CN103605973B - A kind of image character detection and identification method - Google Patents
A kind of image character detection and identification method Download PDFInfo
- Publication number
- CN103605973B CN103605973B CN201310511403.8A CN201310511403A CN103605973B CN 103605973 B CN103605973 B CN 103605973B CN 201310511403 A CN201310511403 A CN 201310511403A CN 103605973 B CN103605973 B CN 103605973B
- Authority
- CN
- China
- Prior art keywords
- image
- character
- black white
- white image
- row
- 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.)
- Active
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a kind of image character detection and identification method, its method and step is：First, make corresponding character standard black white image；2nd, Real-time Collection test image, extracts to character position and is processed as to test black white image；3rd, step 2 is obtained test black white image with the standard black white image that step one makes carry out contrast judge whether qualified.The method can be fully solved the problem of image character None- identified in flat panel TV is tested automatically, and propulsion flat panel TV cartridge assemblies realize Aulomatizeted Detect, Improving The Quality of Products and people's effect.
Description
Technical field
The present invention relates to a kind of image character detection and identification method, more particularly to one kind is applied to flat panel TV and automatically surveys
The method of the image character detection and identification to collection for the examination field.
Background technology
Due to, during flat panel TV cartridge assemblies processing technique, limiting because of supplied materials device with production technology, warp
Cartridge assemblies after the completion of welding have different degrees of defect.The method of existing test-based examination is substantially manual wiring or needle-bar
Connect display screen, the Quality estimation to image deflects is come by artificial range estimation, the method has the mistake that artificially subjective factor leads to
The impact sentence, failed to judge.And the continuous rising with human cost, Aulomatizeted Detect progressively applies.It is right to need in automatic detection
Collection image character detected, in industry more using the character recognition of general purpose O CR by the way of, in actual applications due to
In collection image the problems such as the brightness of character, thickness, stroke fracture or adhesion, cause to have more erroneous judgement in OCR identification, identification is tired
Difficult.
Content of the invention
The technical problem to be solved in the present invention is that offer is a kind of to be judged by accident less, identifies and readily reflect for detecting to image character
Other method.
The technical solution used in the present invention is as follows：A kind of image character detection and identification method, its method and step is：First, make
Make corresponding character standard black white image；2nd, Real-time Collection test image, extracts to character position and is processed as to test artwork master
Picture；3rd, test black white image step 2 being obtained carries out contrast with the standard black white image that step one makes and judges whether to close
Lattice.
As to its further improvement, the concrete grammar step of described step one is：1st, choose a width to pass through to adopt in real time
The testing standard image of collection；2nd, the testing standard image that frame choosing required detection character picture position obtains to step 1 extracts；
3rd, set a gray threshold P, the part that gradation of image is more than gray threshold P is defined as character pixels point, will carry in step 2
The image got carries out binary conversion treatment by gray threshold P and obtains standard black white image.
As to its further improvement, described testing standard image is the position being gone out by image capturing system Real-time Collection
Figure image.
As to its further improvement, described black white image is that character is fixed taking 8Bit-256 colours BITMAP image as a example
Justice is defined as, for completely black, background that gray value is 0, the entirely white 8Bit-256 colours BITMAP image that gray value is 255.
As to its further improvement, the concrete grammar step that described gray threshold P sets as：A, will be by capture card
The RGB bitmap images of collection are converted to gray level image by formula Gray=R*0.299+G*0.587+B*0.114；B, by obtain
Gray level image carries out grey level quantization by N number of gray scale interval, draws N number of rectangular histogram；C, from N number of rectangular histogram choose gray-scale pixels divide
Two the most intensive intervals of cloth, the intermediate value defining this two interval gray scales is respectively Gray1 and Gray2；D, will in step B select
Gray value P=(Gray1+ Gray2)/2 in the middle of fixed two are interval is defined as gray threshold.
As to its further improvement, described N number of be 10.
As to its further improvement, carrying out in described step 3 contrasting concrete grammar it is：Extract the entirety of character
Housing carries out contrast and judges.
As to its further improvement, the concrete grammar that the described overall housing extracting character is contrasted is：a、
Read binaryzation standard black white image and test black white image pixel be identical x row × y row, definition image for f (x,
y)；B, from image coordinate initial point start to standard black white image and test black white image carry out boundary search simultaneously；C, press respectively
Columns and rows scan for, by by row search containing gray value be 0 pixel in row minimum and maximum value be respectively defined as xminWith
xmax；, by by line search to containing gray value be 0 pixel in row minimum and maximum value be respectively defined as yminAnd ymax；D, basis
The character boundary maximum searching in row, column and minima, i.e. f (xmin,ymin) and f (xmax,ymax) 2 points can select figure by frame
As the overall housing of character, pointwise contrast is carried out to the overall housing of standard black white image and test black white image character, unanimously
Property reach setting value be then judged as qualified.
As to its further improvement, the concrete grammar carrying out contrasting judgement in described step 3 also includes：If whole
It is qualified that external frame carries out contrasting, and carries out the point-to-point contrast of single character and judges, its concrete grammar is：From standard black white image and
(the x of test black white image f (x, y)min,ymin) coordinate points start, row or column is scanned for, with 8Bit-256 colours BITMAP image
As a example row search, to standard black white image and black white image according to formula,By row search, when running into first E=0, the value of x may be defined as first character row right margin x1；By
This can be by (xmin,ymin) and (x1,ymin) 2 points of frames select first character；Second character of search is from the pixel of f (x, y)
(x1,ymin) start by row search, it is the left margin of second character when E ≠ 0, the value of record x is x2；Continue by row search
During to next E=0, the value of x is defined as second character row right margin x2；Thus can be by (x2,ymin) and (x3,ymax) two
Second character selected by point frame；The like until search for the pixel (x of f (x, y)max,ymax) can complete to search for；To mark
The character that quasi- black white image and artwork master frame are selected sequentially 1,2,3～N, carry out each pixel contrast, concordance one by one
Reach setting value be then judged as qualified.
Compared with prior art, the invention has the beneficial effects as follows：The method can be fully solved and automatically test in flat panel TV
The problem of middle image character None- identified, propulsion flat panel TV cartridge assemblies realize Aulomatizeted Detect, Improving The Quality of Products and
People is imitated.
Brief description
Fig. 1 is that the binary image housing of a present invention wherein embodiment searches for schematic diagram.
Fig. 2 is the method flow schematic diagram of a present invention wherein embodiment.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, below in conjunction with drawings and Examples, right
The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only in order to explain the present invention, not
For limiting the present invention.
This specification（Including any accessory claim, summary and accompanying drawing）Disclosed in any feature, except non-specifically is chatted
State, all can be replaced by other alternative features equivalent or that there is similar purpose.I.e., unless specifically stated otherwise, each feature
One of simply a series of equivalent or similar characteristics example.
A kind of image character detection and identification method, its method and step is：First, make corresponding character standard black white image；2nd,
Real-time Collection test image, extracts to character position and is processed as to test black white image；3rd, test black and white step 2 being obtained
Image synchronization rapid one making standard black white image carry out contrast judge whether qualified.
In this specific embodiment, the concrete grammar step of described step one is：1st, choose a width and pass through Real-time Collection
The testing standard image of LVDS output；2nd, the testing standard image that frame choosing required detection character picture position obtains to step 1 enters
Row extracts；3rd, set a gray threshold P, the part that gradation of image is more than gray threshold P is defined as character pixels point, will be to step
The image extracting in rapid 2 carries out binary conversion treatment by gray threshold P and obtains standard black white image.
In this specific embodiment, described testing standard image is the 8Bit- being gone out by image capturing system Real-time Collection
256 colours BITMAP image.Black white image be by character definition for completely black, background that gray value is 0 be defined as gray value be 255 complete
White 8Bit-256 colours BITMAP image.
In this specific embodiment, concrete grammar step that described gray threshold P sets as：A, will be gathered by capture card
RGB bitmap images be converted to gray level image by formula Gray=R*0.299+G*0.587+B*0.114；B, by the gray scale obtaining
Image presses 10（Optimum analysis scheme）Gray scale interval carries out grey level quantization, draws 10 rectangular histograms；C, from 10 rectangular histograms
Choose the most intensive two intervals of gray-scale pixels distribution, define this two interval gray scales intermediate value be respectively Gray1 and
Gray2；D, by two selected in step B interval in the middle of gray value P=(Gray1+ Gray2)/2 be defined as gray threshold.
The overall housing extracting character in step 3 carries out contrast judgement.Concrete grammar is：A, reading binaryzation standard
Black white image and test black white image pixel are identical x row × y row, and definition image is f (x, y)；B, from image coordinate
Initial point starts to carry out boundary search to standard black white image and test black white image simultaneously；C, scan for by columns and rows respectively,
By by row search containing gray value be 0 pixel in row minimum and maximum value be respectively defined as xminAnd xmax；, will be by line search
To containing gray value be 0 pixel in row minimum and maximum value be respectively defined as yminAnd ymax；D, according to search in row, column
Character boundary maximum and minima, i.e. f (xmin,ymin) and f (xmax,ymax) 2 points can select the overall outer of image character by frame
Frame, carries out pointwise contrast to the overall housing of standard black white image and test black white image character, concordance reaches setting value then
It is qualified to be judged as.In this specific embodiment, this setting value is 99.9%, is judged as unqualified less than 99.9%, equal to being higher than
99.9% be judged as qualified.
As to its further improvement, the concrete grammar carrying out contrasting judgement in described step 3 also includes：If whole
It is qualified that external frame carries out contrasting, and carries out the point-to-point contrast of single character and judges, its concrete grammar is：From standard black white image and
(the x of test black white image f (x, y)min,ymin) coordinate points start, row or column is scanned for, taking arrange search as a example, to standard
Black white image and black white image are according to formula,By row search, when running into
During first E=0, the value of x may be defined as first character row right margin x1；Thus can be by (xmin,ymin) and (x1,ymin) two
First character selected by point frame；Second character of search is from the pixel (x of f (x, y)1,ymin) start by row search, when E ≠ 0
When be second character left margin, record x value be x2；When continuing by row search extremely next E=0, the value of x is i.e. fixed
Justice is second character row right margin x2；Thus can be by (x2,ymin) and (x3,ymax) 2 points of frames select second character；Class successively
Push away until searching for the pixel (x of f (x, y)max,ymax) can complete to search for；To standard black white image and the choosing of artwork master frame
The character going out sequentially 1,2,3～N, carry out each pixel contrast one by one, concordance reach setting value be then judged as qualified.?
In this specific embodiment, this setting value is 99.99%, is then judged as unqualified less than this value, is then judged as closing equal to higher than this value
Lattice, thus improve accuracy rate further.
Claims (7)
1. a kind of image character detection and identification method, its method and step is：First, make corresponding character standard black white image；2nd, real
When collecting test image, to character position extract and be processed as test black white image；3rd, test artwork master step 2 being obtained
As with step one make standard black white image carry out contrast judge whether qualified；Carry out in described step 3 contrasting concrete grammar
For：The overall housing extracting character carries out contrast judgement；The concrete side that the described overall housing extracting character is contrasted
Method is：A, reading binaryzation standard black white image and test black white image pixel are identical x row × y row, define image
For f (x, y)；B, from image coordinate initial point start to standard black white image and test black white image carry out boundary search simultaneously；c、
Scan for by columns and rows respectively, by by row search containing gray value be 0 pixel in row minimum and maximum value define respectively
For xminAnd xmax；By by line search to containing gray value be 0 pixel in row minimum and maximum value be respectively defined as yminAnd ymax；
D, according to the character boundary maximum searching in row, column and minima, i.e. f (xmin,ymin) and f (xmax,ymax) 2 points can frame
Select the overall housing of image character, pointwise pair is carried out to the overall housing of standard black white image and test black white image character
Than, concordance reach setting value be then judged as qualified.
2. method according to claim 1, the concrete grammar step of described step one is：1st, choose a width to pass through to adopt in real time
The testing standard image of collection；2nd, the testing standard image that frame choosing required detection character picture position obtains to step 1 extracts；
3rd, set a gray threshold P, the part that gradation of image is more than gray threshold P is defined as character pixels point, will carry in step 2
The image got carries out binary conversion treatment by gray threshold P and obtains standard black white image.
3. method according to claim 2, described testing standard image is gone out by image capturing system Real-time Collection
Bitmap images.
4. method according to claim 3, described black white image is that character is fixed taking 8Bit-256 colours BITMAP image as a example
Justice is defined as, for completely black, background that gray value is 0, the entirely white 8Bit-256 colours BITMAP image that gray value is 255.
5. method according to claim 2, the concrete grammar step that described gray threshold P sets as：A, will by collection
The RGB bitmap images of card collection are converted to gray level image by formula Gray=R*0.299+G*0.587+B*0.114；B, will obtain
Gray level image carry out grey level quantization by N number of gray scale interval, draw N number of rectangular histogram；C, from N number of rectangular histogram choose gray-scale pixels
It is distributed two the most intensive intervals, the intermediate value defining this two interval gray scales is respectively Gray1 and Gray2；D, by step B
Gray value P=(Gray1+Gray2)/2 in the middle of selected two are interval is defined as gray threshold.
6. method according to claim 5, described N number of be 10.
7. method according to claim 1, the concrete grammar carrying out contrasting judgement in described step 3 also includes：If whole
It is qualified that external frame carries out contrasting, and carries out the point-to-point contrast of single character and judges, its concrete grammar is：From standard black white image and
(the x of test black white image f (x, y)min,ymin) coordinate points start, row or column is scanned for, with 8Bit-256 colours BITMAP image
As a example row search, to standard black white image and black white image according to formulaBy row search, when running into first E=0, the value of x
It is defined as first character row right margin x1；Thus can be by (xmin,ymin) and (x1,ymin) 2 points of frames select first character；Search
Second character is from the pixel (x of f (x, y)1,ymin) start by row search, it is the left margin of second character when E ≠ 0,
The value of record x is x2；When continuing by row search extremely next E=0, the value of x is defined as second character row right margin
x2；Thus can be by (x2,ymin) and (x3,ymax) 2 points of frames select second character；The like until search for the picture of f (x, y)
Vegetarian refreshments (xmax,ymax) can complete to search for；The character that standard black white image and artwork master frame are selected sequentially 1,2,3～N,
Carry out each pixel contrast one by one, concordance reach setting value be then judged as qualified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310511403.8A CN103605973B (en) | 2013-10-25 | 2013-10-25 | A kind of image character detection and identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310511403.8A CN103605973B (en) | 2013-10-25 | 2013-10-25 | A kind of image character detection and identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103605973A CN103605973A (en) | 2014-02-26 |
| CN103605973B true CN103605973B (en) | 2017-03-01 |
Family
ID=50124193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310511403.8A Active CN103605973B (en) | 2013-10-25 | 2013-10-25 | A kind of image character detection and identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103605973B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104240348B (en) * | 2014-09-12 | 2017-02-15 | 福建省智慧物联网研究院有限责任公司 | Admittance identity authentication method based on image identification |
| CN104504385B (en) * | 2014-12-10 | 2017-12-22 | 上海大学 | The recognition methods of hand-written adhesion numeric string |
| CN105424190A (en) * | 2015-09-30 | 2016-03-23 | 广州超音速自动化科技股份有限公司 | Grayscale detection method of product appearance |
| CN110427885A (en) * | 2019-07-31 | 2019-11-08 | Tcl王牌电器(惠州)有限公司 | Detection method, device and the computer readable storage medium of nameplate |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1567340A (en) * | 2003-06-23 | 2005-01-19 | 中国科学院研究生院 | False proof bill, false proof method of bill and system thereof |
| CN1975804A (en) * | 2006-12-15 | 2007-06-06 | 华南理工大学 | Education robot with character-learning and writing function and character recognizing method thereof |
| CN103049743A (en) * | 2012-12-27 | 2013-04-17 | 天津普达软件技术有限公司 | Detection system and detection method for characters at bottom of bowl |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001235426A (en) * | 1999-12-22 | 2001-08-31 | Novartis Ag | Package inspection instrument |
-
2013
- 2013-10-25 CN CN201310511403.8A patent/CN103605973B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1567340A (en) * | 2003-06-23 | 2005-01-19 | 中国科学院研究生院 | False proof bill, false proof method of bill and system thereof |
| CN1975804A (en) * | 2006-12-15 | 2007-06-06 | 华南理工大学 | Education robot with character-learning and writing function and character recognizing method thereof |
| CN103049743A (en) * | 2012-12-27 | 2013-04-17 | 天津普达软件技术有限公司 | Detection system and detection method for characters at bottom of bowl |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103605973A (en) | 2014-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2016150134A1 (en) | Test paper reading method, and pregnancy test and ovulation test method therefor | |
| CN105913093A (en) | Template matching method for character recognizing and processing | |
| CN104616275A (en) | Defect detecting method and defect detecting device | |
| US20090040215A1 (en) | Interpreting Sign Language Gestures | |
| CN104574353B (en) | The surface defect decision method of view-based access control model conspicuousness | |
| CN103731664B (en) | Full reference video quality appraisal procedure, device and video quality tester | |
| CN106251333B (en) | Element reverse detection method and system | |
| CN103605973B (en) | A kind of image character detection and identification method | |
| CN108764352B (en) | Method and device for detecting repeated page content | |
| CN107480607B (en) | Method for detecting and positioning standing face in intelligent recording and broadcasting system | |
| CN106355579A (en) | Defect detecting method of cigarette carton surface wrinkles | |
| CN113706490B (en) | Wafer defect detection method | |
| CN111127448B (en) | Method for detecting air spring fault based on isolated forest | |
| CN103218603A (en) | Face automatic labeling method and system | |
| WO2018010386A1 (en) | Method and system for component inversion testing | |
| CN109350030A (en) | System and method based on phase enhanced processing face video heart rate signal | |
| TW201317904A (en) | Tag detecting system, apparatus and method for detecting tag thereof | |
| CN106709438A (en) | Method for collecting statistics of number of people based on video conference | |
| CN110222678A (en) | A kind of item analysis method, system, readable storage medium storing program for executing and electronic equipment | |
| CN109886945A (en) | Based on contrast enhancing without reference contrast distorted image quality evaluating method | |
| CN106934770A (en) | A kind of method and apparatus for evaluating haze image defog effect | |
| CN111199194A (en) | Automobile intelligent cabin instrument testing method based on machine vision and deep learning | |
| CN104657714B (en) | Illumination symmetry merged with global illumination intensity without refer to human face light evaluation method | |
| CN110415223B (en) | No-reference spliced image quality evaluation method and system | |
| CN110443803A (en) | A kind of printed matter picture quality detection method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 |