CN109827502A - A kind of line structured light vision sensor high-precision calibrating method of calibration point image compensation - Google Patents
A kind of line structured light vision sensor high-precision calibrating method of calibration point image compensation Download PDFInfo
- Publication number
- CN109827502A CN109827502A CN201811619300.2A CN201811619300A CN109827502A CN 109827502 A CN109827502 A CN 109827502A CN 201811619300 A CN201811619300 A CN 201811619300A CN 109827502 A CN109827502 A CN 109827502A
- Authority
- CN
- China
- Prior art keywords
- point
- striation
- calibration
- target
- coordinate
- 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.)
- Granted
Links
Abstract
The present invention relates to a kind of line structured light vision sensor high-precision calibrating methods of calibration point image compensation, comprising: demarcates to the camera intrinsic parameter in structured light vision sensor；Using the planar metal target for being inlaid with LED luminescence feature point, video camera shooting has the plane target drone image of striation；Extract target characteristic point and striation calibration point coordinate；It calculates separately target characteristic point coordinate and striation demarcates point location uncertainty；It is constraint with positioning feature point uncertainty, the deviations of all characteristic points is solved by nonlinear optimization mode；Compensate target characteristic point and striation calibration point coordinate；More than twice by target movement, three-dimensional coordinate of all striation calibration points under camera coordinates is obtained, is fitted these three-dimensional coordinate points and solves optic plane equations, completes calibration；The present invention is suitble to complicated light environment at the scene, in addition in larger picture noise or positioning feature point deviation still achievable cable architecture visual sensor high-precision calibrating.
Description
Technical field
The present invention relates to the technical fields of transducer calibration, and in particular to a kind of line-structured light view of calibration point image compensation
Feel sensor high-precision calibrating method.
Background technique
Line structured light vision sensor is as important three-dimensional data obtaining means, big, the non-contact, speed with range
Fastly, the advantages that precision is high is becoming widely adopted in on-line dynamic measurement field.Such as detector for train wheel pair size on-line dynamic measurement,
Pantograph wears away on-line measurement and the detection of train Car Body On-line etc..The complicated multiplicity of above-mentioned in-site measurement environment, sensor mark
Fixed and measurement is easy to be influenced by factors such as sensor placement, ambients, it is difficult to meet ideal calibration condition, calibration essence
Spend it is low, this become restrict measurement accuracy problem.Currently, the calibration method based on plane target drone it is high with its precision, at
The advantages that this is low, flexible and convenient is widely used in field calibration.But its stated accuracy is still easy by field calibration environment shadow
It rings, so that image characteristic point or striation calibration point are difficult to meet optimal imaging simultaneously, easily generates deviations.And rely on figure
As processing method can only reduce deviations, can not but eliminate at all, this cannot achieve this method under live complex environment
Calibration.Therefore study one kind is not influenced by picture noise, and can be realized the method demarcated under complicated light condition becomes structure
The new direction of light vision sensor development.
Line structured light vision sensor calibration includes that calibration of camera internal parameters and light-plane parameters demarcate two large divisions.For a long time
Since, the research of camera internal reference calibration is relatively more, wherein there are many research about Calibration of camera intrinsic parameters this respect,
Therefore light-plane parameters calibration process is discussed.Currently have much about the scaling method of optic plane equations.Dewar R. exists
Article " Self-generated targets for spatial calibration of structured light
Optical sectioning sensors with respect to an external coordinate system " is proposed
Wire drawing scaling method is aimed at because bright spot intrinsic brightness is unevenly distributed or highlights phenomena such as reflective with measuring device in space
Bright spot is difficult stringent corresponding with the bright spot in image, therefore the obtained calibration point of this method is few and stated accuracy is lower；In article
Cross ratio invariability is proposed in " Caibration a structured light stripe system:a novel approach "
Theory puts at least three collinear points of known coordinate by three-dimensional target, using cross ratio invariability obtain structural light strip and this
The coordinate of the intersection point of straight line where knowing at 3 points.This method can obtain the calibration point on the optical plane of degree of precision, be suitble to scene mark
It is fixed.But the high-precision three-dimensional target for needing at least two orthogonal planes to constitute.Simultaneously as to illumination between plane
It mutually blocks, it is difficult to obtain the uncalibrated image of high quality, also limit calibration point quantity.Liu et al. is in article " novel
calibration method for multi-sensor visual measurement system based on
Structured light " proposes the method for indicating striation straight line using plucker ' s equation, with the side using less calibration point
Method is compared, and this method can effectively improve stated accuracy.Wei et al. is in article " a novel 1D target-based
calibration method with unknown orientation for structured light vision
Sensor " proposes a kind of structured light vision sensor scaling method based on 1-dimension drone.Using between 1-dimension drone characteristic point
Distance solves the three-dimensional coordinate of the intersection point of optical plane and 1-dimension drone.Optic plane equations are obtained by being fitted multiple intersection points.In recent years
Come, using space geometry as auxiliary constraint, under complex environment on site, the structured light vision sensor with optical filter
Calibration.Liu et al. is in article " calibration method for line-structured light vision sensor
Based a single ball target " proposes the structured light vision sensor scaling method based on ball target.This method
It needs to extract ball target outer profile edge, and then obtains orientation of the ball target under camera coordinates system.In conjunction with optical plane and ball target
Mark intersection obtains conical profile and solves optic plane equations.This method has the ball target profile obtained not by target placement angle
Influence, but need to extract the outer profile of target, be easy the interference by background or the external world.And in " An on-site
It is mentioned in calibration of line-structured light vision in complex light environment "
Parallel bicylindrical target is utilized out, in the case where camera assembles optical filter and external environment is complicated, realizes that structure optical parameter is existing
Field calibration.Intersect corresponding relationship between the parallel elliptic plane in space and the plane of delineation generated in bicylindrical using optical plane, with
Cylindrical radius is equal to a length of constraint of space ellipse short axle, advanced optimizes to obtain optic plane equations.Meanwhile there is scholar to propose
The method for improving stated accuracy, if week et al. is in article " complete calibration of a structured light
One kind is proposed in stripe vison sensor through planar target of unknown orientations "
Light-plane parameters scaling method based on plane target drone.The calibration point in optical plane is obtained by Cross ration invariability, leans on flat target
Indicated weight moves the calibration point three-dimensional coordinate obtained in optical plane again, and fitting obtains optic plane equations.This method is at low cost with its,
Flexibly, the advantages that precision is high, be widely used field of high-precision measurement at the scene.But it still can not get rid of picture noise and draw
Calibrated error caused by the deviations risen, cannot achieve the calibration of higher precision.
Current scaling method is analyzed, is all true as actual imaging point coordinate to extract target characteristic point or striation calibration point
Value solves to obtain optimal solution by minimizing projecting characteristic points error again after obtaining linear solution.Meanwhile also as far as possible using calibration
Space is the calibration mode for measuring space, however inevitably due to complicated light, laser matter when field calibration
Amount, target surface process roughness, target placement angle, and image focus is fuzzy, picture noise, and characteristic point and striation extract inclined
The factors such as difference influence, and reduce stated accuracy.If only looking after target characteristic point image quality when field calibration, striation is easily caused
Brightness decline or thicker；If phenomena such as only treatment striation is imaged, and target feature is easy to appear out of focus fuzzy or under-exposure, two
It is in contradiction shape between person, optimal imaging quality can not be met simultaneously.Therefore the outdoor complex condition in scene, general, no is studied
The structured light vision sensor high-precision calibrating method interfered by picture noise becomes problem urgently to be resolved.
Summary of the invention
The technology of the present invention solves the problems, such as: overcoming the deficiencies of the prior art and provide a kind of cable architecture of calibration point image compensation
Light vision sensor high-precision calibrating method, can at the scene in the case where complicated light environment it is fuzzy especially in the presence of image,
High-precision calibrating is realized in the case of out of focus, noise jamming.
In order to achieve the above objectives, the technical scheme of the present invention is realized as follows:
A kind of line structured light vision sensor high-precision calibrating method of calibration point image compensation, this method comprises:
A, in the case where being not switched on laser, the video camera in line structured light vision sensor is demarcated；
B, using the planar metal target for being inlaid with LED luminescence feature point, video camera shooting has the plane target drone of striation
Image；Extract target characteristic point and striation calibration point coordinate；
C, it calculates separately target characteristic point coordinate and striation demarcates point location uncertainty.With positioning feature point uncertainty
For constraint, the deviations of all characteristic points are solved by nonlinear optimization mode and compensation obtains accurate feature points coordinate；
D, three-dimensional coordinate of all striation calibration points under camera coordinates more than twice by target movement, is obtained, is passed through
Fitting solves optic plane equations after RANSAC picks impurity point.
In step a in the case where not opening laser, the video camera in line structured light vision sensor is demarcated,
Using Liu Zhen et al. improved Zhang Zhengyou scaling method calibrating camera inner parameter proposed and camera lens second order radial distortion system
Number；
The image that striation intersects with plane target drone is shot in step b, and extracts target characteristic point and striation calibration point respectively
Image coordinate, the method is as follows:
(b1) adjustment metal target intersects with space optical plane, guarantees that striation does not pass through the target luminescence feature in space
Point；
(b2) it takes multiple dimensioned optical spot centre coordinate method to extract target characteristic point image coordinate to sit as target characteristic point
Mark initial value；Optical losses point coordinate is extracted, being fitted the target feature point list vertical with striation direction is straight line, and solves it
And striation crosspoint is as striation calibration point coordinate initial value.
Target characteristic point is solved in step c, and the specific method is as follows with striation calibration point location uncertainty:
(c1) acquisition image is handled respectively using multiple Gaussian convolution cores, solves multiple positioning of each characteristic point
Coordinate, statistics obtain the positioning uncertainty of each characteristic point；
(c2) the positioning uncertainty mathematical model for establishing each striation calibration point topography, using mean filter method
Picture noise is solved, and then obtains the positioning uncertainty of each striation calibration point.
Target characteristic point is solved by nonlinear optimization method to constrain with positioning feature point uncertainty in step c
With striation calibration point deviations, accurate feature points coordinate is obtained after overcompensation.
The method of last transducer calibration is as follows in step d:
(d1) it is based on compensated striation calibration point coordinate, striation calibration point is obtained using plane target drone scaling method and is being taken the photograph
Three-dimensional coordinate under camera coordinate system；
(d2) all striation calibration points are being obtained after the three-dimensional coordinate under camera coordinates, is being rejected using RANSAC method
Fitting solves optic plane equations initial value after miscellaneous point；
(d3) optic plane equations optimal solution is acquired using Levenberg-Marquardt nonlinear optimization method, completes line
The calibration of structured light vision sensor.
The advantages of the present invention over the prior art are that:
The present invention proposes a kind of structured light vision sensor scaling method based on uncertainty model.Initially set up target
The positioning uncertainty model of characteristic point, striation calibration point, solves the positioning uncertainty of each characteristic point；Secondly with spy
Levying point location uncertainty is constraint, with striation calibration point back projection to space target and minimum about with target plan range
Shu Jianli target equation, the deviations of each characteristic point are acquired by nonlinear optimization method；Finally after overcompensation, then
It substitutes into plane target drone scaling method and solves high-precision optic plane equations.The present invention can effectively make up target characteristic point and light
Bring deviations when calibration point extracts, be particularly suitable for solving image quality in live complex environment it is bad cause or
Stated accuracy caused by the influence such as out of focus, picture noise declines problem, improves the precision of calibration.
Detailed description of the invention
Fig. 1 is that the present invention is based on the structured light vision sensor high-precision calibrating method flow charts of uncertainty model；
Fig. 2 is that line-structured light visual sensing of the present invention demarcates schematic diagram；
Fig. 3 is that target characteristic point of the present invention and striation demarcate point location uncertainty schematic diagram.
Specific embodiment
The basic idea of the invention is that: target characteristic point and striation calibration point coordinate initial value are extracted, and determines all features
Point location uncertainty.It is constraint with positioning feature point uncertainty, each characteristic point is obtained by nonlinear optimization mode
Deviations.Plane target drone scaling method is substituted into after compensated again, show that striation calibration point is three-dimensional under camera coordinate system
Coordinate.Impurity point is picked using RANSAC method, and is accurately solved by Optimization Method exit plane equation, realizes structure light
The high-precision calibrating of visual sensor.
Below by taking the line structured light vision sensor that a video camera and a laser line generator form as an example, the present invention is made
It is further described.
As shown in Figure 1, the structured light vision sensor high-precision calibrating method the present invention is based on uncertainty model is main
The following steps are included:
Step 11: in the case where laser line generator is not opened, the video camera in line structured light vision sensor being marked
It is fixed.
Here the video camera of visual sensor is demarcated, that is, solves the inner parameter of video camera, specific scaling method
In article [Liu Z, Wu Q, Chen X, the et al.High- for the improved Zhang Zhengyou calibration method that Liu Zhen et al. is proposed
accuracy calibration of low-cost camera using image disturbance factor[J]
.Optics Express, 2016,24 (21): 24321-24336.] in have a detailed description.
Step 12: open laser, put plane target drone immediately ahead of camera so that laser line generator projection optical plane and
Plane target drone intersection, video camera shooting have the plane target drone image of striation.
As shown in Fig. 2, setting OcxcyczcFor camera coordinates system, OtxtytztFor target co-ordinates system, YtFor target plane, Y is sharp
Optical plane, LiFor Y and YtIntersection, liFor LiImaging, Q1j、Q2j、Q3jIndicate three luminous points on target previous column,For target
Punctuate line and LiIntersection point, for subsequent optic plane equations solve.p1j p2j p3jRespectively Q1j、Q2j、Q3jCorresponding image
Point,ForCorresponding picture point is defined as striation calibration point.Optic plane equations are represented by ax+by+cz+d=0,
In
Step 13: extracting characteristic point image coordinate, i.e. target characteristic point and striation calibration point coordinate.
Here, specifically includes the following steps:
Step 131: extract shooting image in target calibration point image coordinate, by multiple dimensioned extracting method (such as text
Chapter " high accuracy positioning [J] the optical precision engineering at the multiple dimensioned dot pattern picture center Liu Zhen, Shang Yanna, 2013,21 (6):
Mentioned in 1586-1591. "), obtain all optimal image coordinates of target characteristic point in image.
Using Steger " Steger C.Unbiased Extraction of Curvilinear Structures
The center of the extraction striation of method described in from 2D and 3D Images [J] .1998. ".
Step 14: calculating target characteristic point and striation demarcates point location uncertainty.
Step 141: as shown in figure 3, the present invention takes multiple dimensioned extracting method to complete target positioning feature point initial value and not
Degree of certainty solves.Image is handled by multiple different Gaussian convolutions, uses statistical method to obtain target after repeatedly extracting central point
Positioning feature point uncertainty, while choosing the corresponding coordinate of best scale and being characterized dot center's coordinate initial value.Choose m not
Same Gaussian convolution verification is respectively handled j characteristic point at the position i, obtains m group target characteristic point coordinate pijAnd group
At point set Pm, such as cross crunode red in the characteristic point in Fig. 2.And then target characteristic point point set is found out in u, the direction v is averaged
CoordinateWith standard deviation sigmau、σv.And by σu、σvFor the positioning uncertainty of target characteristic point.
Step 142: the influence that stimulated light device quality, power, target surface material and ambient light are shone when field calibration, light
Central point, which extracts, to be easy to be influenced by picture noise to generate deviations.The present invention provides under any direction striation calibration point not
Degree of certainty method for solving.The following are striation calibration point coordinate uncertainty solution procedurees.
If obtaining partial derivative after image and Gauss nuclear convolution is respectively gu、gv、guu、guv、gvv.By calculating Hessian square
Battle array, obtains striation normal line vector n (u, v).If (u0,v0) it is image Point Coordinates, striation edge direction (nu,nv) indicate,
And | | (nu,nv) | |=1, orthogonal direction is usedIt indicates.Therefore, striation cross section curve can be along edge direction (nu,
nv) be expressed as,
For line edge, enableIt can obtain:
Therefore, the maximum point of image grayscale namely optical losses point coordinate are (pu,pv)=((tnu+u0),(tnv+
v0))。
With the corresponding optical losses point (0,0) of noise-free picture for origin,With (nu,nv) it is that reference axis establishes o-
Rc coordinate system, if (nu,nv) grey scale curve is h (c, r) on direction, therefore solves (pu,pv) uncertainty be converted into solve h
(c,r)R=0Locate c0Uncertainty.If h (c, r)=I (c, r)+N (c, r), wherein I (c, r) is ideal image, and N (c, r) is equal
Value is zero, and variance isPicture noise.By [CSteger] it is found that h (c0,r0) at first derivative be zero, then:
Wherein,Respectively intensity profile, ideal image distribution, picture noise are by varianceGauss
Data after nuclear convolution,Respectively
For in (c0,r0) at single order, second-order partial differential coefficient to c-axis direction.It is rightTaylor expansion is carried out at (0,0),
Since ideal image can be expressed asM is gray scale maximum value, σwFor Gaussian kernel, and meetAndIt acquires:
Corresponding varianceFor,
Location of the core variance is obtained by formula 8,9For,
Formula (11) substitution formula (10), which is obtained center point coordinate positioning variances, is,
Wherein, σwIt can be by taking ncGrey scale curve is obtained using fittype () Function Fitting in matlab on direction, wherein
Fitting function prototype is
It willIt decomposes under o-uv coordinate system, it can be deduced that striation calibration point point partial uncertainty
Step 15: being constraint with characteristic point uncertainty, together based on the positioning feature point uncertainty determined in step 14
Back projection's error is minimum between Shi Jianli target feature point image and space target, striation calibration point back projection to space target
Plane and point arrive the minimum target equation of plan range, the deviations of characteristic point are obtained by nonlinear optimization method.
Step 151: decomposing target characteristic point imaging process, establish perspective projection point in characteristic point, distortion point and finally make an uproar
Transformation equation between sound point.
If pu=[uu,vu,1]T, pd=[ud,vd,1]TWith pn=[un,vn,1]TIt is respectively undistorted under image coordinate system
Point, distortion point and actual imaging point coordinate.The imaging of space target point Q=[x, y, 1] can divide it can be seen from imaging optical path
Solution is three processes, first be perspective projection model, second be lens distortion model, third be picture noise superposition
Model；Wherein perspective projection model may be expressed as:
Wherein H3×3For the homography matrix between target plane and the plane of delineation, ρ is constant factor, and K is camera intrinsic parameter square
Battle array, u0With v0It is the obliquity factor .r that principal point coordinate γ is image u v coordinate axis1、r2, t be respectively spin matrix first two columns
And translation vector.Lens distortion model may be expressed as:
Whereink1, k2Two rank distortion factors, [x before camera lensn,yn] indicate normalized image coordinate.Root
According to practical experience, preceding two ranks radial distortion precisely enough describes lens distortion, reaches degree of precision.If due to image
Image deviations caused by the reasons such as noise are Δ u, Δ v, then distortion point may be expressed as: to actual imaging point
Under i-th of placement position of target, if j-th point of the target homogeneous coordinates under target co-ordinates system and image coordinate system
Respectively Qj=[xj,yj,1]TWithpijBy formula (14), (15) calculate pu(ij), pu(ij)With Qj=[xj,
yj,1]TH is solved by formula 13iMatrix, wherein Δ u in formula (15)ij,ΔvijInitial value be 0.
Step 152: establishing back projection's error minimum between target feature point image and space target, striation calibration point is counter to be thrown
To space target plane and point arrives the minimum target equation of plan range, obtains determining for characteristic point by nonlinear optimization method
Position deviation.
According to the mapping matrix H of i-th of position target plane to the plane of delineationi, QjIt is obtained j-th of target by formula (13)
The homogeneous coordinates p of characteristic point subpoint under image coordinate systemn(ij).With pijWith pn(ij)Between in distance minimum and image statistics
To establish first aim function as follows apart from minimum constraint for heart point:
Wherein D (pij,pn(ij)) indicate pijWith pn(ij)Distance, M indicate that target placement position number, N indicate target characteristic point
Number.
P is calculated by formula (13)ijThe homogeneous coordinates put under back projection to target co-ordinates systemWith qjWithBetween distance minimum and target point and the subpoint statistics minimum objective function of centre distance establish second target function:
Cross ratio invariability is constrained as the important restrictions condition in projective geometry, can be preferably excellent by target rigid constraint
Change coordinate between image characteristic point, is widely used in the links such as camera intrinsic parameter calibration.The present invention is by target feature point image
Double ratio constraint is established between coordinate and target, and it is as follows to establish third objective function:
CR in formulah、CRvRespectively indicate horizontal and vertical direction double ratio.To reach the intensity of constraint, while improving calculating effect
Rate, horizontal direction, which is appointed, takes four points to calculate double ratio, if shared K kind combination；Vertical direction, which is equally appointed, takes four points to calculate double ratio,
If shared L kind combination, K and L specific value are determined by points, guarantee that target characteristic point uniformly covers entire image planes.
If striation calibration pointForward projection and target intersect at space three-dimensional pointAnd and pijIt is thrown under target co-ordinates system
The homogeneous coordinates of shadow pointThe target plan range of fitting is minimum, establishes the 4th target to distance minimum between plane with point
Function is as follows:
Wherein FiIndicate that target puts the plane equation of i-th of position target point fitting,Indicate point to plane it
Between distance.
Meanwhile ifMiddle kth column fitting a straight line isFitting a straight line isWithIntersection point isWithWith
Between distance minimum to establish the 5th objective function as follows:
Combining four objective functions can obtain:
E (a)=e1+e2+e3+e4+e5 (21)
For Δ uij,Δvij,It joined optimization range constraint, as shown in Equation 22:
Whereinσu(ij)And σv(ij)It is put for i-th of target
Jth point positioning feature point uncertainty in the picture is put at position,For k-th of striation at i position of target
Demarcate point location uncertainty.N is non-zero proportions coefficient, and according to a large amount of repetition tests, the present invention is set as 9.
Step 16: the deviations of each characteristic point being obtained based on the nonlinear optimization in step 15, are obtained after compensated
Accurate profile point location coordinate, by going distortion to obtain undistorted coordinate.
Step 17: striation calibration point is mapped to three-dimensional space by the homography matrix determined by target characteristic point, and it is flat to obtain light
Face three-dimensional coordinate point list.Gross error point is rejected using RANSAC, least square method is recycled to obtain optical plane initial value.Most
The maximum likelihood solution of optical plane is obtained by nonlinear optimization afterwards.
IfFor the target characteristic point coordinate after removal distortion, QijFor target characteristic point coordinate,For target plane and figure
As the homography matrix between plane；Striation calibration point coordinate after distorting for removal,It is sat for striation calibration point in target
Mooring points coordinate is marked,For optical plane and target intersection three-dimensional point coordinate under camera coordinates.Then between picture point and target point
MeetWherein s is nonzero coefficient.The invertibity mapped according to homography matrix, it can be deduced that:
IfIt can be analyzed to spin matrixWith translation vectorThen
If optic plane equations are expressed as ax+by+cz+d=0, whereinIt will
It substitutes into the least square method of RANSAC constraint, obtains optic plane equations and just solvePlane is arrived based on point on optical plane
Apart from minimum constraint, following objective function can be established,
Wherein, a, b, c, d are four coefficients of optic plane equations, xik=[xik,yik,zik, 1] and it is i target position, k table
Show that optical plane intersects resulting k calibration point with target on striation.S indicates that target puts number, and M indicates that each position obtains
Striation calibration point number.The optimal solution of a, b, c, d can be obtained by the maximal possibility estimation of nonlinear optimization method.
Claims (6)
1. a kind of line structured light vision sensor high-precision calibrating method of calibration point image compensation, which is characterized in that realize step
It is rapid as follows:
Step a, in the case where being not switched on laser, the video camera in line structured light vision sensor is demarcated；
Step b, it using the planar metal target for being inlaid with LED luminescence feature point, has been beaten and has been swashed using the video camera shooting demarcated
The planar metal target image for being inlaid with LED luminescence feature point of light striation；Calculate all target LED characteristic points in the picture
The coordinate initial value of coordinate and all striation calibration points；
Step c, solve target characteristic point and striation calibration point uncertainty, and using uncertainty solve target characteristic point and
The deviations of striation calibration point and compensation obtains characteristic point coordinate；
Step d, planar metal target is mobile more than twice, obtain three-dimensional seat of all striation calibration points under camera coordinates
Mark, fitting solves optic plane equations after picking impurity point, completes the calibration of line structured light vision sensor.
2. a kind of line structured light vision sensor high-precision calibrating side of calibration point image compensation according to claim 1
Method, it is characterised in that: in step a in the case where not opening laser condition, the video camera in line structured light vision sensor is carried out
Calibration obtains intrinsic parameters of the camera and camera lens second order coefficient of radial distortion.
3. a kind of line structured light vision sensor high-precision calibrating side of calibration point image compensation according to claim 1
Method, it is characterised in that: shoot the image that striation intersects with plane target drone in step b, calculate the image coordinate of target characteristic point, benefit
Method with the image coordinate initial value of the image coordinate calculating striation calibration point of target characteristic point is as follows:
(b1) the camera lens second order distortion factor obtained using calibration carries out distortion correction to shooting image；
(b2) adjustment planar metal target intersects with space optical plane, guarantees that laser striation does not pass through the target feature in space
Point；
(b3) multiple dimensioned optical spot centre coordinate method is taken to extract the image coordinate of target characteristic point as target characteristic point coordinate
Initial value；Optical losses point coordinate is extracted, the point range straight line of the target characteristic point vertical with striation direction is fitted, and solves the point
Column straight line and striation crosspoint are as striation calibration point coordinate initial value.
4. a kind of line structured light vision sensor high-precision calibrating side of calibration point image compensation according to claim 1
Method, it is characterised in that: target characteristic point is solved in step c, and the specific method is as follows with striation calibration point location uncertainty:
(c1) uncalibrated image is handled respectively using multiple Gaussian convolution cores, solves multiple positioning of each characteristic point respectively
Coordinate, statistics obtain the positioning uncertainty of each characteristic point；
(c2) the positioning uncertainty mathematical model of each striation calibration point topography is established, picture noise is solved and then is obtained
The positioning uncertainty of each striation calibration point.
5. a kind of line structured light vision sensor high-precision calibrating side of calibration point image compensation according to claim 1
Method, it is characterised in that: constrained in step c based on positioning feature point uncertainty, compensate target characteristic point and striation calibration point
Coordinate；
(c1) it is constraint with target characteristic point and the uncertainty of striation anchor point, by the method for nonlinear optimization, solves target
Mark the deviations of characteristic point and striation calibration point；
(c2) with the coordinate of deviations compensation target characteristic point and striation calibration point.
6. a kind of line structured light vision sensor high-precision calibrating side of calibration point image compensation according to claim 1
Method, it is characterised in that: the scaling method of step d centerline construction light vision sensor is as follows:
(d1) it is based on compensated striation calibration point coordinate, using plane target drone scaling method, acquires video camera using calibration
Inner parameter obtains three-dimensional coordinate of the striation calibration point under camera coordinate system；
(d2) all striation calibration points are being obtained after the three-dimensional coordinate under camera coordinates, is picking impurity point using RANSAC method
Fitting solves optic plane equations initial value afterwards；
(d3) according to optic plane equations initial value, optical plane side is acquired using Levenberg-Marquardt nonlinear optimization method
Journey optimal solution completes the calibration of line structured light vision sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811619300.2A CN109827502B (en) | 2018-12-28 | 2018-12-28 | High-precision calibration method for line-structured light vision sensor for calibration point image compensation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811619300.2A CN109827502B (en) | 2018-12-28 | 2018-12-28 | High-precision calibration method for line-structured light vision sensor for calibration point image compensation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109827502A true CN109827502A (en) | 2019-05-31 |
| CN109827502B CN109827502B (en) | 2020-03-17 |
Family
ID=66861331
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811619300.2A Active CN109827502B (en) | 2018-12-28 | 2018-12-28 | High-precision calibration method for line-structured light vision sensor for calibration point image compensation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109827502B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110942434A (en) * | 2019-11-22 | 2020-03-31 | 华兴源创(成都)科技有限公司 | Display compensation system and method of display panel |
| CN111207670A (en) * | 2020-02-27 | 2020-05-29 | 河海大学常州校区 | Line structured light calibration device and method |
| CN111275770A (en) * | 2020-01-20 | 2020-06-12 | 南昌航空大学 | Global calibration method of four-eye stereoscopic vision system based on one-dimensional target rotation motion |
| CN111311686A (en) * | 2020-01-15 | 2020-06-19 | 浙江大学 | Projector out-of-focus correction method based on edge perception |
| CN112229420A (en) * | 2020-08-31 | 2021-01-15 | 南京航空航天大学 | Line laser calibration method for aircraft skin butt seam measurement |
| CN112484746A (en) * | 2020-11-26 | 2021-03-12 | 上海电力大学 | Monocular vision-assisted laser radar odometer method based on ground plane |
| CN112767492A (en) * | 2020-12-25 | 2021-05-07 | 江苏集萃智能光电系统研究所有限公司 | Railway wheel set size detection device and calibration method thereof |
| CN113155057A (en) * | 2021-03-16 | 2021-07-23 | 广西大学 | Line structured light plane calibration method using non-purpose-made target |
| CN113639633A (en) * | 2021-07-26 | 2021-11-12 | 中国航空工业集团公司北京航空精密机械研究所 | Method for aligning angular zero position of clamp in multi-axis vision measuring device |
| WO2022088039A1 (en) * | 2020-10-30 | 2022-05-05 | Harman International Industries, Incorporated | Unified calibration between dvs and camera |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030007159A1 (en) * | 2001-06-27 | 2003-01-09 | Franke Ernest A. | Non-contact apparatus and method for measuring surface profile |
| CN1566906A (en) * | 2003-06-11 | 2005-01-19 | 北京航空航天大学 | Construction optical visual sense transducer calibration method based on plane targets |
| CN101943563A (en) * | 2010-03-26 | 2011-01-12 | 天津大学 | Rapid calibration method of line-structured light vision sensor based on space plane restriction |
| CN104848801A (en) * | 2015-06-05 | 2015-08-19 | 北京航空航天大学 | Line structure light vision sensor calibration method based on parallel bicylindrical target |
| CN106705849A (en) * | 2017-01-25 | 2017-05-24 | 上海新时达电气股份有限公司 | Calibration method of linear-structure optical sensor |
| CN107218904A (en) * | 2017-07-14 | 2017-09-29 | 北京航空航天大学 | A kind of line structured light vision sensor calibration method based on sawtooth target |
| CN107255443A (en) * | 2017-07-14 | 2017-10-17 | 北京航空航天大学 | Binocular vision sensor field calibration method and device under a kind of complex environment |
-
2018
- 2018-12-28 CN CN201811619300.2A patent/CN109827502B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030007159A1 (en) * | 2001-06-27 | 2003-01-09 | Franke Ernest A. | Non-contact apparatus and method for measuring surface profile |
| CN1566906A (en) * | 2003-06-11 | 2005-01-19 | 北京航空航天大学 | Construction optical visual sense transducer calibration method based on plane targets |
| CN101943563A (en) * | 2010-03-26 | 2011-01-12 | 天津大学 | Rapid calibration method of line-structured light vision sensor based on space plane restriction |
| CN104848801A (en) * | 2015-06-05 | 2015-08-19 | 北京航空航天大学 | Line structure light vision sensor calibration method based on parallel bicylindrical target |
| CN106705849A (en) * | 2017-01-25 | 2017-05-24 | 上海新时达电气股份有限公司 | Calibration method of linear-structure optical sensor |
| CN107218904A (en) * | 2017-07-14 | 2017-09-29 | 北京航空航天大学 | A kind of line structured light vision sensor calibration method based on sawtooth target |
| CN107255443A (en) * | 2017-07-14 | 2017-10-17 | 北京航空航天大学 | Binocular vision sensor field calibration method and device under a kind of complex environment |
Non-Patent Citations (8)
| Title |
|---|
| ZHEN LIU等: "On-site calibration of line-structured light vision sensor in complex light environments", 《OPTICS EXPRESS》 * |
| ZHENZHONG WEI等: "Calibration Method for Line Structured Light Vision Sensor Based on Vanish Points and Lines", 《2010 INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION》 * |
| ZHENZHONG WEI等: "Line structured light vision sensor calibration using parallel straight lines features", 《OPTIK》 * |
| 刘冲等: "大视场线结构光视觉传感器的现场标定", 《光电工程》 * |
| 刘珂等: "线结构光传感器标定不确定度估计", 《光电工程》 * |
| 刘震等: "一种高精度线结构光视觉传感器现场标定方法", 《光学学报》 * |
| 林娜等: "基于机器人系统的线结构光视觉传感器标定新方法", 《传感器与微系统》 * |
| 邝泳聪，崔亮纯: "基于线纹尺的线结构光视觉传感器标定新方法", 《华南理工大学学报(自然科学版)》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110942434A (en) * | 2019-11-22 | 2020-03-31 | 华兴源创(成都)科技有限公司 | Display compensation system and method of display panel |
| CN110942434B (en) * | 2019-11-22 | 2023-05-05 | 华兴源创(成都)科技有限公司 | Display compensation system and method of display panel |
| CN111311686A (en) * | 2020-01-15 | 2020-06-19 | 浙江大学 | Projector out-of-focus correction method based on edge perception |
| CN111311686B (en) * | 2020-01-15 | 2023-05-02 | 浙江大学 | Projector defocus correction method based on edge perception |
| CN111275770A (en) * | 2020-01-20 | 2020-06-12 | 南昌航空大学 | Global calibration method of four-eye stereoscopic vision system based on one-dimensional target rotation motion |
| CN111207670A (en) * | 2020-02-27 | 2020-05-29 | 河海大学常州校区 | Line structured light calibration device and method |
| CN112229420A (en) * | 2020-08-31 | 2021-01-15 | 南京航空航天大学 | Line laser calibration method for aircraft skin butt seam measurement |
| WO2022088039A1 (en) * | 2020-10-30 | 2022-05-05 | Harman International Industries, Incorporated | Unified calibration between dvs and camera |
| CN112484746A (en) * | 2020-11-26 | 2021-03-12 | 上海电力大学 | Monocular vision-assisted laser radar odometer method based on ground plane |
| CN112767492A (en) * | 2020-12-25 | 2021-05-07 | 江苏集萃智能光电系统研究所有限公司 | Railway wheel set size detection device and calibration method thereof |
| CN113155057A (en) * | 2021-03-16 | 2021-07-23 | 广西大学 | Line structured light plane calibration method using non-purpose-made target |
| CN113639633A (en) * | 2021-07-26 | 2021-11-12 | 中国航空工业集团公司北京航空精密机械研究所 | Method for aligning angular zero position of clamp in multi-axis vision measuring device |
| CN113639633B (en) * | 2021-07-26 | 2023-07-07 | 中国航空工业集团公司北京航空精密机械研究所 | Clamp angular zero alignment method in multi-axis vision measurement device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109827502B (en) | 2020-03-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109827502A (en) | A kind of line structured light vision sensor high-precision calibrating method of calibration point image compensation | |
| CN111414798B (en) | Head posture detection method and system based on RGB-D image | |
| CN105956539B (en) | A kind of Human Height measurement method of application background modeling and Binocular Vision Principle | |
| CN103971353B (en) | Splicing method for measuring image data with large forgings assisted by lasers | |
| CN104484648B (en) | Robot variable visual angle obstacle detection method based on outline identification | |
| CN107255443A (en) | Binocular vision sensor field calibration method and device under a kind of complex environment | |
| CN109598762A (en) | A kind of high-precision binocular camera scaling method | |
| CN105894574B (en) | A kind of binocular three-dimensional reconstruction method | |
| CN108986070B (en) | Rock crack propagation experiment monitoring method based on high-speed video measurement | |
| CN106705849B (en) | Calibrating Technique For The Light-strip Sensors | |
| CN103578088B (en) | A kind of starry sky image processing method | |
| CN110332887A (en) | A kind of monocular vision pose measurement system and method based on characteristic light punctuate | |
| CN111563878B (en) | Space target positioning method | |
| CN108662987B (en) | Calibration method of 2D camera type laser measuring head | |
| CN104266608B (en) | Field calibration device for visual sensor and calibration method | |
| CN103278138A (en) | Method for measuring three-dimensional position and posture of thin component with complex structure | |
| CN106996748A (en) | A kind of wheel footpath measuring method based on binocular vision | |
| CN107356202A (en) | A kind of laser scanning measurement system target sights method automatically | |
| CN110060304B (en) | Method for acquiring three-dimensional information of organism | |
| CN108491810A (en) | Vehicle limit for height method and system based on background modeling and binocular vision | |
| CN107595388A (en) | A kind of near infrared binocular visual stereoscopic matching process based on witch ball mark point | |
| CN109631912A (en) | A kind of deep space spherical object passive ranging method | |
| CN107869954A (en) | A kind of binocular vision volume weight measuring system and its implementation | |
| CN109727291A (en) | A kind of high-precision online calibration method of zoom camera | |
| CN110223355A (en) | A kind of feature mark poiX matching process based on dual epipolar-line constraint |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |