification module.Its
In, table detection module includes:Image correction unit and table detection unit.
The operation principle of this system is as follows:
First, image correction unit is used on financial statement image, obtains the financial statement image of correction.Then high-ranking officers
Image after just is passed to table and detects detection unit, obtains table area image, and determines the position of each cell in image.
Finally, by the incoming digital identification module of the image in cell, digital identification module uses the number based on deep learning network
Identification model identifies the number in each cell.
Table detection module is first corrected image, then, table line, this method is detected in the image of correction
It can be effectively prevented from because the flase drop that table line distorts or tilting band is come influences, the accuracy rate of table line detecting be improved, into one
Step improves the accuracy of number identification.
The specific detection method of table detection unit is:Image is carried out binaryzation by table detection unit first, then structure
It makes horizontal, vertical two kinds of strips corrosion verification bianry image to be filtered, lateral encroaching core handles to obtain the horizontal line of table, vertical to corrode
Core handles to obtain the vertical line of table, the horizontal line that wherein red line indicates, the vertical line that blue line indicates.Then the cross that will be obtained
Line and vertical line are combined, and obtain the position of all cells.
Digital identification model acquisition methods are as follows:
S1, training data prepare:Obtain different background, the table cell digital picture of content and noise;
S2, data mark:The different numbers in cell are marked and assign to corresponding label using rectangle frame, wherein
Label 1~10 indicates that number 0~9, label 0 indicate background respectively;
S3, model training:Using the training data marked, digital identification model of the training based on deep learning network
(common knowledge does not repeat hereby);
The specific detection method of digital identification module includes:Obtained list cell table images are inputted into number identification mould
Type obtains the position belonging to one group of rectangle frame and corresponding label;Rectangle frame is successively read its label by sequence from left to right,
By tag combination at final recognition result.The content of all cells is all identified in aforementioned manners, integration obtains
The digital recognition result of whole table.
The implementation detailed process of the present invention is including as follows as shown in Fig. 2, a kind of digit recognition method for financial statement
Step:
S1, financial statement image is downloaded from server;
S2, image rectification is carried out to the financial statement image using based on Hough straight-line method, keeps table change horizontal;
The method of described image correction is as follows:
S21,50 straight lines random in the financial statement image are obtained using Hough straight-line method, is taken if less than 50
All straight lines;
S22, the slope according to 50 straight lines of acquisition, calculate G-bar;
S23, image is corrected according to obtained G-bar, keeps table change horizontal;
S3, table line, and extraction unit table images are detected using morphological image;
The method of the extraction unit table images is as follows:
S31, binary conversion treatment is carried out to the financial statement image;
S32, the horizontal, vertical two kinds of strips of construction corrode core, are filtered to bianry image, lateral encroaching core handles to obtain table
Horizontal line, the vertical core that corrodes handles to obtain the vertical line of table;
The horizontal line and vertical line that S33, basis obtain, determine the position of each cell in table area;
S4, the number in the unit table images is identified using the digital identification model based on deep learning network successively,
The content of whole table of final output;
The obtaining step of the number identification model is as follows:
Unit table images under S41, acquisition different location, different noises;
S42, the position that each number in the unit table images is marked using rectangle frame, and record respective classes mark
Label;
S43, number identification deep neural network model is trained using above-mentioned digital data sets, obtain digital identification model.
The advantages of basic principles and main features and this programme of this programme have been shown and described above.The technology of the industry
Personnel are it should be appreciated that this programme is not restricted to the described embodiments, and the above embodiments and description only describe this
The principle of scheme, under the premise of not departing from this programme spirit and scope, this programme will also have various changes and improvements, these changes
Change and improve and both falls within the scope of claimed this programme.This programme be claimed range by appended claims and its
Equivalent thereof.
Claims (5)
1. a kind of digital recognition system for financial statement, which is characterized in that including table detection module and number identification mould
Block;The table detection module includes image correction unit and table detection unit;
Described image corrects unit, for the financial statement image to be corrected, keeps its table change horizontal;
The table detection unit, for detecting the table line in the financial statement image after correcting, extraction unit table images;
The number identification module, for identification content in the unit table images.
2. a kind of digit recognition method for financial statement, which is characterized in that include the following steps:
S1, financial statement image is downloaded from server;
S2, image rectification is carried out to the financial statement image using based on Hough straight-line method, keeps table change horizontal;
S3, table line, and extraction unit table images are detected using morphological image;
S4, the number in the unit table images is identified using the digital identification model based on deep learning network successively, finally
Export the content of whole table.
3. digit recognition method as claimed in claim 2, which is characterized in that the method for described image correction is as follows:
S21,50 straight lines random in the financial statement image are obtained using Hough straight-line method, is taken if less than 50 all
Straight line;
S22, the slope according to 50 straight lines of acquisition, calculate G-bar;
S23, image is corrected according to obtained G-bar, keeps table change horizontal.
4. digit recognition method as claimed in claim 2, which is characterized in that the method for the extraction unit table images is as follows:
S31, binary conversion treatment is carried out to the financial statement image;
S32, the horizontal, vertical two kinds of strips of construction corrode core, are filtered to bianry image, lateral encroaching core handles to obtain the cross of table
Line, the vertical core that corrodes handle to obtain the vertical line of table;
The horizontal line and vertical line that S33, basis obtain, determine the position of each cell in table area.
5. digit recognition method as claimed in claim 2, which is characterized in that the obtaining step of the number identification model is such as
Under:
Unit table images under S41, acquisition different location, different noises;
S42, the position that each number in the unit table images is marked using rectangle frame, and record respective classes label;
S43, number identification deep neural network model is trained using above-mentioned digital data sets, obtain digital identification model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810228641.0A CN108470164A (en) | 2018-03-20 | 2018-03-20 | A kind of digital recognition system and method for financial statement |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810228641.0A CN108470164A (en) | 2018-03-20 | 2018-03-20 | A kind of digital recognition system and method for financial statement |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108470164A true CN108470164A (en) | 2018-08-31 |
Family
ID=63265613
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810228641.0A Pending CN108470164A (en) | 2018-03-20 | 2018-03-20 | A kind of digital recognition system and method for financial statement |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108470164A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110059687A (en) * | 2019-03-19 | 2019-07-26 | 平安科技(深圳)有限公司 | Pictorial information recognition methods, device, computer equipment and storage medium |
CN110458070A (en) * | 2019-08-01 | 2019-11-15 | 上海眼控科技股份有限公司 | Method and system based on motor vehicle annual test check table picture recognition amount of testing |
CN110929580A (en) * | 2019-10-25 | 2020-03-27 | 北京译图智讯科技有限公司 | Financial statement information rapid extraction method and system based on OCR |
CN111079756A (en) * | 2018-10-19 | 2020-04-28 | 杭州萤石软件有限公司 | Method and equipment for extracting and reconstructing table in document image |
CN111814598A (en) * | 2020-06-22 | 2020-10-23 | 吉林省通联信用服务有限公司 | Financial statement automatic identification method based on deep learning framework |
CN111881769A (en) * | 2020-07-03 | 2020-11-03 | 苏州开心盒子软件有限公司 | Method and system for table labeling |
-
2018
- 2018-03-20 CN CN201810228641.0A patent/CN108470164A/en active Pending
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111079756A (en) * | 2018-10-19 | 2020-04-28 | 杭州萤石软件有限公司 | Method and equipment for extracting and reconstructing table in document image |
CN111079756B (en) * | 2018-10-19 | 2023-09-19 | 杭州萤石软件有限公司 | Form extraction and reconstruction method and equipment in receipt image |
CN110059687A (en) * | 2019-03-19 | 2019-07-26 | 平安科技(深圳)有限公司 | Pictorial information recognition methods, device, computer equipment and storage medium |
CN110059687B (en) * | 2019-03-19 | 2024-05-28 | 平安科技(深圳)有限公司 | Picture information identification method, device, computer equipment and storage medium |
CN110458070A (en) * | 2019-08-01 | 2019-11-15 | 上海眼控科技股份有限公司 | Method and system based on motor vehicle annual test check table picture recognition amount of testing |
CN110929580A (en) * | 2019-10-25 | 2020-03-27 | 北京译图智讯科技有限公司 | Financial statement information rapid extraction method and system based on OCR |
CN111814598A (en) * | 2020-06-22 | 2020-10-23 | 吉林省通联信用服务有限公司 | Financial statement automatic identification method based on deep learning framework |
CN111881769A (en) * | 2020-07-03 | 2020-11-03 | 苏州开心盒子软件有限公司 | Method and system for table labeling |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108470164A (en) | A kind of digital recognition system and method for financial statement | |
CN109784326A (en) | A kind of vehicle chassis detection method based on deep learning | |
CN108491789A (en) | A kind of Chinese characters recognition method and device for financial statement | |
CN108491788A (en) | A kind of intelligent extract method and device for financial statement cell | |
CN107958201A (en) | A kind of intelligent checking system and method for vehicle annual test insurance policy form | |
CN107145810A (en) | A kind of comprehensive bar code identifying device and method | |
CN107818322A (en) | A kind of vehicle VIN code tampering detection system and methods for vehicle annual test | |
CN105447530A (en) | Power transmission line hidden risk and fault detection method based on image identification technology | |
CN106127205A (en) | A kind of recognition methods of the digital instrument image being applicable to indoor track machine people | |
CN110728225A (en) | High-speed face searching method for attendance checking | |
CN108491787A (en) | A kind of date detection and recognition methods and device for financial statement | |
CN107784469A (en) | A kind of articles from the storeroom checking method based on unmanned plane image technique | |
CN109784343B (en) | Resource allocation method and terminal based on deep learning model | |
CN107886047A (en) | A kind of detecting system and method for vehicle annual test survey report | |
CN109815864A (en) | A kind of facial image age recognition methods based on transfer learning | |
CN110245583A (en) | A kind of intelligent identification Method of Vehicular exhaust survey report | |
CN107818321A (en) | A kind of watermark date recognition method for vehicle annual test | |
CN111144749A (en) | Intelligent labeling crowdsourcing platform for power grid images and working method | |
CN112465809A (en) | Mold defect detection method based on image recognition, computer-readable storage medium and device | |
CN104298993A (en) | Pole number positioning and identification method suitable for railway line in complex scene | |
CN104112123A (en) | Defect characteristic extraction and identification method of AOI system used for bullet apparent defect detection | |
CN103699876A (en) | Method and device for identifying vehicle number based on linear array CCD (Charge Coupled Device) images | |
CN114862821A (en) | Automatic monitoring method, system and equipment for relay protection hard pressing plate | |
CN116664431B (en) | Image processing system and method based on artificial intelligence | |
CN109447285A (en) | A kind of protection pressing plate state identification method based on self comparison of figure |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180831 |
|
WD01 | Invention patent application deemed withdrawn after publication |