CN100428270C - Line detecting method and device - Google Patents
Line detecting method and device Download PDFInfo
- Publication number
- CN100428270C CN100428270C CNB2005101287847A CN200510128784A CN100428270C CN 100428270 C CN100428270 C CN 100428270C CN B2005101287847 A CNB2005101287847 A CN B2005101287847A CN 200510128784 A CN200510128784 A CN 200510128784A CN 100428270 C CN100428270 C CN 100428270C
- Authority
- CN
- China
- Prior art keywords
- line
- pixel
- gradient
- grad
- candidate
- 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
Images
Abstract
The invention provides a novel line detecting method and device more effective and accurate than routine methods, such as Hough Transform Method. And the invention is based on edge direction classification, and concretely, edges are divided into horizontal and vertical edges, thus causing real lines to form long connected elements but texts or other targets to form some separate elements in each directional edge image. The method comprises the steps of: calculating directional edge image to calculate horizontal and vertical edge images based on input image; candidate line determining to determine candidate lines based on each calculated horizontal edge image and each calculated vertical edge image; and line determining to determine line position based on the candidate lines.
Description
Technical field
The present invention relates to the program of line detecting method and equipment, detection line and write down the medium of this program.
Background technology
It is the basic problem of Flame Image Process that line detects, and more generally, relates to the curve detection problem.Existing multiple application need line detects.Such as:
In pre-scan images, line may be detected to determine the edge of paper.
Can provide with reference to lookup table, important numeral and signature at finance data Flame Image Process midline position.
Be used for the OCR of digital camera images (optical character identification), the image border line detects and can be used for detection position and pitch angle to be used for image rectification.
Detect people for line and made many effort.As the Hough Transform that describes in US 6292583 and US6111993 may be most important and the most widely used line detecting method.But known to everyone, this method is very infirm for complex image.In addition, if expectation is obtained the exact position of line then it has bigger calculated amount.And digital picture is all very big usually, and has complicated content.
In addition, US 6654495 is by use stroke analyzing and processing detection line, but this use is vertical and the stroke analysis of horizontal throw only may detection of vertical and horizontal line.
Consider the above-mentioned defective of prior art, require to find other more effective and simple method.
Summary of the invention
An object of the present invention is to provide a kind of than conventional method such as the Hough transform method method and apparatus of the novelty of detection line more effectively and accurately.This method and equipment are based on the classification of edge direction, and particularly, the edge is divided into level or vertical edge, have caused thus containing in the direction outline map at each, and solid line has formed long connection element, and text or other object have formed the element of some separation.
According to an aspect of the present invention, a kind of line detecting method comprises:
Contain direction outline map calculation procedure, be used for based on input picture calculated level outline map and vertical edge figure;
Candidate's line determining step, the edge pixel that is used for horizontal edge figure that calculates based on each and the vertical edge figure that is calculated is determined candidate's line; And
The line determining step is used for determining based on described candidate's line the position of line.
The present invention is categorized as two with outline map and contains the direction outline map, and promptly horizontal edge figure and vertical edge figure can detect than the simpler line that carries out exactly of classic method thus.Particularly,
(1) edge pixel of the same line only can appear at a kind of containing in the direction outline map, so the present invention merges candidate's line that this contains in the direction outline map, and need not to influence candidate's line of other direction.
(2) second, marginal classification is highly beneficial to identification candidate line.For many cross spiders form line for example, horizontal line and perpendicular line are intersected with each other in outline map.Which connection element is the connection elemental analysis method can not discern is the connection element at solid line edge.Yet the present invention is separated to perpendicular line and horizontal line in the different outline maps after marginal classification, so perpendicular line and horizontal line can be not intersected with each other, and contains easier identification line in the direction outline map at each, therefore can be identified exactly by the present invention.
(3) it is very favourable that other marginal classification is combined candidate's line equally, and this is that the region of search will tail off in the direction outline map because of containing at each, and the quantity of candidate's line also tails off.
Preferably, describedly contain the gradient map that direction outline map calculation procedure comprises the steps: to calculate described input picture; With based on gradient map calculated level outline map that is calculated and vertical edge figure.
Preferably, use the Sobel operator to calculate the gradient map of described input picture.
Preferably, the described direction outline map calculation procedure that contains comprises the steps:
Use following equation calculated level gradient and VG (vertical gradient):
gradX(i，j)＝?grad(i，j+1)-grad(i，j-1)
gradY(i，j)＝grad(i+1，j)-grad(i-1，j)
Here gradX (i, j) feeling the pulse with the finger-tip mark pixel (i, horizontal gradient j), gradY (i, i) feeling the pulse with the finger-tip mark pixel (i, VG (vertical gradient) j), (i j+1) refers to pixel (i, gradient j+1) to grad, (i j-1) refers to pixel (i, gradient j-1) to grad, (i+1 j) refers to pixel (i+1, gradient i) to grad, and grad (i-1 j) refers to pixel (i-1, gradient j);
Use following expression calculated level outline map and vertical edge figure:
Here, EdgeX (i, j) mean horizontal edge figure object pixel (i, j), EdgeY (i, j) mean vertical edge figure object pixel (i, j).
Preferably, wherein said candidate's line determining step comprises the steps: by connecting each horizontal edge figure that calculates and the edge pixel scounting line segment among the vertical edge figure; The line segment that merging searches is as candidate's line.
Preferably, described candidate's line determining step comprises the line segment of its length of removing less than the intended pixel scope.Preferably, this intended pixel scope is a 6-10 pixel.
Preferably, described combining step merges the line segment based on relation of the position between the line segment and pitch angle thereof.
Preferably, if adjacent line segment has less than the gap of intended pixel scope and has identical pitch angle, then merge adjacent line segment.Preferably, the intended pixel scope is a 4-8 pixel.
Containing the present invention of direction outline map for each has also introduced according to the search of position and pitch angle and has merged the processing of the line fragment that belongs to same line.The difficult point that this processing can be eliminated line breakpoint, inclination, bending and intersect.
Particularly, the edge pixel Connection Step comprises based on relation of the position between candidate's line and pitch angle merging line segment thereof.If adjacent line segment has less than the gap of intended pixel scope and has identical pitch angle, then merge adjacent line segment, the intended pixel scope is a 4-8 pixel.
Use these measures position line very accurately easily, be particularly suitable for line disconnection, inclination, little bending and intersection.
On the other hand, the invention provides a kind of line checkout equipment, comprising:
Contain direction outline map calculation element, be used for based on input picture calculated level outline map and vertical edge figure;
Candidate's line is determined device, and the edge pixel that is used for horizontal edge figure that calculates based on each and the vertical edge figure that is calculated is determined candidate's line; And
Line is determined device, is used for the position that described candidate's line is determined line.
Preferably, the described direction outline map calculation element that contains is configured to: the gradient map of calculating described input picture; With based on gradient map calculated level outline map that is calculated and vertical edge figure.
Preferably, use the Sobel operator to calculate the gradient map of described input picture.
Preferably, the described direction outline map calculation element that contains is configured to:
Use following equation calculated level gradient and VG (vertical gradient):
GradX (i, j)=grad (i, j+1)-grad (i, j-1) and
gradY(i，j)＝grad(i+1，j)-grad(i-1，j)
Here gradX (i, j) feeling the pulse with the finger-tip mark pixel (i, horizontal gradient j), gradY (i, j) feeling the pulse with the finger-tip mark pixel (i, VG (vertical gradient) j), (i j+1) refers to pixel (i, gradient j+1) to grad, (i j-1) refers to pixel (i, gradient j-1) to grad, (i+1 j) refers to pixel (i+1, gradient j) to grad, and grad (i-1 j) refers to pixel (i-1, gradient j); And
Use following expression calculated level outline map and vertical edge figure:
Here, EdgeX (i, j) mean horizontal edge figure object pixel (i, j), EdgeY (i, i) mean vertical edge figure object pixel (i, j).
Preferably, described candidate's line determines that device is configured to: by connecting each horizontal edge figure that calculates and the edge pixel scounting line segment among the vertical edge figure; The line segment that searches with merging is as candidate's line.
Preferably, described candidate's line determines that device further is configured to remove the line segment of its length less than the intended pixel scope.Preferably, the intended pixel scope is a 6-10 pixel.
Preferably, described merging device merges the line segment based on relation of the position between the line segment and pitch angle thereof.
Preferably, if adjacent line segment has less than the gap of intended pixel scope and has identical pitch angle, then merge adjacent line segment.Preferably, Yu Ding pixel coverage is a 4-8 pixel.
The present invention also provides the medium of the program of the step that a kind of program of carrying out line detecting method and storage be used to carry out above-mentioned line detecting method.
Will know other features and advantages of the present invention by description taken together with the accompanying drawings hereinafter.
Description of drawings
The accompanying drawing of incorporating in this manual and constituting its part shows embodiments of the invention, and is used from and explains principle of the present invention together with describing one.
Accompanying drawing 1 is depicted as the process flow diagram that detects the line in input picture according to the present invention.
Accompanying drawing 2a and 2b are depicted as VG (vertical gradient) pixel that forms horizontal edge and the horizontal gradient pixel that forms vertical edge respectively.
Accompanying drawing 3 is depicted as the vertical and horizontal edge figure that is formed by the original table table images.
Accompanying drawing 4 is depicted as the vertical and horizontal edge figure that is formed by analog image.
Accompanying drawing 5a-5d is depicted as the line segment that is produced by 4 kinds of real images respectively.
Accompanying drawing 6 is depicted as an example based on relation of the position between candidate's line and pitch angle merging candidate line thereof.
Accompanying drawing 7 is depicted as segment and merges processing.
Accompanying drawing 8a-8c is depicted as and is using according to form line, business card image and the profile of defiber and some testing results of barcode lines after the line detecting method of the present invention.
Accompanying drawing 9 is depicted as application of the present invention.
Accompanying drawing 10 is depicted as the line checkout equipment of enforcement according to line detecting method of the present invention.
Embodiment
The hereinafter with reference accompanying drawing describes preferential embodiment of the present invention in detail now.
Accompanying drawing 1 is depicted as the process flow diagram according to most preferred embodiment detection line of the present invention, and its key step is as follows:
Step 1: the gradient map of calculating input image.
Step 2: calculating contains the direction outline map.
Step 3: the connection element of search candidate line in each outline map.
Step 4: position-based and pitch angle merge candidate's line.
Step 5: judge whether candidate's line is solid line.
The step 1 of accompanying drawing 1 relates to the gradient map of calculating input image.
The rate of change that various edge detection operators can be used for checking the neighbouring relations of each pixel and limiting its gray level is with edge calculation figure, and these are general knowledges, therefore is not described in detail at this.This most preferred embodiment uses the Sobel operator to calculate the gradient map of tablet pattern, uses Otsu ' s method calculated threshold, and obtains outline map according to gradient map and threshold value.Can store gradient map in order to using afterwards, this will be described below.
The step 2 of accompanying drawing 1 relates to calculating and contains the direction outline map.
Edge pixel can be divided into two classes according to gradient direction: horizontal edge pixel and vertical edge pixel, and shown in accompanying drawing 2a and 2b.Usually, several vertical edge pixels that are provided with regularly appear on the horizontal position, and the horizontal edge pixel appears on the perpendicular line position.
Based on above-mentioned knowledge, the present invention can obtain horizontal edge figure and vertical edge figure from gradient map.
2.1) different operator calculated level gradient map " gradX " and the vertical gradient map " gradY " of use:
gradX(i，j)＝grad(i，j+1)-grad(i，j-1)
gradY(i，j)＝?grad(i+1，j)-grad(i-1，j)
Here gradX (i, j) feeling the pulse with the finger-tip mark pixel (i, horizontal gradient j),
GradY (i, j) feeling the pulse with the finger-tip mark pixel (i, VG (vertical gradient) j),
Grad (i, j+1) refer to pixel (i, gradient j+1),
Grad (i, j-1) refer to pixel (i, gradient j-1),
Grad (i+1, j) refer to pixel (i+1, gradient j) and
(i-1 j) refers to pixel (i-1, gradient j) to grad.
2.2) use following expression that gradient map is divided into horizontal edge figure " EdgeX " and vertical edge figure " EdgeY ";
Here, EdgeX (i, j) mean object pixel (i j) is confirmed as the horizontal edge pixel, and EdgeY (i j) means that (i j) is confirmed as the vertical edge pixel to object pixel.
Therefore, if gradX (i, absolute value j) greater than gradY (i, absolute value j), then as can be known pixel (i j) is the horizontal edge pixel, and Edge (i, j)=1, and EdgeY (i, j)=0, vice versa.
Then, what obtain two binaryzations contains the direction outline map, wherein searches out the pixel that is provided with the regularly candidate's line as the back.Accompanying drawing 3 is depicted as the vertical and horizontal edge figure of actual digital camera images, and wherein white point is an edge pixel.Accompanying drawing 4 is depicted as the vertical and horizontal edge figure of analog image.
At this, the invention provides and use edge gradient to define level or vertical edge, it does not require that level or vertical edge are abswolute levels or vertical.In the present embodiment, have with respect to the reference levels line and be defined as horizontal edge, and these edges that have greater than 45 ° angle are defined as vertical edge, as shown in Figure 4 less than these edges of 45 ° angle.Therefore, the present invention not only can detect the line of almost vertical and level, and can easily detect all lines of various angles.
The step 3 of accompanying drawing 1 relates to respectively at two connection elements that contain search candidate line in the direction outline map.The field of search is selected as 8 neighborhoods in the present embodiment.For those of ordinary skills clearly, except being communicated with the element localization method as used herein, the edge pixel connection processing of other of the connection border that formation is closed also can be used.
As optimal steps, the step 4 of accompanying drawing 1 is that these line segments as candidate's line are merged.
From above-mentioned outline map, only the edge pixel of line is arranged in connected region as can be seen, and those text edges are dispersed in the different zones.Therefore, usually be communicated with element and enough locate these lines by search.
But in application, we must consider following practical factor:
Have some breakpoints in online.
Line normally tilts, and the line of inclination has caused serrate.
Some little curves may be present in the line.
The form line is to intersect and connect.
Above-mentioned these features of line may cause line to occur as the short-term of some separation, promptly in this said line segment, these only may occur online in.Accompanying drawing 5a is depicted as some breakpoints of the line that is positioned on the edge; Accompanying drawing 5b is depicted as zigzag two lines; Accompanying drawing 5c is depicted as two approximate horizontal lines that intersect and be connected with other the line of approximate vertical.
Therefore need to solve the method for these difficult points, this present embodiment proposed new processing as optimal steps 4 these short-term segments are merged in the long line according to these features.
The position relation
Two line segments that belong to same line are close to each other, and one rear end must be near another front end.
The pitch angle
Two line segments that belong to same line must have the roughly angle (θ 1 in accompanying drawing 6 and θ 2) of inclination, and the segment that merges also has identical pitch angle (θ in the accompanying drawing 6).
Accompanying drawing 7 is depicted as the detailed process of the combining step 4 in accompanying drawing 1.
Entire process comprises:
Step 1: its length is carried out elimination or removing less than those segments (connection element) of 8 pixels;
Step 2: judge whether any segment in two segments satisfies above-described two merging conditions, promptly;
Segment gap＜6 (pixel);
θ1＝θ2＝θ；
Step 3: if two line segments satisfy this two conditions, then the present invention will think that they are arranged on the same line and with them and merge to one, as shown in Figure 6.
Step 4: judge whether all segments have all been merged.
Whether the segment gap between the pixel of the segment of removing and the online segment all can change according to the requirement of reality (clear such as image, and whether image comprises form etc.).Usually, the length of the segment that remove (connection element) is selected as less than 6-10 pixel, preferably less than 8 pixels.The segment gap is selected less than 4-8 pixel, preferably less than 6 pixels.
The segment of this novelty merges handles the difficult point that can solve line breakpoint, inclination, bending and intersect.
After having merged all line segments, in the step 5 of accompanying drawing 1, can determine by its length whether the segment that merges is actual line in horizontal edge figure or vertical edge figure.If judge that the segment that merges is a solid line, then we have obtained line position.Otherwise, will throw away segment after the merging according to current line detecting method.
Note that the physical length that is used to judge solid line reference length can by those of ordinary skills rule of thumb and/or other application requirements determine, will no longer describe in the present invention.
The result
Accompanying drawing 8a-8c is depicted as some testing results of using above-mentioned line detecting method, can find out respectively that from accompanying drawing 8a-8c the profile of form line, business card image and defiber and barcode lines are detected satisfactorily.
Use
The present invention can be used for for example different line or the form lines in location in projection DC-Bina (digital camera images binaryzation) as shown in Figure 9, and line that these are different or form line may disturb text detection or text to merge, but in this application of the invention it are deleted.
In above-mentioned description, the present invention has described in the preferred embodiment as method or software program.About the present invention, be understood that easily the present invention is preferred for any known computer system such as on the personal computer.Therefore, computer system will no longer go through.It should be appreciated that image can be directly inputted to computer system (for example passing through digital camera) or digitizing before being input to computer system (for example by scanning).
In addition, just as used herein, have therein the computer-readable storage medium that storage is used to carry out the computer program of said method and for example can comprise that magnetic storage media is such as disk (such as floppy disk) or tape; Optical storage media is such as CD, light belt or machine-readable bar code; The solid-state electronic memory device is such as random-access memory (ram); Or be used for any other physical device or the medium of storage computation machine program.
In addition, those of ordinary skills recognize easily also can be with the equivalent of the above-mentioned software of hardware design.
Accompanying drawing 10 is depicted as the line checkout equipment of implementing said method, comprising:
Contain direction outline map calculation element 100, be used for based on input picture calculated level outline map and vertical edge figure;
Candidate's line is determined device 101, is used for horizontal edge figure that calculates based on each and the vertical edge figure that is calculated and determines candidate's line; And
Line is determined device 102, is used for determining based on described candidate's line the position of line.
Contain direction outline map calculation element 100 and can utilize Sobel wave filter compute gradient figure and calculated threshold, obtain outline map according to gradient map and threshold value thus.Line determines that device 102 comprises with searcher by being connected each horizontal edge figure that calculates and the edge pixel scounting line segment among the vertical edge figure and merges the line segment that the searches merging device as candidate's line.
With reference to certain embodiments the present invention has been described.It should be understood that the present invention is not limited to description above, those of ordinary skills can carry out various changes and modification to the present invention under the premise without departing from the spirit and scope of the present invention.
Claims (14)
1. line detecting method comprises:
First calculation procedure is used for the gradient of the pixel that calculating input image comprises;
Second calculation procedure is used to use following equation calculated level gradient and VG (vertical gradient):
GradX (i, j)=grad (i, j+1)-grad (i, j-1) and
gradY(i，j)＝grad(i+1，j)-grad(i-1，j)
Here gradX (i, j) feeling the pulse with the finger-tip mark pixel (i, horizontal gradient j), gradY (i, j) feeling the pulse with the finger-tip mark pixel (i, VG (vertical gradient) j), (i j+1) refers to pixel (i, gradient j+1) to grad, (i j-1) refers to pixel (i, gradient j-1) to grad, (i+1 j) refers to pixel (i+1, gradient j) to grad, and grad (i-1 j) refers to pixel (i-1, gradient j);
The 3rd calculation procedure is used to use following expression calculated level outline map and vertical edge figure:
Here, EdgeX (i, j) refer to horizontal edge figure object pixel (i, j), EdgeY (i, j) refer to vertical edge figure object pixel (i, j);
Candidate's line determining step is used for determining candidate's line by being incorporated in each horizontal edge figure that calculates and vertical edge figure based on the line segment that the connection element searches; And
The line determining step is used for judging according to each length of candidate's line whether in described candidate's line each is solid line, and determines to be judged as the position of candidate's line of solid line.
2. line detecting method according to claim 1, wherein the gradient of the pixel that comprises in the input picture uses the Sobel operator to calculate in described first calculation procedure.
3. line detecting method according to claim 1, wherein said candidate's line determining step also comprise removes the line segment of its length less than the intended pixel scope.
4. line detecting method according to claim 3, wherein the intended pixel scope is a 6-10 pixel.
5. line detecting method according to claim 1 wherein in described candidate's line determining step, merges the line segment based on relation of the position between the line segment and pitch angle thereof.
6. line detecting method according to claim 5 if wherein adjacent line segment has less than the gap of intended pixel scope and has identical pitch angle, then merges adjacent line segment.
7. line detecting method according to claim 6, wherein said intended pixel scope are 4-8 pixels.
8. line checkout equipment comprises:
First calculation element is used for the gradient of the pixel that calculating input image comprises;
Second calculation element is used to use following equation calculated level gradient and VG (vertical gradient):
GradX (i, j)=grad (i, j+1)-grad (i, j-1) and
gradY(i，j)＝grad(i+1，j)-grad(i-1，j)
Here gradX (i, j) feeling the pulse with the finger-tip mark pixel (i, horizontal gradient j), gradY (i, j) feeling the pulse with the finger-tip mark pixel (i, VG (vertical gradient) j), (i j+1) refers to pixel (i, gradient j+1) to grad, (i j-1) refers to pixel (i, gradient j-1) to grad, (i+1 j) refers to pixel (i+1, gradient j) to grad, and grad (i-1 j) refers to pixel (i-1, gradient j);
The 3rd calculation element is used to use following expression calculated level outline map and vertical edge figure:
Here, EdgeX (i, j) refer to horizontal edge figure object pixel (i, j), EdgeY (i, j) refer to vertical edge figure object pixel (i, j);
Candidate's line is determined device, is used for determining candidate's line by being incorporated in each horizontal edge figure that calculates and vertical edge figure based on the line segment that the connection element searches; And
Line is determined device, is used for judging according to each length of candidate's line whether in described candidate's line each is solid line, and determines to be judged as the position of candidate's line of solid line.
9. line checkout equipment according to claim 8, wherein said first calculation element use the gradient of the pixel that comprises in the Sobel operator calculating input image.
10. line checkout equipment according to claim 8, wherein said candidate's line determine that device further is configured to remove the line segment of its length less than the intended pixel scope.
11. line checkout equipment according to claim 10, wherein the intended pixel scope is a 6-10 pixel.
12. line checkout equipment according to claim 8 is wherein determined device by described candidate's line, merges the line segment based on relation of the position between the line segment and pitch angle thereof.
13. line checkout equipment according to claim 12 if wherein adjacent line segment has less than the gap of intended pixel scope and has identical pitch angle, then merges adjacent line segment.
14. line checkout equipment according to claim 13, wherein the intended pixel scope is a 4-8 pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101287847A CN100428270C (en) | 2005-12-02 | 2005-12-02 | Line detecting method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101287847A CN100428270C (en) | 2005-12-02 | 2005-12-02 | Line detecting method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1979528A CN1979528A (en) | 2007-06-13 |
| CN100428270C true CN100428270C (en) | 2008-10-22 |
Family
ID=38130683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005101287847A Expired - Fee Related CN100428270C (en) | 2005-12-02 | 2005-12-02 | Line detecting method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100428270C (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100552698C (en) * | 2007-12-11 | 2009-10-21 | 中国科学院长春光学精密机械与物理研究所 | Real-time detection method to airfield runway in the image of taking photo by plane |
| CN104166977B (en) * | 2013-05-17 | 2017-10-10 | 中国航空工业集团公司洛阳电光设备研究所 | A kind of Image Matching Similarity Measurement Method and its image matching method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0584662A1 (en) * | 1992-08-25 | 1994-03-02 | NOKIA TECHNOLOGY GmbH | Method and coupling for doubling the vertical and horizontal frequency of a video signal |
| JPH08111852A (en) * | 1994-10-12 | 1996-04-30 | Sharp Corp | Motion adaptive scanning line interpolation device |
| US5835614A (en) * | 1992-06-26 | 1998-11-10 | Honda Giken Kogyo Kabushiki Kaisha | Image processing apparatus |
| CN1393827A (en) * | 2001-06-26 | 2003-01-29 | 中国科学院遥感应用研究所 | Method for using computer to recreate image containing spot noise |
| CN1440192A (en) * | 2002-02-21 | 2003-09-03 | 三星电子株式会社 | Marginal correcting method and device |
-
2005
- 2005-12-02 CN CNB2005101287847A patent/CN100428270C/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5835614A (en) * | 1992-06-26 | 1998-11-10 | Honda Giken Kogyo Kabushiki Kaisha | Image processing apparatus |
| EP0584662A1 (en) * | 1992-08-25 | 1994-03-02 | NOKIA TECHNOLOGY GmbH | Method and coupling for doubling the vertical and horizontal frequency of a video signal |
| JPH08111852A (en) * | 1994-10-12 | 1996-04-30 | Sharp Corp | Motion adaptive scanning line interpolation device |
| CN1393827A (en) * | 2001-06-26 | 2003-01-29 | 中国科学院遥感应用研究所 | Method for using computer to recreate image containing spot noise |
| CN1440192A (en) * | 2002-02-21 | 2003-09-03 | 三星电子株式会社 | Marginal correcting method and device |
Non-Patent Citations (4)
| Title |
|---|
| 一种改进的基于小波变换的图像边缘提取算法. 高国荣，刘冉，羿旭明.武汉大学学报(理学版)，第31卷第5期. 2005 |
| 一种改进的基于小波变换的图像边缘提取算法. 高国荣，刘冉，羿旭明.武汉大学学报(理学版)，第31卷第5期. 2005 * |
| 新的梯度边缘检测方法. 赵春江，施文康，邓勇.光电工程，第32卷第4期. 2005 |
| 新的梯度边缘检测方法. 赵春江，施文康，邓勇.光电工程，第32卷第4期. 2005 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1979528A (en) | 2007-06-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10885644B2 (en) | Detecting specified image identifiers on objects | |
| US7519214B2 (en) | System and method of determining image skew using connected components | |
| JP5050075B2 (en) | Image discrimination method | |
| Aradhye | A generic method for determining up/down orientation of text in roman and non-roman scripts | |
| US20080226171A1 (en) | Correcting device and method for perspective transformed document images | |
| JPH09179937A (en) | Method for automatically discriminating boundary of sentence in document picture | |
| CN108805116A (en) | Image text detection method and its system | |
| CN106960208A (en) | A kind of instrument liquid crystal digital automatic segmentation and the method and system of identification | |
| CN108960221B (en) | Bank card identification method and device based on image | |
| Shivakumara et al. | New gradient-spatial-structural features for video script identification | |
| CN110598566A (en) | Image processing method, device, terminal and computer readable storage medium | |
| CN111259891A (en) | Method, device, equipment and medium for identifying identity card in natural scene | |
| US11216905B2 (en) | Automatic detection, counting, and measurement of lumber boards using a handheld device | |
| CN100428270C (en) | Line detecting method and device | |
| JPH07105312A (en) | Method and device for eliminating dirt from character image in optical character reader | |
| Dubey | Edge based text detection for multi-purpose application | |
| Bhaskar et al. | Implementing optical character recognition on the android operating system for business cards | |
| WO2022019018A1 (en) | Segmentation of continuous dynamic scans | |
| CN113158755A (en) | Method for improving accuracy of bank pipelining recognition | |
| Amarnath et al. | Automatic localization and extraction of tables from handheld mobile-camera captured handwritten document images | |
| KR20070078509A (en) | Method character recognition in a giro paper teller machine | |
| JPH03225578A (en) | Method for detecting and segmenting character | |
| EP3872707A1 (en) | Automatic detection, counting, and measurement of lumber boards using a handheld device | |
| US20170200383A1 (en) | Automated review of forms through augmented reality | |
| JPH0573718A (en) | Area attribute identifying system |
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 | ||
| 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: 20081022 Termination date: 20211202 |