CN100377169C - Method for picture binaryzation - Google Patents
Method for picture binaryzation Download PDFInfo
- Publication number
- CN100377169C CN100377169C CNB2005100800506A CN200510080050A CN100377169C CN 100377169 C CN100377169 C CN 100377169C CN B2005100800506 A CNB2005100800506 A CN B2005100800506A CN 200510080050 A CN200510080050 A CN 200510080050A CN 100377169 C CN100377169 C CN 100377169C
- Authority
- CN
- China
- Prior art keywords
- binaryzation
- image
- histogram
- gray shade
- shade scale
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000013213 extrapolation Methods 0.000 claims description 10
- 230000002902 bimodal effect Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 4
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 206010000234 Abortion spontaneous Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 208000015994 miscarriage Diseases 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 208000000995 spontaneous abortion Diseases 0.000 description 1
Images
Abstract
The present invention relates to an image processing technique in computer information processing fields, particularly to a method for image binaryzation. In the prior art, when a binaryzation method is carried out to digitized images through computer systems, a gradation histogram is processed to normal double-humped images, and the effect is relatively good generally, but when the double humps have obvious deviation, for example, when the brightness of the images is too high, too shallow or too deep images after binaryzation are caused by that a binaryzation field value can not be accurately selected. Thus, rear continuous processing to the images is influenced, such as character recognition. The method of the present invention improves the quality of the images after the binaryzation through compensating the gradation histogram of the binaryzation images and using a conventional binaryzation field value determining method. The method of the present invention can effectively improve the binaryzation effect of the double-humped deviation images in the gradation histogram.
Description
Technical field
The present invention relates to the image processing techniques in computer information processing field, be specifically related to a kind of method of image binaryzation.
Background technology
Image binaryzation is a kind of very useful image processing techniques, is used for gray level image is converted into monochrome image.Image is carried out binaryzation, can effectively reduce the storage space that image takies, convenient transmission; Simultaneously a lot of correlation techniques must be used the image after the binaryzation, as literal identification (OCR).The quality quality of image binaryzation not only has influence on the subjective quality evaluation of image, also can directly have influence on the subsequent treatment link, directly has influence on the discrimination of OCR such as meeting.
In the prior art, the image binaryzation basic ideas are by determining a thresholding, the point in the image is carried out pointwise judge that less than the stain that is of (or equaling) thresholding, greater than thresholding is white point.Therefore the key of binaryzation is determining of thresholding.Determining of binaryzation thresholding generally is that grey level histogram computing by image obtains, and basic ideas are that the point in the image is divided into two classes, look for the binaryzation thresholding under the minimum probability of miscarriage of justice.Document " image binaryzation algorithm research and realization thereof " [scientific and technological information exploitation and economy, 2004 the 14th the 12nd phases of volume, author Lv Junzhe] has been summed up the method that the binaryzation thresholding is chosen preferably.Also have a lot of documents to relate to multiple domain value choosing method, overall and local thresholding choosing method, and corresponding improvement strategy, as document " a kind of improved text image binaryzation algorithm " [computer engineering, the 29th the 13rd phase of volume, in August, 2003, author Chen Dan etc.] described a kind of of local thresholding and improved one's methods, patent " method of image binaryzation: " [Chinese patent application number 00808969.8] has been described the image binaryzation method of multi-grey level for another example, and patent " gray level image binary processing system and method " [Chinese patent application number 98119135] has been described and a kind ofly put the local thresholding method of determining according to neighbours.
The digitized image that general digitizer produces, its grey level histogram majority show as the not bimodal normally or multimodal feature of skew, and prior art can be handled this class image preferably.But when the grey level histogram of digital picture produces skew, make color obviously shallow partially or dark partially, show in the grey level histogram, be that the leftmost side or the rightmost side are when unusual peak occurring, it is dark or shallow excessively that the binaryzation result that prior art is handled can cause image to be crossed, and causes foreground/background effectively not separate and make image quality decrease.This general who has surrendered down greatly influences subjective assessment or influences follow-up processing procedure, as causes the literal discrimination to descend.
Summary of the invention
At in the prior art to the deficiency of image binaryzation, the objective of the invention is to propose a kind of improved image binaryzation method, this method can have the better binary conversion result to the image that skew takes place histogram, thereby the image subjective quality is improved, and can effectively improve the discrimination of OCR.
For realizing above purpose, the technical solution used in the present invention is: a kind of method of image binaryzation may further comprise the steps:
(1) at first adds up the gray shade scale histogram of pending digital picture;
(2) check that whether the gray shade scale histogram exists skew, advances the gray shade scale histogram that is offset
Row histogram edge compensation is asked the binaryzation thresholding to the gray shade scale histogram after the compensation;
(3) image is carried out binaryzation by thresholding.
In step (1), establishing gray shade scale is N, and the statistic histogram that obtains is H[N], each element H[n in the statistic histogram] represent that gray shade scale is the number of pixels of n in this image, n=0 wherein, 1 ... N-1.
In step (2), by checking two end points H[0 of gray shade scale histogram] and H[N-1] whether numerical value be higher than near the extrapolated value of the point homonymy, if, assert that then there is skew in histogram, need compensation, and the histogram after the compensation asked the binaryzation thresholding, specifically comprise following step:
1) get M the numerical value that border is deducted in gray shade scale histogram left side, H[i], i=1,2 ..., M infers the 0th numerical value H as the reference point of extrapolating
1
2) calculate left side penalty coefficient k
1=c*H[0]/H
1, wherein c is a compensating factor;
3) get M the numerical value that border is deducted on gray shade scale histogram right side, H[N-M+i-1 with method], i=1,2 ..., M infers the numerical value H that N-1 is ordered as the reference point of extrapolation
r
4) calculate right side penalty coefficient k
r=c*H[N-1]/H
r
5) the big Tianjin method that use to change is asked the binaryzation thresholding, and promptly the square of leftmost side point is by H[0] * 0 changes into H[0] * (0-k
1), and the square of rightmost side point is by H[N-1] * (N-1) changes into H[N-1] * (N-1+k
r), obtain binaryzation thresholding d.
Further, in the step 1) and step 3) in the step (2), use linear extrapolation or curve extrapolation method to calculate the estimated value of end points during extrapolation, perhaps use the estimated value of arithmetic mean as end points.In the step 5) in the step (2), use big Tianjin method, bimodal method, process of iteration when asking the binaryzation thresholding or based on the binarization method of square.
Further, have better effect for making the present invention, in step (2) and step (3), the method for asking binaryzation is overall binarization method, or local binarization method; This method can adopt single domain value binaryzation when binaryzation, also can adopt multiple domain value binaryzation.
Effect of the present invention is: adopt method of the present invention, can the better binary conversion result can be arranged to the image that histogram takes place to be offset, thereby the image subjective quality is improved, and can effectively improve the discrimination of OCR.
Principle of the present invention is: the peak of supposing histogram right side end points is because having more the pixel of high brightness be truncated to maximum gray shade scale causes, and the peak of histogram left side end points is to cause because there is the pixel of more low-light level to be truncated to the minimal gray grade.By being similar to the pixel on high brightness level more and peak, partition left side to low brightness levels more, the pixel on partition peak, right side reduces the histogram of the normal peak that meets statistical law, use the histogram after reducing to determine the binaryzation thresholding, thereby obtain more real binaryzation thresholding, finally use this binaryzation thresholding that original image is carried out binaryzation.
Description of drawings
Fig. 1 is the process flow diagram of the method for the invention;
Fig. 2 is pending image specimen page;
Fig. 3 is the statistic histogram for the image specimen page;
Fig. 4 is image and the recognition result after conventional big Tianjin method binaryzation;
Fig. 5 is to use image and the recognition result after the binaryzation of the inventive method;
Embodiment
The present invention is described in further detail below in conjunction with drawings and embodiments.
As shown in Figure 1, a kind of improved image binaryzation method may further comprise the steps:
(1) grey level histogram of statistical figure image
As Fig. 2, shown in Figure 3, earlier later to digitizing gray level image (Fig. 2) statistics grey level histogram (Fig. 3) can see that there is obvious peak in the histogram right side, and in the present embodiment, gray shade scale is 256 grades (N=256);
(2) calculate the binaryzation thresholding according to compensation method
Get M the numerical value that border is deducted in histogram left side, H[i], i=1,2 ..., M infers the 0th numerical value H as the reference point of extrapolating
1Get M the numerical value that border is deducted on the histogram right side, H[N-M+i-1], i=1,2 ..., M infers the numerical value H that N-1 is ordered as the reference point of extrapolation
rIn the present embodiment, M gets 5, and the arithmetic mean that the Extrapolation method use is 5 is as the estimated value of frontier point.Calculating the binaryzation thresholding by revised big Tianjin method is 254;
(3) image is carried out binaryzation
As Fig. 4, shown in Figure 5, in the present embodiment, carry out binaryzation by overall single domain value binarization method, the binary image that obtains is shown in Fig. 5 left side.As a comparison, the binaryzation thresholding of not using method of the present invention directly to use big Tianjin method to obtain is 185, and the binary image that obtains is shown in Fig. 4 left side;
(4) use OCR software to carry out literal identification to the binary image that obtains, the right side of Fig. 4 and Fig. 5 shows recognition result.
Can see that image binaryzation of the present invention has obviously improved the quality of image, the disconnected pen of literal obviously reduces, and the output result can effectively improve discrimination when using for OCR software.
Method of the present invention is not limited to the embodiment described in the embodiment, and those skilled in the art's technical scheme according to the present invention draws other embodiment, belongs to technological innovation scope of the present invention equally.
Claims (7)
1. the method for an image binaryzation may further comprise the steps:
(1) at first adds up the gray shade scale histogram of pending digital picture;
(2) check whether the gray shade scale histogram exists skew, and the gray shade scale histogram that is offset is carried out the histogram edge compensation, and the gray shade scale histogram after the compensation is asked the binaryzation thresholding;
(3) image is carried out binaryzation by thresholding.
2. the method for a kind of image binaryzation as claimed in claim 1, it is characterized in that: in step (1), if gray shade scale is N, the statistic histogram that obtains is H[N], each element H[n in the statistic histogram] represent that gray shade scale is the number of pixels of n in this image, n=0 wherein, 1, ..., N-1.
3. the method for a kind of image binaryzation as claimed in claim 1, it is characterized in that: in step (2), by checking two end points H[0 of gray shade scale histogram] and H[N-1] whether numerical value be higher than near the extrapolated value of the point homonymy, if, assert that then there is skew in the gray shade scale histogram, need compensation, and the gray shade scale histogram after the compensation asked the binaryzation thresholding, specifically comprise following step:
1) get M the numerical value that border is deducted in gray shade scale histogram left side, H[i], i=1,2 ..., M infers the 0th numerical value H as the reference point of extrapolating
1
2) calculate left side penalty coefficient k
1=c
*H[0]/H
1, wherein c is a compensating factor;
3) get M the numerical value that border is deducted on gray shade scale histogram right side, H[N-M+i-1 with method], i=1,2 ..., M infers the numerical value H that N-1 is ordered as the reference point of extrapolation
r
4) calculate right side penalty coefficient k
r=c
*H[N-1]/H
r
5) use the big Tianjin method that changes to ask the binaryzation thresholding, promptly the square of leftmost side point is by H[0]
*0 changes into H[0]
*(0-k
1), and the square of rightmost side point is by H[N-1]
*(N-1) change into H[N-1]
*(N-1+k
r), obtain binaryzation thresholding d.
4. the method for a kind of image binaryzation as claimed in claim 3, it is characterized in that: in the step 1) and step 3) in step (2), use linear extrapolation or curve extrapolation method to calculate the estimated value of end points during extrapolation, perhaps use the estimated value of arithmetic mean as end points.
5. the method for a kind of image binaryzation as claimed in claim 3 is characterized in that: in the step 5) in step (2), use big Tianjin method, bimodal method, process of iteration when asking the binaryzation thresholding or based on the binarization method of square.
6. the method for a kind of image binaryzation as claimed in claim 1, it is characterized in that: the method for asking binaryzation is overall binarization method, or local binarization method.
7. the method for a kind of image binaryzation as claimed in claim 1, it is characterized in that: this method can adopt single domain value binaryzation when binaryzation, also can adopt multiple domain value binaryzation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100800506A CN100377169C (en) | 2005-06-28 | 2005-06-28 | Method for picture binaryzation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100800506A CN100377169C (en) | 2005-06-28 | 2005-06-28 | Method for picture binaryzation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1694119A CN1694119A (en) | 2005-11-09 |
| CN100377169C true CN100377169C (en) | 2008-03-26 |
Family
ID=35353073
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100800506A Expired - Fee Related CN100377169C (en) | 2005-06-28 | 2005-06-28 | Method for picture binaryzation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100377169C (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100363942C (en) * | 2005-12-21 | 2008-01-23 | 北大方正集团有限公司 | Binary method and system for image |
| CN101127207B (en) * | 2007-09-26 | 2010-06-02 | 北大方正集团有限公司 | Method and device for promoting grey scale font display quality |
| CN102169580B (en) * | 2011-04-08 | 2012-10-10 | 中国船舶重工集团公司第七○二研究所 | Self-adaptive image processing method utilizing image statistic characteristics |
| CN102350057A (en) * | 2011-10-21 | 2012-02-15 | 上海魔迅信息科技有限公司 | System and method for realizing operation and control of somatic game based on television set top box |
| CN110288626B (en) * | 2019-06-13 | 2021-05-25 | 北京云测信息技术有限公司 | Method and apparatus for detecting text in native electronic images |
| CN110266729A (en) * | 2019-07-18 | 2019-09-20 | 倪玉根 | Cloud Server login method and device based on image encryption |
| CN115546241B (en) * | 2022-12-06 | 2023-04-28 | 成都数之联科技股份有限公司 | Edge detection method, edge detection device, electronic equipment and computer readable storage medium |
-
2005
- 2005-06-28 CN CNB2005100800506A patent/CN100377169C/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN1694119A (en) | 2005-11-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100377169C (en) | Method for picture binaryzation | |
| JP5701182B2 (en) | Image processing apparatus, image processing method, and computer program | |
| Bataineh et al. | An adaptive local binarization method for document images based on a novel thresholding method and dynamic windows | |
| CN101453575B (en) | Video subtitle information extracting method | |
| JP4590471B2 (en) | Method and system for estimating background color | |
| US8355566B2 (en) | Method and device for use in converting a colour image into a grayscale image | |
| CN105374015A (en) | Binary method for low-quality document image based on local contract and estimation of stroke width | |
| US20080181496A1 (en) | Methods and Systems for Detecting Character Content in a Digital Image | |
| ATE71787T1 (en) | METHOD OF CONVERTING AN ORIGINAL VIDEO IMAGE WITH MANY GRAY LEVELS INTO A BINARY IMAGE. | |
| CN106096610A (en) | A kind of file and picture binary coding method based on support vector machine | |
| CN110210440A (en) | A kind of form image printed page analysis method and system | |
| CN109377464A (en) | A kind of Double plateaus histogram equalization method and its application system of infrared image | |
| CN101751575B (en) | Image comparison system and method | |
| CN100479484C (en) | File and picture binary coding method | |
| CN110210467B (en) | Formula positioning method of text image, image processing device and storage medium | |
| US20080205753A1 (en) | Methods and Systems for Refining Text Detection in a Digital Image | |
| CN110826360A (en) | OCR image preprocessing and character recognition | |
| KR100955180B1 (en) | Method and System for Computing Quality Value of Image | |
| CN103839231A (en) | Image enhancement method based on maximization of human vision minimum detection probability | |
| CN111223050A (en) | Real-time image edge detection algorithm | |
| CN114267035A (en) | Document image processing method and system, electronic device and readable medium | |
| CN102799880A (en) | Device and method used for acquiring rolling fingerprint image | |
| Salagar et al. | Analysis of PCA usage to detect and correct skew in document images | |
| Boiangiu et al. | Bitonal image creation for automatic content conversion | |
| Wei | Research of neural networks in image anomaly identification application |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Assignee: Beijing Founder Broadband Network Technology Co.,Ltd. Assignor: PEKING UNIVERSITY FOUNDER GROUP Co.,Ltd.|PEKING University FOUNDER R & D CENTER Contract fulfillment period: 2007.12.26 to 2013.12.25 Contract record no.: 2008110000197 Denomination of invention: Method for picture binaryzation Granted publication date: 20080326 License type: Exclusive license Record date: 20081212 |
|
| LIC | Patent licence contract for exploitation submitted for record |
Free format text: EXCLUSIVE LICENSE; TIME LIMIT OF IMPLEMENTING CONTACT: 2007.12.26 TO 2013.12.25; CHANGE OF CONTRACT Name of requester: BEIJING BEIDAFANGZHENG BROADBAND NETWORKS SCIENCE Effective date: 20081212 |
|
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: Beijing Founder Broadband Network Technology Co.,Ltd. Assignor: PEKING UNIVERSITY FOUNDER GROUP Co.,Ltd.|PEKING University FOUNDER R & D CENTER Contract record no.: 2008110000197 Date of cancellation: 20111209 |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220919 Address after: 3007, Hengqin international financial center building, No. 58, Huajin street, Hengqin new area, Zhuhai, Guangdong 519031 Patentee after: New founder holdings development Co.,Ltd. Patentee after: PEKING University FOUNDER R & D CENTER Address before: 100871 Beijing City, Haidian District Road No. 298 building, founder of Ayutthaya Patentee before: PEKING UNIVERSITY FOUNDER GROUP Co.,Ltd. Patentee before: PEKING University FOUNDER R & D CENTER |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080326 |