CN109460758A - A kind of automatic wisdom grain depot form recognition method and its system - Google Patents
A kind of automatic wisdom grain depot form recognition method and its system Download PDFInfo
- Publication number
- CN109460758A CN109460758A CN201811406943.9A CN201811406943A CN109460758A CN 109460758 A CN109460758 A CN 109460758A CN 201811406943 A CN201811406943 A CN 201811406943A CN 109460758 A CN109460758 A CN 109460758A
- Authority
- CN
- China
- Prior art keywords
- information
- grain depot
- region
- image
- module
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
Abstract
The invention discloses a kind of automatic wisdom grain depot form recognition method and its systems, and this method comprises the following steps: (1) obtaining the image information of grain depot document；(2) image procossing is carried out to the image information of acquisition；(3) by treated, image carries out region division, divides are as follows: bill name region, information area and table area；(4) contours extract is carried out to the font in each region and identifies corresponding character contour, the character contour of corresponding region is saved in corresponding information storage area.The present invention can carry out automatic input to the information of grain depot document, carry out manual typing without staff, improve work efficiency, reduce cost of labor.
Description
Technical field
The present invention relates to grain depot data input technical fields, and in particular to a kind of automatic wisdom grain depot form recognition method and
Its system.
Background technique
China is grain-production and consumption big country and grain reserves big country, grain are the commodity concerning national economy, is protected
The development that its abundance supply is related to social stability and national economy with quality safety is demonstrate,proved, China all establishes have for a long time thus
Corresponding grain reserve system.
With the development for calculating industry, various grain depot Application of Monitoring System and give birth to, but present grain depot monitors system base
This is copied on computers by being transcribed again by the table that staff will fill in after manually filling in a form, this mode not only working efficiency
It is low, also add cost of labor.
Summary of the invention
The purpose of the present invention is to provide a kind of automatic wisdom grain depot documents that work efficiency is high and can reduce cost of labor
Recognition methods and its system.
The purpose of the present invention is what is be achieved through the following technical solutions:
A kind of automatic wisdom grain depot form recognition method, includes the following steps:
(1) image information of grain depot document is obtained；
(2) image procossing is carried out to the image information of acquisition；
(3) by treated, image carries out region division, divides are as follows: bill name region, information area and table area；
(4) contours extract is carried out to the font in each region and identifies corresponding character contour, by the word of corresponding region
Body profile is saved in corresponding information storage area.
Further, the acquisition of described image information is obtained by camera or scanning means.
Further, described image processing refers to that image information is first carried out to gray proces carries out binary conversion treatment again.
Further, the font progress contours extract in each region, which refers to, first passes through convolutional neural networks module
The space characteristics in images to be recognized are extracted, space characteristics are then input to abstraction sequence feature in nested LSTM module, most
It is entered and left afterwards according to space characteristics and sequence signature to RPN module and obtains character contour.
Further, the corresponding character contour of the identification refer to by the character contour of extraction and the character contour that prestores into
Row comparison obtains corresponding text information.
The present invention also provides a kind of automatic wisdom grain depot form recognition systems, comprising:
Image collection module, for obtaining the image information of grain depot document；
Image processing module, for carrying out image procossing to image information；
Region division module divides are as follows: bill name area for carrying out region division to the image information after image procossing
Domain, information area and table area；
Text information identification module, for carrying out contours extract to the text in each region and identifying its text information；
Typing preserving module, to get to corresponding in the information storage area for thinking the text information typing of identification
The text information in region completes the data input and preservation of a grain depot document.
The invention has the following advantages:
(1) automatic input can be carried out to the information of grain depot document, carries out manual typing without staff, improves work
Efficiency reduces cost of labor；
(2) by first carrying out gray proces binary conversion treatment again to image information, after an action of the bowels it is subsequent division and text wheel
Wide extraction；
(3) space characteristics in convolutional neural networks module extraction images to be recognized are first passed through, it is then that space characteristics are defeated
Enter the abstraction sequence feature into nested LSTM module, is finally entered and left according to space characteristics and sequence signature to RPN module and obtain word
Body profile, recognition speed is fast and effect is good；
(4) text information is obtained compared with the text profile prestored by text profile, is applicable not only to type-script text
Apply also for the text of handwritten form.
Detailed description of the invention
Fig. 1 is automatic wisdom grain depot form recognition method flow schematic diagram of the invention.
Specific embodiment
Embodiment 1
Automatic wisdom grain depot form recognition method provided in this embodiment includes the following steps:
(1) document of grain depot is scanned by camera or scanning means to obtain the image of grain depot document letter
Breath；
(2) image procossing is carried out to the image information of acquisition, described image processing refers to first carry out gray proces after into
Row binary conversion treatment, obtains gray level image information；
(3) gray level image information is subjected to region division, divided are as follows: bill name region, information area and table area；
The bill name region is title information area, specially goes out storage registration form, advice of settlement or other records
Deng the information area is the information area filled in, and the information area of each table has all differences, can be practical according to each
Table is configured, and the table area is also the region filled in, such as weight, inspection result etc., after being divided more
It is suitble to text information to read, the region after division can also be divided again, quickly to obtain text information；
(4) contours extract is carried out to the font in each region and identifies corresponding character contour, by the word of corresponding region
Body profile is saved in corresponding information storage area, and it is by first rolling up that the font in each region, which carries out contours extract,
Product neural network module extracts the space characteristics in images to be recognized, and then space characteristics are input in nested LSTM module and are mentioned
Sequence signature is taken, is finally entered and left according to space characteristics and sequence signature to RPN module and obtains character contour, this identification text wheel
Wide recognition speed is fast and effect is good, is compared after extracting text profile with the text profile prestored, can quickly obtain text
Word information, the locating text profile prestored can be computerized print body can also be with handwritten form, and the information for being more suitable for grain depot document obtains
It taking, text information is stored in corresponding information storage area to get to the grain depot document of an electronic edition after obtaining, with
And its data are filled in, it is finally stored, grain depot document is opened in typing one.
Embodiment 2
The present invention also provides a kind of automatic wisdom grain depot form recognition systems to include:
Image collection module, for obtaining the image information of grain depot document, by camera or scanning means to grain depot
Document be scanned to obtain the image information of grain depot document；
Image processing module carries out image procossing to the image information of acquisition for carrying out image procossing to image information,
Described image processing is carrying out binary conversion treatment after referring to first progress gray proces, obtains gray level image information；
Region division module divides are as follows: bill name area for carrying out region division to the image information after image procossing
Domain, information area and table area, the bill name region be title information area, specially go out storage registration form,
Advice of settlement or other records etc., the information area is the information area filled in, and the information area of each table is
It is different, it can be actually configured according to each table, the table area is also the region filled in, such as weight, inspection
Come to an end fruit etc., and text information reading is more suitable for after being divided, can also be divided again to the region after division, with quick
Acquisition text information；
Text information identification module, for carrying out contours extract to the text in each region and identifying its text information,
The font in each region carries out contours extract and first passes through in convolutional neural networks module extraction images to be recognized
Then space characteristics are input to abstraction sequence feature in nested LSTM module, finally according to space characteristics and sequence by space characteristics
Column feature enters and leaves to RPN module and obtains character contour.The recognition speed of this identification text profile is fast and effect is good, extracts text
It is compared after profile with the text profile prestored, can quickly show that text information, the locating text profile prestored can be
Computerized print body can also be more suitable for the acquisition of information of grain depot document with handwritten form, and text information is stored to corresponding after obtaining
Information storage area in get to the grain depot document of electronic edition and its fill in data.
Typing preserving module, to get to corresponding in the information storage area for thinking the text information typing of identification
The text information in region completes the data input and preservation of a grain depot document.
The above is only the preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto, any
The transformation and replacement carried out based on technical solution provided by the present invention and inventive concept should all be covered in protection model of the invention
In enclosing.
Claims (6)
1. a kind of automatic wisdom grain depot form recognition method, it is characterised in that include the following steps:
(1) image information of grain depot document is obtained；
(2) image procossing is carried out to the image information of acquisition；
(3) by treated, image carries out region division, divides are as follows: bill name region, information area and table area；
(4) contours extract is carried out to the font in each region and identifies corresponding character contour, by the font wheel of corresponding region
Exterior feature is saved in corresponding information storage area.
2. automatic wisdom grain depot form recognition method according to claim 1, it is characterised in that: described image information obtains
Take is obtained by camera or scanning means.
3. automatic wisdom grain depot form recognition method according to claim 1, it is characterised in that: described image processing refers to
Image information is first carried out to gray proces and carries out binary conversion treatment again.
4. automatic wisdom grain depot form recognition method according to claim 1, it is characterised in that: described in each region
Font carry out contours extract refer to first pass through convolutional neural networks module extract images to be recognized in space characteristics, then will
Space characteristics are input to abstraction sequence feature in nested LSTM module, are finally entered and left according to space characteristics and sequence signature to RPN
Module obtains character contour.
5. automatic wisdom grain depot form recognition method according to claim 4, it is characterised in that: the corresponding word of identification
Body profile, which refers to compare the character contour of extraction with the character contour prestored, obtains corresponding text information.
6. a kind of automatic wisdom grain depot form recognition system, characterized by comprising:
Image collection module, for obtaining the image information of grain depot document；
Image processing module, for carrying out image procossing to image information；
Region division module is divided for carrying out region division to the image information after image procossing are as follows: bill name region,
Information area and table area；
Text information identification module, for carrying out contours extract to the text in each region and identifying its text information；
Typing preserving module arrives corresponding region in the information storage area for thinking the text information typing of identification
Text information complete data input and the preservation of grain depot document.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406943.9A CN109460758A (en) | 2018-11-23 | 2018-11-23 | A kind of automatic wisdom grain depot form recognition method and its system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811406943.9A CN109460758A (en) | 2018-11-23 | 2018-11-23 | A kind of automatic wisdom grain depot form recognition method and its system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109460758A true CN109460758A (en) | 2019-03-12 |
Family
ID=65611505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811406943.9A Pending CN109460758A (en) | 2018-11-23 | 2018-11-23 | A kind of automatic wisdom grain depot form recognition method and its system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109460758A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460959A (en) * | 2020-03-27 | 2020-07-28 | 北京大学 | Document management method and related device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203416A (en) * | 2016-07-08 | 2016-12-07 | 武汉工程大学 | A kind of express delivery document information input method based on scanogram and input device |
| CN107679531A (en) * | 2017-06-23 | 2018-02-09 | 平安科技（深圳）有限公司 | Licence plate recognition method, device, equipment and storage medium based on deep learning |
| US10032072B1 (en) * | 2016-06-21 | 2018-07-24 | A9.Com, Inc. | Text recognition and localization with deep learning |
-
2018
- 2018-11-23 CN CN201811406943.9A patent/CN109460758A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10032072B1 (en) * | 2016-06-21 | 2018-07-24 | A9.Com, Inc. | Text recognition and localization with deep learning |
| CN106203416A (en) * | 2016-07-08 | 2016-12-07 | 武汉工程大学 | A kind of express delivery document information input method based on scanogram and input device |
| CN107679531A (en) * | 2017-06-23 | 2018-02-09 | 平安科技（深圳）有限公司 | Licence plate recognition method, device, equipment and storage medium based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 贾旸: "单证识别领域的OCR技术应用探索", 《金融电子化》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460959A (en) * | 2020-03-27 | 2020-07-28 | 北京大学 | Document management method and related device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108664996B (en) | Ancient character recognition method and system based on deep learning | |
| CN105138652B (en) | A kind of enterprise's incidence relation recognition methods and system | |
| CN108805076B (en) | Method and system for extracting table characters of environmental impact evaluation report | |
| CN107301244A (en) | Method, device, system and the trade mark memory of a kind of trade mark point card processing | |
| US20100290701A1 (en) | Document image segmentation system | |
| CN110795919A (en) | Method, device, equipment and medium for extracting table in PDF document | |
| CN103093240A (en) | Calligraphy character identifying method | |
| CN107944718A (en) | A kind of business policy assessment system and method | |
| CN104123550A (en) | Cloud computing-based text scanning identification method | |
| CN109784330B (en) | Signboard content identification method, device and equipment | |
| CN105117741A (en) | Recognition method of calligraphy character style | |
| CN104463091B (en) | A kind of facial image recognition method based on image LGBP feature subvectors | |
| CN101118592B (en) | Printers evidence obtaining method based on character printing feature | |
| CN109460758A (en) | A kind of automatic wisdom grain depot form recognition method and its system | |
| CN114463767A (en) | Credit card identification method, device, computer equipment and storage medium | |
| CN112200789B (en) | Image recognition method and device, electronic equipment and storage medium | |
| Pratama et al. | Indonesian ID card recognition using convolutional neural networks | |
| Arsenovic et al. | Deep learning driven plates recognition system | |
| Dhandra et al. | Kannada handwritten vowels recognition based on normalized chain code and wavelet filters | |
| Antony et al. | A framework for recognition of handwritten South Dravidian Tulu script | |
| Prommas et al. | CNN-based Thai handwritten OCR: an application for automated mail sorting | |
| CN106709502B (en) | Multi-feature fusion recognition method based on voting method | |
| CN114926842A (en) | Dongba pictograph recognition method and device | |
| CN112149654A (en) | Invoice text information identification method based on deep learning | |
| CN113657279A (en) | Bill image layout analysis 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190312 |