CN104766043A - Method of fast identification of ballot image - Google Patents
Method of fast identification of ballot image Download PDFInfo
- Publication number
- CN104766043A CN104766043A CN201410545341.7A CN201410545341A CN104766043A CN 104766043 A CN104766043 A CN 104766043A CN 201410545341 A CN201410545341 A CN 201410545341A CN 104766043 A CN104766043 A CN 104766043A
- Authority
- CN
- China
- Prior art keywords
- image
- ballot paper
- ballot
- paper
- rectangular code
- 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
Landscapes
- Character Input (AREA)
Abstract
The invention discloses a method of fast identification of a ballot image. A ballot image is acquired by a high-speed scanner based on related position data recorded in ballot design. After the ballot image is preprocessed, a determined reference point is found, the position data in ballot design is extracted and converted into an image position relative to the reference point, and form lines are searched up and down and left and right based on the position to determine the accurate position of the image to be identified. A handwritten symbol image is intercepted based on the position. After a series of processing is performed on the image, the structure features of the image are extracted so as to realize symbol identification. By adopting the method of the invention, the speed of image positioning is increased, and the identification rate for handwritten symbol images is greatly improved.
Description
Technical field
The present invention relates to field of image recognition, particularly relate to a kind of ballot paper image and know method for distinguishing fast.
Background technology
In present democratic society, the decision-making of a lot of problem all needs to use election, traditional electoral machinery is, each participant chooses corresponding position on papery ballot paper, and then by manually carrying out calling out the names of those voted for while counting ballot-slips and count of votes, such way to elect can cause the problem wasted time and energy, the development of science and technology at any time, the development of OCR (Optical Character Recognition, optical character identification) technology, utilizing OCR technology to develop poll system becomes focus.Due to OCR technology, make system can fully demonstrate the justice, just and open of ballot, have fast, accurately, the feature such as strong security.System adopts papery ballot paper can control error rate preferably, and also can process well soft image, but the existing method based on OCR is higher for the requirement of papery, to the requirement of filling in people also more, recognition speed and recognition accuracy not high, greatly limit the development that it is follow-up.
Summary of the invention
In order to solve the problem, on the basis fully combining ballot paper format and symbolic construction feature, the present invention proposes the method for quickly identifying of hand-written symbol in ballot paper image.The relative position data of record during the method utilizes ballot paper to design, after ballot paper Image semantic classification, find a reference point determined, extract the position data in ballot paper design again, it is converted into the picture position relative to this reference point, utilize this location finding form line, thus determine the accurate location wanting recognition image; Based on this position, intercept hand-written symbol image, after a series of process is carried out to this image, extract its architectural feature, thus realize Symbol recognition.The locating speed that not only improve image of method of the present invention, and substantially increase the discrimination of hand-written symbol image.
The present invention adopts following technical scheme:
A kind of ballot paper image knows method for distinguishing fast, comprises the following steps:
S1: utilize the ballot paper design module of host side to carry out ballot paper format design, comprising: paper size, margin, table position, boxhead are high, gauge outfit columns, fill in hurdle column number, fill in that hurdle is wide, candidate hurdle is wide, the columns of each row in capable high, the table body of table body line number, table body; Described ballot paper format design is provided with symmetrical ballot paper sequence number rectangular code and ballot paper page number rectangular code in the fixed position, base of ballot paper page, the frame line of two rectangular codes is thick lines, and each rectangular code is made up of the little square of 6 same sizes;
S2: after ballot paper has designed, sort by election item, candidate's form, alternative people form in the area to be targeted of every page of ballot paper, and LSN Index; With paper bottom left vertex for initial point, calculate the paper position P from a nearest form bottom left vertex of initial point; Calculate the position of center in paper of each area to be targeted, find out from the nearest data d0 of P, then look for the next data d1 nearest from d0, is sorted successively in the center of areas to be targeted all in ballot paper, the sequence number Index that record area to be targeted is corresponding, obtains data D; All data relevant with ballot paper are all recorded hereof;
S3: ballot paper formatted data and the data D relevant with ballot paper are sent to ballot box end by host side;
S4: ballot box end utilizes high speed scanner to scan ballot paper to be identified, obtains the scan image Pic0 of ballot paper;
S5: the scan image Pic0 of ballot box end to ballot paper carries out pre-service; Described pre-service comprises Pic0 is converted into 24 colored DIB Pic1, carries out image gray processing, binaryzation, goes discrete noise and be set effective ballot paper image periphery for white to Pic1;
S6: carry out horizontal projection to through pretreated scan image Pic1, search for from bottom to up, find the position on rectangular code bottom margin image base, then refinement is carried out to ballot paper image, obtain image Pic2; Carry out bottom sectional drawing to image Pic2, sectional drawing image comprises rectangular code, carries out from top to bottom by line scanning to sectional drawing image, removes any word except rectangular code or picture; And calculate the slope of sectional drawing, utilize slope to carry out slant correction to sectional drawing image;
S7: carry out horizontal projection and vertical projection to sectional drawing image at the bottom of corrected ballot paper page, determines the position of ballot paper sequence number rectangular code left upper apex and ballot paper page number rectangular code right vertices; Utilize two vertex positions obtained to navigate to each little square, and judge there is nil in each little square, utilize the weights distribution principle of rectangular code, determine the sequence number of ballot paper kind and the page number;
S8: the sequence number utilizing ballot paper kind and the page number thereof obtained in S7, carries out slant correction according to ballot paper formatted data, the data D relevant with ballot paper and slope by whole ballot paper image; The position of ballot paper sequence number rectangular code left upper apex is converted into its position at whole ballot paper image, namely determines the position of reference point; Utilize the position of reference point in ballot paper image, ballot paper kind and page number sequence number, calculate approximate location that is to be identified or screenshot area, utilize this location finding form line thus determine accurate location that is to be identified or screenshot area;
S9: utilize the to be identified or screenshot area position data obtained in S8, symbol area to be identified is intercepted in direct scan image after calibration, a left side, top, the right side, first black pixel point at the end in searching image, determine the position of hand-written symbol, intercept hand-written symbol image, it be normalized, image expansion process and thinning processing;
S10: extract the hand-written feature meeting image;
S11: judge it is which kind of hand-written symbol according to the characteristic use decision-making technique of image.
Preferably, in described step S1, the position of ballot paper sequence number rectangular code and ballot paper page number rectangular code centered beneath is also provided with the page number; Ballot paper sequence number rectangular code is on the left side of the page number rectangular code of ballot paper.
Preferably, described step S6 medium dip rate is the ratio of the true altitude of rectangular code in average height half of about rectangular code in sectional drawing image and ballot paper page formatted data.
Preferably, the normalized in described step S9 refers to: image is changed into 32 × 32 pixel images; That image expansion adopts is self-defined structure structure [3] [3]={ 0,1,0,1,1,1,0,1,0};
Preferably, extract the hand-written feature meeting image in described step S10 to refer to: in horizontal direction, line by line scan from top to bottom, once there be continuous 8 provisional capitals to have black pixel point, just illustrate it is hand-written symbol, the situation that record is often gone: single-point represents with-1, a difference=rear black pixel point column-previous black pixel point column-1, the difference of two continuous black pixel points is 0,-1 and difference are recorded in array successively, and this array is exactly the architectural feature of hand-written symbol in horizontal direction; In like manner obtain the architectural feature of this hand-written symbol in vertical direction.
Preferably, the decision-making technique described in described step S11 refers to the difference and single-point information that utilize horizontal and vertical direction, first determines whether thick stick, then determines whether fork; When being judged as thick stick, the feature of thick stick is directly utilized to judge; When being judged as fork, if there is difference 0, will be removed it to calculate the mean difference of top, middle part, bottom, and meet the feature of fork, and just can be judged as fork; Otherwise, then determine whether to hook or circle.
Beneficial effect of the present invention:
The relative position data of record during the present invention utilizes ballot paper to design, after ballot paper Image semantic classification, find a reference point determined, extract the position data in ballot paper design again, it is converted into the picture position relative to this reference point, utilize this location finding form line, thus determine the accurate location wanting recognition image; Based on this position, intercept hand-written symbol image, after a series of process is carried out to this image, extract its architectural feature, thus realize Symbol recognition.Method of the present invention not only improves the locating speed of image, and substantially increases the discrimination of hand-written image.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of an embodiment of the present invention;
Fig. 2 is sectional drawing schematic diagram at the bottom of the page of an embodiment of the present invention;
Fig. 3 carries out the rear schematic diagram of removal interference to Fig. 2;
The hand-written preprocessing process schematic diagram meeting image of Fig. 4;
Fig. 5 is difference and single-point schematic diagram;
Fig. 6 is that in Fig. 5,4 figure characteristics of correspondence extract schematic diagram;
Fig. 7 hooks and circle three partial difference schematic diagram;
Fig. 8 is the single-point of fork and upper and lower two three partial difference schematic diagram.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is described in detail.
As shown in Figure 1, a kind of ballot paper image knows method for distinguishing fast, comprises the following steps:
S1: utilize the ballot paper design module of host side to carry out ballot paper format design, comprising: paper size, margin, table position, boxhead are high, gauge outfit columns, fill in hurdle column number, fill in that hurdle is wide, candidate hurdle is wide, the columns of each row in capable high, the table body of table body line number, table body; Described ballot paper format design is provided with symmetrical ballot paper sequence number rectangular code and ballot paper page number rectangular code in the fixed position, base of ballot paper page, the frame line of two rectangular codes is thick lines, and each rectangular code is made up of the little square of 6 same sizes;
S2: after ballot paper has designed, sort by election item, candidate's form, alternative people form in the area to be targeted of every page of ballot paper, and LSN Index; With paper bottom left vertex for initial point, calculate the paper position P from a nearest form bottom left vertex of initial point; Calculate the position of center in paper of each area to be targeted, find out from the nearest data d0 of P, then look for the next data d1 nearest from d0, is sorted successively in the center of areas to be targeted all in ballot paper, the sequence number Index that record area to be targeted is corresponding, obtains data D; All data relevant with ballot paper are all recorded hereof;
S3: ballot paper formatted data and the data D relevant with ballot paper are sent to ballot box end by host side;
S4: ballot box end utilizes high speed scanner to scan ballot paper to be identified, obtains the scan image Pic0 of ballot paper;
S5: the scan image Pic0 of ballot box end to ballot paper carries out pre-service; Described pre-service comprises Pic0 is converted into 24 colored DIB Pic1, carries out image gray processing, binaryzation, goes discrete noise and be set effective ballot paper image periphery for white to Pic1;
S6: carry out horizontal projection to through pretreated scan image Pic1, search for from bottom to up, find the position on rectangular code bottom margin image base, then refinement is carried out to ballot paper image, obtain image Pic2; Carry out bottom sectional drawing to image Pic2, sectional drawing image comprises rectangular code, carries out from top to bottom by line scanning to sectional drawing image, removes any word except rectangular code or picture; And calculate the slope of sectional drawing, utilize slope to carry out slant correction to sectional drawing image;
S7: carry out horizontal projection and vertical projection to sectional drawing image at the bottom of corrected ballot paper page, determines the position of ballot paper sequence number rectangular code left upper apex and ballot paper page number rectangular code right vertices; Utilize two vertex positions obtained to navigate to each little square, and judge there is nil in each little square, utilize the weights distribution principle of rectangular code, determine the sequence number of ballot paper kind and the page number;
S8: the sequence number utilizing ballot paper kind and the page number thereof obtained in S7, carries out slant correction according to ballot paper formatted data, the data D relevant with ballot paper and slope by whole ballot paper image; The position of ballot paper sequence number rectangular code left upper apex is converted into its position at whole ballot paper image, namely determines the position of reference point; Utilize the position of reference point in ballot paper image, ballot paper kind and page number sequence number, calculate approximate location that is to be identified or screenshot area, utilize this location finding form line thus determine accurate location that is to be identified or screenshot area;
S9: utilize the to be identified or screenshot area position data obtained in S8, symbol area to be identified is intercepted in direct scan image after calibration, a left side, top, the right side, first black pixel point at the end in searching image, determine the position of hand-written symbol, intercept hand-written symbol image, it be normalized, image expansion process and thinning processing;
S10: extract the hand-written feature meeting image;
S11: judge it is which kind of hand-written symbol according to the characteristic use decision-making technique of image.
Fig. 2 and Fig. 3 correspond respectively in step S6 bottom sectional drawing is carried out to image Pic2 after and remove any word except rectangular code or picture.
In the present embodiment, described rectangular code refers to the patent No.: CN1437157, and patent name is a kind of rectangular code coding method and based on rectangular code said in the rectangular code of the method.The length of side of rectangular code is 5mm.
In the present embodiment, the position of ballot paper sequence number rectangular code and ballot paper page number rectangular code centered beneath is also provided with the page number (this page number only occurs in multipage ballot paper, and single page ballot paper is this page number not); Ballot paper sequence number rectangular code is on the left side of the page number rectangular code of ballot paper.
Normalized in the present embodiment in step S9 refers to: image is changed into 32 × 32 pixel images; That image expansion adopts is self-defined structure structure [3] [3]={ 0,1,0,1,1,1,0,1,0}; As shown in Figure 4.
In the present embodiment, before carrying out image characteristics extraction to image, need to do following pre-service:
In the horizontal direction: from top to bottom, carry out line scanning, record often row black pixel point and column thereof, this has 3 kinds of situations.
1) only have 1 black pixel point, do not deal with.
2) there are 2 black pixel points, then judge whether these 2 black pixel points are continuous print.If so, a rear black pixel point (it is white for putting this pixel) is removed; Otherwise, do not deal with.
3) have more than 2 black pixel points, will judge whether these points are continuous print.If so, just retain 2 continuous print black pixel points of foremost, remove remaining; Otherwise, retain foremost and rearmost black pixel point, in the middle of removing.
In vertical direction: from left to right, carry out column scan, record often row black pixel point and being expert at.1) with 2) both of these case is identical with the process in horizontal direction.3) this situation, if continuous print, just retains first black pixel point of foremost, removes remaining; Otherwise, retain foremost and rearmost black pixel point, in the middle of removing.That is, vertical direction only have single-point and two points separated.
In horizontal direction and vertical direction, if there are two continuous black pixel points the centre of black pixel point, just remove latter one, retain previous.Last treatment effect as shown in Figure 6.
After obtaining such image, just the architectural feature of symbol can be extracted.Method is: in horizontal direction, lines by line scan from top to bottom, once there be continuous 8 provisional capitals to have black pixel point, just illustrates that this has been hand-written symbol.The situation that record is often gone: single-point represents with-1, a difference=rear black pixel point column-previous black pixel point column-1.Like this, the difference of two continuous black pixel points is 0.-1 and difference are recorded in array successively, and this array is exactly the architectural feature of hand-written symbol in horizontal direction.According to same method, the architectural feature of this hand-written symbol in vertical direction also can be obtained.
The number being more than or equal to 0 is used in horizontal direction architectural feature array to calculate mean difference.But after image procossing, likely there is following situation, as shown in Figure 9.That is, difference 0 is interference.Decision process is for this reason: first determine whether thick stick, then determines whether fork.When being judged as thick stick, the feature of thick stick is directly utilized to judge.When being judged as fork, if there is difference 0, will be removed it to calculate the mean difference of top, middle part, bottom, and meet the feature of fork, and just can be judged as fork; Otherwise, then determine whether to hook or circle.The introducing of this difference 0 calculates, and probably hook is judged as circle, for this reason, will uses the feature of vertical direction symbol.Obviously, in vertical direction, the single-point sum of hook is many, and difference sum is few, encloses then contrary.Therefore, after the mean difference calculating the top of horizontal direction, middle part, bottom, also to carry out comparing of single-point sum and difference sum in vertical direction, just can distinguish hook and enclose.Not meeting above-mentioned condition, is other symbol.
Use the computing machine of CPU frequency 2.0GHz, internal memory 2GB, choose a kind of ballot paper and test.This ballot paper is single page ballot paper, and it has, and the candidate of 2 election items, election item 1 is not sub-category, election item 2 contains 2 candidate's classifications, and have 3 candidate's forms, 43 form grid need fixation and recognition, and 23 grid fill in symbol.This ballot paper A4 paper prints, ballot paper image to be resolution be 1200 × 1600 scan image, 43 complete needs of grid fixation and recognition 2.250 seconds.If (single page, elects item, 2 candidate's classifications, 2 candidate's forms for 1,2 alternative people forms, the grid of 59 need location, and wherein 6 grid select other people else, and 3 grid fill in alternative people name to choose another kind of ballot paper again; 53 grid need distinguished symbol, and 40 grid fill in symbol.), then need 2.437 seconds.Fixation and recognition speed is still than faster.
To those skilled in the art, obviously the invention is not restricted to the details of above-mentioned one exemplary embodiment, and when not deviating from spirit of the present invention or essential characteristic, the present invention can be realized in other specific forms.Therefore, no matter from which point, all should embodiment be regarded as exemplary, and be nonrestrictive, scope of the present invention is limited by claims instead of above-mentioned explanation, and all changes be therefore intended in the implication of the equivalency by dropping on claim and scope are included in the present invention.Any Reference numeral in claim should be considered as the claim involved by limiting.
In addition, be to be understood that, although this instructions is described according to embodiment, but not each embodiment only comprises an independently technical scheme, this narrating mode of instructions is only for clarity sake, those skilled in the art should by instructions integrally, and the technical scheme in each embodiment also through appropriately combined, can form other embodiments that it will be appreciated by those skilled in the art that.
Claims (6)
1. ballot paper image knows a method for distinguishing fast, it is characterized in that: comprise the following steps:
S1: utilize the ballot paper design module of host side to carry out ballot paper format design, comprising: paper size, margin, table position, boxhead are high, gauge outfit columns, fill in hurdle column number, fill in that hurdle is wide, candidate hurdle is wide, the columns of each row in capable high, the table body of table body line number, table body; Described ballot paper format design is provided with symmetrical ballot paper sequence number rectangular code and ballot paper page number rectangular code in the fixed position, base of ballot paper page, the frame line of two rectangular codes is thick lines, and each rectangular code is made up of the little square of 6 same sizes;
S2: after ballot paper has designed, sort by election item, candidate's form, alternative people form in the area to be targeted of every page of ballot paper, and LSN Index; With paper bottom left vertex for initial point, calculate the paper position P from a nearest form bottom left vertex of initial point; Calculate the position of center in paper of each area to be targeted, find out from the nearest data d0 of P, then look for the next data d1 nearest from d0, is sorted successively in the center of areas to be targeted all in ballot paper, the sequence number Index that record area to be targeted is corresponding, obtains data D; All data relevant with ballot paper are all recorded hereof;
S3: ballot paper formatted data and the data D relevant with ballot paper are sent to ballot box end by host side;
S4: ballot box end utilizes high speed scanner to scan ballot paper to be identified, obtains the scan image Pic0 of ballot paper;
S5: the scan image Pic0 of ballot box end to ballot paper carries out pre-service; Described pre-service comprises Pic0 is converted into 24 colored DIB Pic1, carries out image gray processing, binaryzation, goes discrete noise and be set effective ballot paper image periphery for white to Pic1;
S6: carry out horizontal projection to through pretreated scan image Pic1, search for from bottom to up, find the position on rectangular code bottom margin image base, then refinement is carried out to ballot paper image, obtain image Pic2; Carry out bottom sectional drawing to image Pic2, sectional drawing image comprises rectangular code, carries out from top to bottom by line scanning to sectional drawing image, removes any word except rectangular code or picture; And calculate the slope of sectional drawing, utilize slope to carry out slant correction to sectional drawing image;
S7: carry out horizontal projection and vertical projection to sectional drawing image at the bottom of corrected ballot paper page, determines the position of ballot paper sequence number rectangular code left upper apex and ballot paper page number rectangular code right vertices; Utilize two vertex positions obtained to navigate to each little square, and judge there is nil in each little square, utilize the weights distribution principle of rectangular code, determine the sequence number of ballot paper kind and the page number;
S8: the sequence number utilizing ballot paper kind and the page number thereof obtained in S7, carries out slant correction according to ballot paper formatted data, the data D relevant with ballot paper and slope by whole ballot paper image; The position of ballot paper sequence number rectangular code left upper apex is converted into its position at whole ballot paper image, namely determines the position of reference point; Utilize the position of reference point in ballot paper image, ballot paper kind and page number sequence number, calculate approximate location that is to be identified or screenshot area, utilize this location finding form line thus determine accurate location that is to be identified or screenshot area;
S9: utilize the to be identified or screenshot area position data obtained in S8, symbol area to be identified is intercepted in direct scan image after calibration, a left side, top, the right side, first black pixel point at the end in searching image, determine the position of hand-written symbol, intercept hand-written symbol image, it be normalized, image expansion process and thinning processing;
S10: the feature extracting hand-written symbol image;
S11: judge it is which kind of hand-written symbol according to the characteristic use decision-making technique of image.
2. a kind of ballot paper image according to claim 1 knows method for distinguishing fast, it is characterized in that: in described step S1, the position of ballot paper sequence number rectangular code and ballot paper page number rectangular code centered beneath is also provided with the page number; Ballot paper sequence number rectangular code is on the left side of the page number rectangular code of ballot paper.
3. a kind of ballot paper image according to claim 1 knows method for distinguishing fast, it is characterized in that: described step S6 medium dip rate is the ratio of the true altitude of rectangular code in average height half of about rectangular code in sectional drawing image and ballot paper page formatted data.
4. a kind of ballot paper image according to claim 1 knows method for distinguishing fast, it is characterized in that: the normalized in described step S9 refers to: image is changed into 32 × 32 pixel images; That image expansion adopts is self-defined structure structure [3] [3]={ 0,1,0,1,1,1,0,1,0};
5. a kind of ballot paper image according to claim 1 knows method for distinguishing fast, it is characterized in that: the feature extracting hand-written symbol image in described step S10 refers to: in horizontal direction, line by line scan from top to bottom, once there be continuous 8 provisional capitals to have black pixel point, just illustrate it is hand-written symbol, the situation that record is often gone: single-point represents with-1, a difference=rear black pixel point column-previous black pixel point column-1, the difference of two continuous black pixel points is 0,-1 and difference are recorded in array successively, this array is exactly the architectural feature of hand-written symbol in horizontal direction, in like manner obtain the architectural feature of this hand-written symbol in vertical direction.
6. a kind of ballot paper image according to claim 1 knows method for distinguishing fast, it is characterized in that: the decision-making technique described in described step S11 refers to the difference and single-point information that utilize horizontal and vertical direction, first determines whether thick stick, then determines whether fork; When being judged as thick stick, the feature of thick stick is directly utilized to judge; When being judged as fork, if there is difference 0, will be removed it to calculate the mean difference of top, middle part, bottom, and meet the feature of fork, and just can be judged as fork; Otherwise, then determine whether to hook or circle.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410545341.7A CN104766043A (en) | 2014-10-15 | 2014-10-15 | Method of fast identification of ballot image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410545341.7A CN104766043A (en) | 2014-10-15 | 2014-10-15 | Method of fast identification of ballot image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104766043A true CN104766043A (en) | 2015-07-08 |
Family
ID=53647859
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410545341.7A Pending CN104766043A (en) | 2014-10-15 | 2014-10-15 | Method of fast identification of ballot image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104766043A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194391A (en) * | 2017-04-21 | 2017-09-22 | 广东清华文通科技有限公司 | A kind of method for the intellectual analysis positioning electronic ballot paper for resisting image stretch skew |
| CN108475335A (en) * | 2016-01-27 | 2018-08-31 | 霍尼韦尔国际公司 | The Method and kit for of the postmortem analysis of tripping field device in process industrial for using optical character identification ＆ intelligent character recognitions |
| CN110308346A (en) * | 2019-06-24 | 2019-10-08 | 中国航空无线电电子研究所 | Cockpit display system automatic test approach and system based on image recognition |
| CN110532938A (en) * | 2019-08-27 | 2019-12-03 | 海南阿凡题科技有限公司 | Papery operation page number recognition methods based on Faster-RCNN |
| CN112801016A (en) * | 2021-02-08 | 2021-05-14 | 联仁健康医疗大数据科技股份有限公司 | Vote data statistical method, device, equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1804905A (en) * | 2006-01-12 | 2006-07-19 | 北京飞利信电子技术有限公司 | Voting paper recognition method for electronic voting boxes |
| CN101436321A (en) * | 2008-12-12 | 2009-05-20 | 中科院成都信息技术有限公司 | Method for processing ballot information |
| CN101447017A (en) * | 2008-11-27 | 2009-06-03 | 浙江工业大学 | Method and system for quickly identifying and counting votes on the basis of layout analysis |
| CN103761518A (en) * | 2014-01-10 | 2014-04-30 | 安徽大学 | Method for processing vote information |
-
2014
- 2014-10-15 CN CN201410545341.7A patent/CN104766043A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1804905A (en) * | 2006-01-12 | 2006-07-19 | 北京飞利信电子技术有限公司 | Voting paper recognition method for electronic voting boxes |
| CN101447017A (en) * | 2008-11-27 | 2009-06-03 | 浙江工业大学 | Method and system for quickly identifying and counting votes on the basis of layout analysis |
| CN101436321A (en) * | 2008-12-12 | 2009-05-20 | 中科院成都信息技术有限公司 | Method for processing ballot information |
| CN103761518A (en) * | 2014-01-10 | 2014-04-30 | 安徽大学 | Method for processing vote information |
Non-Patent Citations (1)
| Title |
|---|
| 胡俐蕊等: ""选票图像识别的快速方法"", 《计算机工程与设计》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108475335A (en) * | 2016-01-27 | 2018-08-31 | 霍尼韦尔国际公司 | The Method and kit for of the postmortem analysis of tripping field device in process industrial for using optical character identification ＆ intelligent character recognitions |
| CN108475335B (en) * | 2016-01-27 | 2022-10-14 | 霍尼韦尔国际公司 | Method for post-inspection analysis of tripped field devices in process industry using optical character recognition, smart character recognition |
| CN107194391A (en) * | 2017-04-21 | 2017-09-22 | 广东清华文通科技有限公司 | A kind of method for the intellectual analysis positioning electronic ballot paper for resisting image stretch skew |
| CN107194391B (en) * | 2017-04-21 | 2022-11-08 | 广东清华文通科技有限公司 | Method for intelligently analyzing and positioning electronic ballot capable of resisting image stretching deviation |
| CN110308346A (en) * | 2019-06-24 | 2019-10-08 | 中国航空无线电电子研究所 | Cockpit display system automatic test approach and system based on image recognition |
| CN110308346B (en) * | 2019-06-24 | 2021-10-26 | 中国航空无线电电子研究所 | Automatic testing method and system for cockpit display system based on image recognition |
| CN110532938A (en) * | 2019-08-27 | 2019-12-03 | 海南阿凡题科技有限公司 | Papery operation page number recognition methods based on Faster-RCNN |
| CN112801016A (en) * | 2021-02-08 | 2021-05-14 | 联仁健康医疗大数据科技股份有限公司 | Vote data statistical method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106156761B (en) | Image table detection and identification method for mobile terminal shooting | |
| CN107679078B (en) | Bayonet image vehicle rapid retrieval method and system based on deep learning | |
| CN108197538B (en) | Bayonet vehicle retrieval system and method based on local features and deep learning | |
| US9396404B2 (en) | Robust industrial optical character recognition | |
| CN102360419B (en) | Method and system for computer scanning reading management | |
| CN104392205B (en) | A kind of recognition methods and system of improper license plate | |
| Berchmans et al. | Optical character recognition: an overview and an insight | |
| US9324001B2 (en) | Character recognition device and character segmentation method | |
| CN103870803A (en) | Vehicle license plate recognition method and system based on coarse positioning and fine positioning fusion | |
| JP2012008791A (en) | Form recognition device and form recognition method | |
| CN106875546A (en) | A kind of recognition methods of VAT invoice | |
| JP5176763B2 (en) | Low quality character identification method and apparatus | |
| CN106056751A (en) | Prefix number identification method and system | |
| CN105574063A (en) | Image retrieval method based on visual saliency | |
| CN105809205B (en) | A kind of classification method and its system of high spectrum image | |
| CN1343339A (en) | Video stream classifiable symbol isolation method and system | |
| CN102388392A (en) | Pattern recognition device | |
| CN107145487A (en) | Image search method and device | |
| CN104700062A (en) | Method and equipment for identifying two-dimension code | |
| CN105139011B (en) | A kind of vehicle identification method and device based on mark object image | |
| CN103914680A (en) | Character image jet-printing, recognition and calibration system and method | |
| CN103136527B (en) | Image processing equipment and method | |
| CN106874901B (en) | Driving license identification method and device | |
| CN104766043A (en) | Method of fast identification of ballot image | |
| Chen et al. | Robust text line segmentation for historical manuscript images using color and texture |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate 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: 20150708 |