CN101271576A - Gridiron pattern recognition locating method under complex illumination and surface condition - Google Patents
Gridiron pattern recognition locating method under complex illumination and surface condition Download PDFInfo
- Publication number
- CN101271576A CN101271576A CNA2008100371044A CN200810037104A CN101271576A CN 101271576 A CN101271576 A CN 101271576A CN A2008100371044 A CNA2008100371044 A CN A2008100371044A CN 200810037104 A CN200810037104 A CN 200810037104A CN 101271576 A CN101271576 A CN 101271576A
- Authority
- CN
- China
- Prior art keywords
- pattern
- angle point
- point
- gridiron pattern
- angle
- 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
Images
Abstract
The invention relates to an identification and location method of a checkerboard mode under a condition of a complicated illumination and a complicated surface in the technical field of a computer vision. An angular point formed by four black-and-white areas in the checkerboard mode is detected; a boundary position of the black-and-white area in the angular point is recorded; through the position of the angular point and the corresponding boundary position, the angular points belonging to the black area or the white area in the same checkerboard are connected to get a set of the angular points interconnected; the number of the angular points in the checkerboard mode is confirmed through the known line number and the known column number of the checkerboard; thereby the set of the angular points in the checkerboard mode can be confirmed based on whether the number of the angular point in the obtained set of the angular points interconnected is equal to the number of the angular points in the checkerboard mode to identify the checkerboard mode. The identification and location method of the checkerboard mode overcomes the problems that the angular point of the checkerboard mode is difficult to identify on the complicated surface and the surface of an irregular object, and the complicated illumination environment is processed, thereby being capable of identifying and locating the checkerboard mode under the complicated illumination and on the complicated surface accurately.
Description
Technical field
The present invention relates to a kind of recognition positioning method of technical field of computer vision, specifically is gridiron pattern pattern recognition locating method under a kind of complex illumination and the surface condition.
Background technology
Proofread and correct or how much registrations through being commonly used to do by video camera in computer vision system for the gridiron pattern pattern.Use as the typical case in the camera chain, people such as Zhang Zhengyou in 2000 propose to use black and white round dot or pattern lattice to carry out video camera and proofread and correct.And in augmented reality system, people such as Raskar then used the black and white gridiron pattern to obtain to reflect the homography matrix of spatial alternation relation between projector and the video camera in 2002.In addition, undertaken by structured light in the research of modeling, the gridiron pattern pattern also can be used as a kind of ad hoc structured light pattern and carries out three-dimensional reconstruction.In all above-mentioned application a vital step all need camera acquisition to image in the gridiron pattern pattern is discerned and is located.In this respect, practical at present method seldom, cvFindChessboardCorners method in the OpenCV vision storehouse of having only Intel Company of comparative maturity practicality, and present method original intention all is the gridiron pattern pattern on the good illumination condition lower plane in identification location, do not consider the identification of gridiron pattern pattern on on-plane surface under the relatively poor illumination condition or complex object surface, and this is actual vision system just, the most applications of being run in the image intensifying reality system.Therefore, for the gridiron pattern pattern, study a kind of can be healthy and strong under complex illumination and surface condition identification and the method for location have great significance.
Find through literature search prior art, Zhongshi Wang etc. are in " Applied Mathematics andComputation " (" applied mathematics and calculating ", total the 185th phase 894-906 page or leaf in 2007) " the Recognition and location of the internal corners of planar checkerboardcalibration pattern image " that delivers on (identification and the location of angle point in the gridiron pattern correction mode image of plane), identification to gridiron pattern mode image on the plane has been proposed in this article, concrete grammar is: the local light characteristic based on angle point in the gridiron pattern pattern is carried out Corner Detection, detects angle point in the gridiron pattern that is centered on by four black and white grids; Then by angle point in the gridiron pattern being regarded as the joining of tessellated grid boundary line, carry out the gridiron pattern angle point and must discern and mate.Its deficiency is: this method can only be discerned the gridiron pattern pattern angle point on the plane, for irregular surface, such as any complex surface and irregularity surface, can not effectively discern, and not consider for the photoenvironment of complexity, lacks robustness.
Summary of the invention
The present invention be directed to the deficiencies in the prior art, gridiron pattern pattern recognition locating method under a kind of complex illumination and the arbitrary surfaces condition is provided, make it overcome the difficulty of identification gridiron pattern pattern angle point on complex surface and the irregularity surface, and the photoenvironment to complexity is handled, and can stalwartness discern location gridiron pattern pattern exactly on complex illumination and complex surface.
The present invention is achieved by the following technical solutions, the present invention detects the angle point that is made of four black and white zones in the gridiron pattern pattern, and the boundary position in record black and white zone, angle point place, be communicated with the angle point that belongs to the black or white region of same gridiron pattern by corner location and corresponding boundary position, thereby the angle point that obtains being interconnected set, determine angle point number in the gridiron pattern pattern by known gridiron pattern line number and columns, thereby the angle point number that whether equals the gridiron pattern pattern according to angle point number in the angle point set of gained connection is determined the set of gridiron pattern angle point, thereby identification gridiron pattern pattern.
The present invention includes for two steps:
The first step is the detection of regional angle point;
Second step was the coupling of gridiron pattern pattern.
The detection of described regional angle point, be specially: have a few in the traversing graph picture, to each picture point, organizing the one dimension distribution signals 2D signal of circumjacent neighborhood point around it is converted into according to the distance of distance center point more, on one-dimensional signal, adopt the mean value binarization method to carry out binaryzation, to the one-dimensional signal after the binaryzation, carry out annular morphology open with closed operation to eliminate noise, by ask after the denoising all points first derivative values on the signal and, obtain the sum in chequered with black and white zone on the former 2D signal of this one-dimensional signal representative.If the interior many groups one-dimensional signal that 2D signal transformed of the certain neighborhood of picture point has the signal more than specified quantity that 4 chequered with black and white zones are arranged, this picture point is detected as the angle point of gridiron pattern pattern.
Described annular morphological operation, be meant the one-dimensional signal that is transformed by 2D signal originally for, when carrying out morphological operation, regard the head and the tail point of one-dimensional signal as on the geometric position adjacent, also be about to one-dimensional signal and regard an end to end ring as and carry out morphological operation.
Whether the coupling of described gridiron pattern pattern is specially: the result points of Corner Detection will be carried out the operation of denoising and cancellation redundant points, have around the angle point angle point of setting quantity to determine that whether this put noise point in the setting regions by detecting.By the angle point of phase mutual edge distance less than setpoint distance is divided in the same subclass, use the barycenter of subclass to replace whole angle point subclass to eliminate redundant points.Definition angle point vector is an angle point zone boundary vector to the difference of its corresponding zone boundary location point vector, if the angle point zone boundary vector angle of two different angle points is within the specific limits near 180 ° or-180 °, and distance is not more than the gridiron pattern grid area size of determining according to picture size and gridiron pattern ranks number between two angle points, thinks that then two angle points can be connected.After all angle points are connected test, obtain several connected components of whole angle point set, calculate true angle point number in the pattern according to the ranks number of known gridiron pattern pattern, the connected component that will have true several angle points of angle point is considered as the angle point set of gridiron pattern pattern.
The present invention can carry out the image pre-service as required before carrying out above-mentioned two step process, be about to image and do not change in the resolving range that length breadth ratio zooms to common video camera.
The inventive method is by being converted into the distribution of the point of two dimension the layer of one dimension in regional Corner Detection, thereby makes the provincial characteristics of regional angle point be more prone to detect, and healthy and strong more.And, do not need to consider that the rotation in the two dimension influences because the zone detection is on the one dimension layer, therefore natural robustness is arranged for rotational transform.
Description of drawings
The gridiron pattern pattern example that Fig. 1 is one 20 * 20.
Two complex illuminations of Fig. 2 and subsurface gridiron pattern pattern photo example.
Fig. 3 adopts two complex illuminations of the present invention and subsurface gridiron pattern pattern photo example;
Identification and positioning result in the photo, the regional angle point in the first step marks with cross, and real gridiron pattern angle point is marked with circle.
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
Fig. 1 is the example of a undeformed original gridiron pattern pattern, is used for the source images of our method projector institute projection.
Among Fig. 2 the gridiron pattern pattern photos of two pairs under complex illumination and surface condition.Fig. 2 (a) is an A4 paper that is printed on the gridiron pattern pattern, by its torsional deformation is taken, Fig. 2 (b) is in indoor corner and the ceiling intersection gridiron pattern pattern by the projector projection, and as can be seen because a plurality of planes distortion in corner, the gridiron pattern pattern is by gross distortion.
By on the paper of an A4 size, printing 26 * 22 chequered with black and white tessellated images of a pair, can obtain the paper among Fig. 2 (a), with the random torsional deformation of paper, guarantee that in the imaging scope of video camera as seen the gridiron pattern pattern can both obtain the effect of Fig. 2 (a).
Adopt common projector, intersection with 26 * 22 chequered with black and white tessellated image projection to indoor ceilings of a pair and corner, guarantee gridiron pattern pattern that projection goes out can both shooting squeeze in the imaging scope as seen, obtained the effect of Fig. 2 (b).
Use above-mentioned two width of cloth pictures of common camera collection, do not need, use the inventive method that it is carried out the identification and the location of gridiron pattern pattern through special processing:
If target gridiron pattern pattern has the capable c row point of r, i.e. (r+1) * (c+1) individual black and white lattice.R=22-1=21 herein, c=26-1=25.
At first pre-service does not change length breadth ratio with image and zooms in 300 * 200 to 2100 * 1600 scopes.
Angle point (regional angle point) detects in 1 gridiron pattern
(1) with every bit p (x on the window mode traversing graph picture
p, y
p), window R (p w) is defined as: and R (p, w)={ p
i‖ x
Pi-x
p|≤w, | y
Pi-y
p|≤w}, W gets 10.
(2) (p w), is divided into 1 ~ w layer with the picture point that is covered from inside to outside successively, and definition is layered as: L to R
r(p)=and R (p, r)-R (p, r-1), 1≤r≤w.
(3) at R (p, w) interior each layer L
r(p) carry out the mean value binaryzation on, adopt ring type morphology to open and closed operation (Ring-Morphology) denoising then.The ring type morphological operation is defined as:
(4) L after the denoising of detection binaryzation
rThe number in the zone that black and white replaces (p) is if having black and white alternately 4 zones, then L
r(p) be regional angle point layer.Define the notion of regional angle point layer thus.
(5) if (p has the regional angle point layer of α=80% in w), then p (x to R
p, y
p) be regional angle point.Write down outermost region angle point layer L after the binaryzation simultaneously
r(p) position of the point of last 4 place's black and white intersections is to set B
p, be called p (x
p, y
p) zone boundary point set.
(6) searching loop is had a few, and regional angle point set is designated as N.
2 gridiron pattern pattern match
(1) for any 2 point among the N, if distance is less than W between the two
Se, W
Se=5, think that then be connected at 2, (N, E), E gathers for the limit that is communicated with pie graph G thus.Calculate barycenter for each connected component among the G, obtain barycenter set N '.The zone boundary point set of barycenter is combined into the zone boundary point set of any point that connects barycenter in the connected component recently.
(2) to any 2 p among the N '
1, p
2, its zone boundary point set is combined into B
P1, B
P2If there is p '
1∈ B
P1, p '
2∈ B
P2, make with p
1, p '
1, p '
2, p
2The order conllinear, then think p
1, p
2Be connected.Thereby obtain figure G ' (N ', E '), E ' is for being communicated with the limit set.
(3) if there is connected component C among the G ', number of vertex among the C | C|=r * c, then the summit is the gridiron pattern angle point among the C.Thereby find gridiron pattern pattern and corner location thereof.The results are shown in accompanying drawing 3, the regional angle point barycenter of the connection that all obtain in the Corner Detection of method first step zone all is detected with the cross curve mark, and the angle point that belongs to the gridiron pattern pattern that obtains in the second step gridiron pattern pattern matching process is all marked with circle.All angle points all detection and location are correct, do not omit and unnecessary angle point.The validity and the accuracy of the inventive method have been proved.
Claims (4)
1, gridiron pattern pattern recognition locating method under a kind of complex illumination and the surface condition, it is characterized in that, detect the angle point that constitutes by four black and white zones in the gridiron pattern pattern, and the boundary position in record black and white zone, angle point place, be communicated with the angle point that belongs to the black or white region of same gridiron pattern by corner location and corresponding boundary position, thereby the angle point that obtains being interconnected set, determine angle point number in the gridiron pattern pattern by known gridiron pattern line number and columns, the angle point number that whether equals the gridiron pattern pattern according to angle point number in the angle point set of gained connection is determined the set of gridiron pattern angle point, thus identification gridiron pattern pattern.
2, gridiron pattern pattern recognition locating method under complex illumination according to claim 1 and the surface condition is characterized in that, comprises two steps:
The first step is the detection of regional angle point,
Second step was the coupling of gridiron pattern pattern,
The detection of described regional angle point, be specially: have a few in the traversing graph picture, to each picture point, organizing the one dimension distribution signals 2D signal of circumjacent neighborhood point around it is converted into according to the distance of distance center point more, on one-dimensional signal, adopt the mean value binarization method to carry out binaryzation, to the one-dimensional signal after the binaryzation, carry out annular morphology open with closed operation to eliminate noise, by ask after the denoising all points first derivative values on the signal and, obtain the sum in chequered with black and white zone on the former 2D signal of this one-dimensional signal representative, if the interior many groups one-dimensional signal that 2D signal transformed of the certain neighborhood of picture point has the signal more than specified quantity that 4 chequered with black and white zones are arranged, this picture point is detected as the angle point of gridiron pattern pattern.
The coupling of described gridiron pattern pattern, be specially: the result points of Corner Detection will be carried out the operation of denoising and cancellation redundant points, whether there is around the angle point angle point of setting quantity to determine that whether this put noise point in the setting regions by detecting, by the angle point of phase mutual edge distance less than setpoint distance is divided in the same subclass, use the barycenter of subclass to replace whole angle point subclass to eliminate redundant points, definition angle point vector is an angle point zone boundary vector to the difference of its corresponding zone boundary location point vector, if the angle point zone boundary vector angle of two different angle points in setting range near 180 ° or-180 °, and distance is less than or equal to the gridiron pattern grid area size of determining according to picture size and gridiron pattern ranks number between two angle points, think that then two angle points can be connected, after all angle points are connected test, obtain several connected components of whole angle point set, calculate true angle point number in the pattern according to the ranks number of known gridiron pattern pattern, the connected component that will have true several angle points of angle point is considered as the angle point set of gridiron pattern pattern.
3, gridiron pattern pattern recognition locating method under complex illumination according to claim 2 and the surface condition, it is characterized in that, in the detection of described regional angle point, used annular morphological operation, promptly for an one-dimensional signal that is transformed by 2D signal originally, when carrying out morphological operation, regard the head and the tail point of one-dimensional signal as on the geometric position adjacent, be about to one-dimensional signal and regard an end to end ring as and carry out morphological operation.
4, gridiron pattern pattern recognition locating method under complex illumination according to claim 2 and the surface condition, it is characterized in that, before the detection of carrying out described regional angle point, carry out the image pre-service, be about to image and keep length breadth ratio to zoom in the resolving range of common video camera.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100371044A CN101271576A (en) | 2008-05-08 | 2008-05-08 | Gridiron pattern recognition locating method under complex illumination and surface condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008100371044A CN101271576A (en) | 2008-05-08 | 2008-05-08 | Gridiron pattern recognition locating method under complex illumination and surface condition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101271576A true CN101271576A (en) | 2008-09-24 |
Family
ID=40005528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008100371044A Pending CN101271576A (en) | 2008-05-08 | 2008-05-08 | Gridiron pattern recognition locating method under complex illumination and surface condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101271576A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102184544A (en) * | 2011-05-18 | 2011-09-14 | 北京联合大学生物化学工程学院 | Method for correcting deformity and identifying image of go notation |
| CN103489192A (en) * | 2013-09-30 | 2014-01-01 | 北京林业大学 | Method for detecting number of Arabidopsis leaves and distance between cusp and center of mass of each leaf |
| CN104036244A (en) * | 2014-06-09 | 2014-09-10 | 西安邮电大学 | Checkerboard pattern corner point detecting method and device applicable to low-quality images |
| CN105894518A (en) * | 2016-04-22 | 2016-08-24 | 安徽师范大学 | Cross-stitch template identification method based on lattice point arrangement |
| CN107358221A (en) * | 2017-08-08 | 2017-11-17 | 大连万和海拓文化体育产业有限公司 | The chessboard localization method of spectrum is remembered in a kind of go based on video identification technology automatically |
| CN107480678A (en) * | 2017-09-29 | 2017-12-15 | 北京深度奇点科技有限公司 | A kind of chessboard recognition methods and identifying system |
| CN108955573A (en) * | 2018-06-05 | 2018-12-07 | 天津大学 | A kind of lossless bearing calibration of the order of coded structured light |
| CN109859226A (en) * | 2019-01-10 | 2019-06-07 | 上海理工大学 | A kind of detection method of the X-comers sub-pix of figure segmentation |
| CN110717920A (en) * | 2019-09-03 | 2020-01-21 | 歌尔股份有限公司 | Method and device for extracting target image of projector galvanometer test and electronic equipment |
| CN112465760A (en) * | 2020-11-19 | 2021-03-09 | 深圳惠牛科技有限公司 | Checkerboard corner point identification method, device, equipment and storage medium |
| WO2022204864A1 (en) * | 2021-03-29 | 2022-10-06 | Harman International Industries, Incorporated | A method for corner detection on unified calibration board between dvs and camera |
-
2008
- 2008-05-08 CN CNA2008100371044A patent/CN101271576A/en active Pending
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102184544A (en) * | 2011-05-18 | 2011-09-14 | 北京联合大学生物化学工程学院 | Method for correcting deformity and identifying image of go notation |
| CN103489192A (en) * | 2013-09-30 | 2014-01-01 | 北京林业大学 | Method for detecting number of Arabidopsis leaves and distance between cusp and center of mass of each leaf |
| CN103489192B (en) * | 2013-09-30 | 2017-01-11 | 北京林业大学 | Method for detecting number of Arabidopsis leaves and distance between cusp and center of mass of each leaf |
| CN104036244A (en) * | 2014-06-09 | 2014-09-10 | 西安邮电大学 | Checkerboard pattern corner point detecting method and device applicable to low-quality images |
| CN104036244B (en) * | 2014-06-09 | 2017-05-10 | 西安邮电大学 | Checkerboard pattern corner point detecting method and device applicable to low-quality images |
| CN105894518A (en) * | 2016-04-22 | 2016-08-24 | 安徽师范大学 | Cross-stitch template identification method based on lattice point arrangement |
| CN105894518B (en) * | 2016-04-22 | 2019-04-16 | 安徽师范大学 | A kind of cross-stitch template recognition methods based on lattice point arrangement |
| CN107358221B (en) * | 2017-08-08 | 2020-10-09 | 大连万和海拓文化体育产业有限公司 | Video recognition technology-based chessboard positioning method for automatically recording chess manual of go |
| CN107358221A (en) * | 2017-08-08 | 2017-11-17 | 大连万和海拓文化体育产业有限公司 | The chessboard localization method of spectrum is remembered in a kind of go based on video identification technology automatically |
| CN107480678A (en) * | 2017-09-29 | 2017-12-15 | 北京深度奇点科技有限公司 | A kind of chessboard recognition methods and identifying system |
| CN108955573A (en) * | 2018-06-05 | 2018-12-07 | 天津大学 | A kind of lossless bearing calibration of the order of coded structured light |
| CN108955573B (en) * | 2018-06-05 | 2020-03-06 | 天津大学 | Order lossless correction method of coded structured light |
| CN109859226A (en) * | 2019-01-10 | 2019-06-07 | 上海理工大学 | A kind of detection method of the X-comers sub-pix of figure segmentation |
| CN109859226B (en) * | 2019-01-10 | 2022-06-17 | 上海理工大学 | Detection method of checkerboard corner sub-pixels for graph segmentation |
| CN110717920A (en) * | 2019-09-03 | 2020-01-21 | 歌尔股份有限公司 | Method and device for extracting target image of projector galvanometer test and electronic equipment |
| CN110717920B (en) * | 2019-09-03 | 2022-06-07 | 歌尔光学科技有限公司 | Method and device for extracting target image of projector galvanometer test and electronic equipment |
| CN112465760A (en) * | 2020-11-19 | 2021-03-09 | 深圳惠牛科技有限公司 | Checkerboard corner point identification method, device, equipment and storage medium |
| WO2022204864A1 (en) * | 2021-03-29 | 2022-10-06 | Harman International Industries, Incorporated | A method for corner detection on unified calibration board between dvs and camera |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101271576A (en) | Gridiron pattern recognition locating method under complex illumination and surface condition | |
| WO2021138995A1 (en) | Fully automatic detection method for checkerboard corners | |
| US7019713B2 (en) | Methods and measurement engine for aligning multi-projector display systems | |
| CN106856003B (en) | The expansion bearing calibration of shaft-like workpiece side surface defects detection image | |
| WO2017092631A1 (en) | Image distortion correction method for fisheye image, and calibration method for fisheye camera | |
| CN103048331A (en) | Printing defect detection method based on flexible template registration | |
| CN110569857B (en) | Image contour corner detection method based on centroid distance calculation | |
| CN105701827A (en) | Method and device for jointly calibrating parameters of visible light camera and infrared camera | |
| TWI500925B (en) | Check the device, check the method and check the program | |
| Bu et al. | Concentric circle grids for camera calibration with considering lens distortion | |
| Borrmann et al. | Mutual calibration for 3D thermal mapping | |
| CN103795935B (en) | A kind of camera shooting type multi-target orientation method and device based on image rectification | |
| US9946947B2 (en) | System and method for finding saddle point-like structures in an image and determining information from the same | |
| KR101759798B1 (en) | Method, device and system for generating an indoor two dimensional plan view image | |
| CN103106661A (en) | Solving parabolic catadioptric camera parameters through two intersected straight lines in space | |
| CN102156986A (en) | Calibration method of near-infrared camera | |
| CN103824298A (en) | Intelligent body visual and three-dimensional positioning method based on double cameras and intelligent body visual and three-dimensional positioning device based on double cameras | |
| CN109325381A (en) | The positioning of QR code and correcting algorithm at a kind of view finding pattern lacks one | |
| CN113538583A (en) | Method for accurately positioning position of workpiece on machine tool and vision system | |
| CN111707187A (en) | Measuring method and system for large part | |
| Xing et al. | A new calibration technique for multi-camera systems of limited overlapping field-of-views | |
| CN103942784A (en) | Method for solving parameters in parabolic catadioptric camera through three unparallel straight lines in checkerboard | |
| Zhang et al. | A new algorithm for accurate and automatic chessboard corner detection | |
| CN114187363A (en) | Method and device for obtaining radial distortion parameter value and mobile terminal | |
| JP5274173B2 (en) | Vehicle inspection device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20080924 |