CN106251311A - A kind of feature extraction algorithm of cross - Google Patents
A kind of feature extraction algorithm of cross Download PDFInfo
- Publication number
- CN106251311A CN106251311A CN201610645240.6A CN201610645240A CN106251311A CN 106251311 A CN106251311 A CN 106251311A CN 201610645240 A CN201610645240 A CN 201610645240A CN 106251311 A CN106251311 A CN 106251311A
- Authority
- CN
- China
- Prior art keywords
- image
- cross
- feature
- processing method
- image processing
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 11
- 238000003672 processing method Methods 0.000 claims abstract description 21
- 238000000034 method Methods 0.000 claims description 16
- 238000009499 grossing Methods 0.000 claims description 11
- 238000001514 detection method Methods 0.000 claims description 8
- 230000004807 localization Effects 0.000 abstract description 4
- 239000002184 metal Substances 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000002902 bimodal effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to cut automatic field, the feature extraction algorithm of a kind of cross.The feature extraction algorithm of a kind of cross, specifically comprises the following steps that the gray level image of input camera shooting;The grey level histogram of statistical picture;Judge that image, the need of image enhaucament, is, carries out image enhancement processing according to rectangular histogram, carry out image processing method one;Image after image processing method one obtains feature contour, simulates four straight lines of cross outline;Judge whether four straight lines simulating cross outline constitute cross feature, be to terminate algorithm;Image after image processing method two obtains marginal information, simulates four straight lines of cross feature;Terminate algorithm.Vision localization application characteristic for spectacle-frame cutting industry, it is possible to improve greatly speed and the adaptability identifying cross feature, actually used during, as long as containing cross in image, centre coordinate and the anglec of rotation can be given quickly and accurately.
Description
Technical field
The present invention relates to cut automatic field, the feature extraction algorithm of a kind of cross.
Background technology
The spectacle-frame cutting demand that the present invention is originally derived from cut automatic field.Industry is cut at spectacle-frame
In, upper one procedure, is to corrode at sheet metal surface with liquid medicine decorative pattern and mark feature, then with the side of vision localization
Formula, on the most corresponding with sheet metal for the processing drawing of second operation work.Among these, a most important link, it is simply that fast
Speed identifies the mark feature in shot by camera picture, and returns mark center and other parameters accurately.And ten
Cabinet frame is a kind of mark feature that application is more.
General identification cross algorithm, uses the mode of template matching, is merely able to process the anglec of rotation well at 5 degree
Within cross feature, for the anglec of rotation excessive in the case of, or can not process, or speed is very slow, expends the time
More than 500ms.And, it is desirable to mark feature is obvious with background contrast, does not has the interference of impurity, is i.e. merely able to processing feature effect
Preferably image, concordance and stability requirement to characteristics of image are the highest.
Summary of the invention
The present invention is for overcoming the deficiencies in the prior art, it is provided that the feature extraction algorithm of a kind of cross, cuts for spectacle-frame
Cut the vision localization application characteristic of industry, it is possible to improve speed and the adaptability identifying cross feature greatly, actually used
During, as long as containing cross in image, centre coordinate and the anglec of rotation can be given quickly and accurately and same
Width image, the error repeatedly searched, in positive negative one pixel coverage, i.e. has preferable stability.
For achieving the above object, the feature extraction algorithm of a kind of cross is designed, it is characterised in that: specifically comprise the following steps that
(1) gray level image of input camera shooting;
(2) grey level histogram of statistical picture;
(3) judge that image, the need of image enhaucament, is according to rectangular histogram, carry out image enhancement processing, then carry out at image
Reason method one;The most directly carry out image processing method one;
(4) image after image processing method one obtains feature contour information, simulates four straight lines of cross outline;
(5) judge whether four straight lines simulating cross outline constitute cross feature, be to terminate algorithm;Otherwise carry out
Image processing method two;
(6) image after image processing method two obtains marginal information, simulates four straight lines of cross feature;
(7) algorithm is terminated.
Described image processing method one is as follows:
(1) original image is carried out Gaussian smoothing, remove noise and interference;
(2) obtaining bianry image, be wherein characterized as white, gray value is 255, and background is black, and gray value is 0;
(3) bianry image after processing carries out morphology opening operation process, removes burr little around profile;
(4) search all of profile in image, it is believed that maximum profile is cross characteristics profile, the coordinate of largest contours point is believed
Breath preserves.
Described image processing method two is as follows:
(1) image is carried out Gaussian smoothing, remove noise and interference;
(2) then the image after Gaussian smoothing carries out estimating the number of edges of desired output, and utilize canny operator to enter
Row rim detection;
(3) image after traversal rim detection, by the point that all gray values are 255, saves as marginal information, wherein most
Marginal information for cross characteristics.
Described image enhancement processing is that histogram equalization processes.
The present invention is directed to the vision localization application characteristic of spectacle-frame cutting industry, it is possible to improve greatly and identify that cross is special
The speed levied and adaptability, actually used during, as long as containing cross in image, in can being given quickly and accurately
Heart coordinate and the anglec of rotation.
Accompanying drawing explanation
Fig. 1 is flow chart of the present invention.
Detailed description of the invention
Below according to accompanying drawing, the present invention is described further.
As it is shown in figure 1, specifically comprise the following steps that
(1) gray level image of input camera shooting;
(2) grey level histogram of statistical picture;
(3) judge that image, the need of image enhaucament, is according to rectangular histogram, carry out image enhancement processing, then carry out at image
Reason method one;The most directly carry out image processing method one;
(4) image after image processing method one obtains feature contour information, simulates four straight lines of cross outline;
(5) judge whether four straight lines simulating cross outline constitute cross feature, be to terminate algorithm;Otherwise carry out
Image processing method two;
(6) image after image processing method two obtains marginal information, simulates four straight lines of cross feature;
(7) algorithm is terminated.
Image processing method one is as follows:
(1) original image is carried out Gaussian smoothing, remove noise and interference;
(2) obtaining bianry image, be wherein characterized as white, gray value is 255, and background is black, and gray value is 0;
(3) bianry image after processing carries out morphology opening operation process, removes burr little around profile;
(4) search all of profile in image, it is believed that maximum profile is cross characteristics profile, the coordinate of largest contours point is believed
Breath preserves.
Image processing method two is as follows:
(1) image is carried out Gaussian smoothing, remove noise and interference;
(2) then the image after Gaussian smoothing carries out estimating the number of edges of desired output, and utilize canny operator to enter
Row rim detection;
(3) image after traversal rim detection, by the point that all gray values are 255, saves as marginal information, wherein most
Marginal information for cross characteristics.
Image enhancement processing is that histogram equalization processes.
In the gray-scale map of normal input, main information is two parts, the low intensity value ranges with cross characteristics as representative,
High intensity value ranges with background as representative.It is therefore contemplated that in grey level histogram, there are two crests.Calculate two ripples
Distance between peak, when less than a certain threshold value, then it is assumed that feature is inconspicuous with background difference, needs to carry out image enhaucament.As
, there are not two crests in the most original rectangular histogram, its data can carry out the smooth operation of certain radius size, when iterating to
Certain number of times, when there is not yet bimodal, then directly returns and need not image enhaucament.
Image, through Gaussian smoothing, thresholding, morphology operations, after extracting profile, obtains the institute of cross feature outline
There is coordinate points.Preserving two parts of coordinate points, first part of coordinate points, according to x coordinate size, is divided into two classes, and second part of coordinate points is according to y
Coordinate size, is divided into two classes, obtains four class coordinate points.The method being respectively adopted stochastic sampling, removal is unsatisfactory for fitting a straight line and wants
The impure point asked, then carries out the fitting a straight line of method of least square to remaining point, i.e. obtains four straight lines.
Randomly select straight line as basis reference, judge the relation of remaining straight line and reference line successively, one
In the range of determining angular error, it is necessary to be parallel or vertical relation.If there being straight line to be unsatisfactory for condition, then it is assumed that not
Become cross feature, otherwise proceed to judge.Then, obtain the distance between parallel lines respectively, obtain small distance with
The ratio of relatively large distance, if less than certain threshold value, then it is assumed that become cross feature not, otherwise it is assumed that it is special to constitute cross
Levy.
After Gaussian smoothing and rim detection, obtain the marginal information of cross feature.Marginal information is with profile information not
With, profile is continuous print Guan Bi, and edge is not necessarily continuous print Guan Bi.First, in the marginal point of magnanimity, use
The mode of stochastic sampling, through multiple repairing weld, finds out by the Article 1 straight line that multiple spot simulates, and it represents with x-axis angle
For, and these points are deleted from marginal point.Then, the geometric properties of cross is utilized, it may be determined that other three straight lines
Angle with x-axis、、, then in certain angle range of error, search straight line two, three, four successively, often find out one
Straight line, it is possible to will meet all point deletions of this straight line, to improve the efficiency that lower bar straight line is searched.Finally, ten have been obtained
Four straight line information after the rim detection of cabinet frame feature.
Claims (4)
1. the feature extraction algorithm of a cross, it is characterised in that: specifically comprise the following steps that
(1) gray level image of input camera shooting;
(2) grey level histogram of statistical picture;
(3) judge that image, the need of image enhaucament, is according to rectangular histogram, carry out image enhancement processing, then carry out at image
Reason method one;The most directly carry out image processing method one;
(4) image after image processing method one obtains feature contour information, simulates four straight lines of cross outline;
(5) judge whether four straight lines simulating cross outline constitute cross feature, be to terminate algorithm;Otherwise carry out
Image processing method two;
(6) image after image processing method two obtains marginal information, simulates four straight lines of cross feature;
(7) algorithm is terminated.
The feature extraction algorithm of a kind of cross the most according to claim 1, it is characterised in that: described image processing method
Method one is as follows:
(1) original image is carried out Gaussian smoothing, remove noise and interference;
(2) obtaining bianry image, be wherein characterized as white, gray value is 255, and background is black, and gray value is 0;
(3) bianry image after processing carries out morphology opening operation process, removes burr little around profile;
(4) search all of profile in image, it is believed that maximum profile is cross characteristics profile, the coordinate of largest contours point is believed
Breath preserves.
The feature extraction algorithm of a kind of cross the most according to claim 1, it is characterised in that: described image processing method
Method two is as follows:
(1) image is carried out Gaussian smoothing, remove noise and interference;
(2) then the image after Gaussian smoothing carries out estimating the number of edges of desired output, and utilize canny operator to enter
Row rim detection;
(3) image after traversal rim detection, by the point that all gray values are 255, saves as marginal information, wherein most
Marginal information for cross characteristics.
The feature extraction algorithm of a kind of cross the most according to claim 1, it is characterised in that: at described image enhaucament
Manage and process for histogram equalization.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610645240.6A CN106251311A (en) | 2016-08-09 | 2016-08-09 | A kind of feature extraction algorithm of cross |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610645240.6A CN106251311A (en) | 2016-08-09 | 2016-08-09 | A kind of feature extraction algorithm of cross |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106251311A true CN106251311A (en) | 2016-12-21 |
Family
ID=58079141
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610645240.6A Pending CN106251311A (en) | 2016-08-09 | 2016-08-09 | A kind of feature extraction algorithm of cross |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106251311A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178210A (en) * | 2019-12-21 | 2020-05-19 | 中国电波传播研究所(中国电子科技集团公司第二十二研究所) | Image identification and alignment method for cross mark |
CN117990072A (en) * | 2024-04-03 | 2024-05-07 | 中交天津港湾工程研究院有限公司 | Automatic monitoring method for tunnel surrounding rock convergence |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101315699A (en) * | 2008-07-10 | 2008-12-03 | 哈尔滨工业大学 | Incremental variation level set fast medical image partition method |
CN101901477A (en) * | 2010-07-27 | 2010-12-01 | 中国农业大学 | Method and system for extracting field image edges of plant leaves |
US8670154B2 (en) * | 2011-04-28 | 2014-03-11 | Xerox Corporation | Data architecture for mixed resolution interleaved cross-channel data flow and format |
US20140341426A1 (en) * | 2013-05-14 | 2014-11-20 | Kabushiki Kaisha Toshiba | Image processing apparatus, image processing method and medical imaging device |
CN104794720A (en) * | 2015-05-04 | 2015-07-22 | 成都唐源电气有限责任公司 | Method and system for positioning steady ears in catenary images |
CN104966302A (en) * | 2015-07-09 | 2015-10-07 | 深圳中科智酷机器人科技有限公司 | Detecting and positioning method for laser cross at any angle |
-
2016
- 2016-08-09 CN CN201610645240.6A patent/CN106251311A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101315699A (en) * | 2008-07-10 | 2008-12-03 | 哈尔滨工业大学 | Incremental variation level set fast medical image partition method |
CN101901477A (en) * | 2010-07-27 | 2010-12-01 | 中国农业大学 | Method and system for extracting field image edges of plant leaves |
US8670154B2 (en) * | 2011-04-28 | 2014-03-11 | Xerox Corporation | Data architecture for mixed resolution interleaved cross-channel data flow and format |
US20140341426A1 (en) * | 2013-05-14 | 2014-11-20 | Kabushiki Kaisha Toshiba | Image processing apparatus, image processing method and medical imaging device |
CN104794720A (en) * | 2015-05-04 | 2015-07-22 | 成都唐源电气有限责任公司 | Method and system for positioning steady ears in catenary images |
CN104966302A (en) * | 2015-07-09 | 2015-10-07 | 深圳中科智酷机器人科技有限公司 | Detecting and positioning method for laser cross at any angle |
Non-Patent Citations (2)
Title |
---|
肖世昌等: "用于网版印刷自动套印的十字线检测新方法", 《中国印刷与包装研究》 * |
陈显峰: "机器视觉***中图像对准方法的研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178210A (en) * | 2019-12-21 | 2020-05-19 | 中国电波传播研究所(中国电子科技集团公司第二十二研究所) | Image identification and alignment method for cross mark |
CN117990072A (en) * | 2024-04-03 | 2024-05-07 | 中交天津港湾工程研究院有限公司 | Automatic monitoring method for tunnel surrounding rock convergence |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109785316B (en) | Method for detecting apparent defects of chip | |
CN105067638B (en) | Tire fetal membrane face character defect inspection method based on machine vision | |
CN107808378B (en) | Method for detecting potential defects of complex-structure casting based on vertical longitudinal and transverse line profile features | |
CN110163853B (en) | Edge defect detection method | |
CN103871053B (en) | Vision conspicuousness-based cloth flaw detection method | |
CN105260693A (en) | Laser two-dimensional code positioning method | |
CN102982545B (en) | A kind of image depth estimation method | |
CN112132153B (en) | Tomato fruit identification method and system based on clustering and morphological processing | |
CN109767445B (en) | High-precision PCB defect intelligent detection method | |
CN112053376B (en) | Workpiece weld joint identification method based on depth information | |
CN115115612B (en) | Surface defect detection method and system for mechanical parts | |
CN110717872A (en) | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning | |
CN111754538B (en) | Threshold segmentation method for USB surface defect detection | |
CN114399522A (en) | High-low threshold-based Canny operator edge detection method | |
CN110070523B (en) | Foreign matter detection method for bottle bottom | |
CN105741281B (en) | Method for detecting image edge based on neighborhood dispersion | |
CN104778707A (en) | Electrolytic capacitor detecting method for improving general Hough transform | |
CN112991374B (en) | Canny algorithm-based edge enhancement method, canny algorithm-based edge enhancement device, canny algorithm-based edge enhancement equipment and storage medium | |
CN112883881B (en) | Unordered sorting method and unordered sorting device for strip-shaped agricultural products | |
CN115147448A (en) | Image enhancement and feature extraction method for automatic welding | |
CN112884746A (en) | Character defect intelligent detection algorithm based on edge shape matching | |
CN112232222A (en) | Bullet train axle box end cover bolt loss fault detection method based on image processing | |
CN106251311A (en) | A kind of feature extraction algorithm of cross | |
CN113781413B (en) | Electrolytic capacitor positioning method based on Hough gradient method | |
CN106447683A (en) | Feature extraction algorithm of circles |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB02 | Change of applicant information | ||
CB02 | Change of applicant information |
Address after: 200240 No. 953 lane, Jianchuan Road, Minhang District, Shanghai 322 Applicant after: Shanghai Pak Chu electronic Polytron Technologies Inc Address before: 200240 west two floor, 2 building, 940 Jianchuan Road, Minhang District, Shanghai. Applicant before: Shanghai Bochu Electronic Technology Co., Ltd. |
|
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20161221 |