CN106091983A - Comprise the complete scaling method of Vision Measuring System With Structured Light Stripe of scanning direction information - Google Patents

Abstract

The complete scaling method of Vision Measuring System With Structured Light Stripe comprising scanning direction information belongs to the scaling method field of line-structured light visual system, the method is to incorporate the conversion thinking of relative motion coordinate system under the auxiliary of relative motion caliberating device based on line slide rail system, the Light-plane calibration process directly utilizing routine disposably obtains several reverse relative motion uncalibrated images simultaneously including Rhizoma Dioscoreae (peeled) positional information and scanning direction information, thus while simplifying demarcating steps, decrease the operand calculated needed for system solution structure optical plane space equation the most very significantly.The method of the present invention can accurately solve and simulate scanning direction vector equation, and scanning direction can be any direction, and final whole new definition comprises, with establishing, the most complete Vision Measuring System With Structured Light Stripe scaling method determining scan vector step.The present invention has raising and demarcates efficiency and reduce the double effects of calibrated error.

Description

Comprise the complete scaling method of Vision Measuring System With Structured Light Stripe of scanning direction information

Technical field

The invention belongs to the scaling method field of line-structured light visual system, be specifically related to one and comprise scanning direction information The complete scaling method of Vision Measuring System With Structured Light Stripe.

Background technology

Vision measurement technology can be well adapted for new standard that workpiece configurations size detection proposed by modern industry and want Ask, be a kind of contactless profile measurement means having both precision and efficiency.Line-structured light visual system is vision measurement technology One of Typical Representative, its ultimate principle is that to utilize linear laser diode to project spatial attitude on space object to be detected complete Known laser plane, forms projection Rhizoma Dioscoreae (peeled) on a surface of an, and makes this projection Rhizoma Dioscoreae (peeled) move along known scanning direction, By high accuracy video camera, aforementioned projection Rhizoma Dioscoreae (peeled) diverse location in scanning process is shot multiple photos respectively again.Subsequently, Utilize optical losses detection technique, the pixel coordinate of the Rhizoma Dioscoreae (peeled) characteristic point of extract surface；Then utilize and demarcate the knot obtained Structure light visual system parameter, projects to the pixel coordinate of features described above point in camera coordinate system, thus obtains body surface 3D contour feature.

For realizing the Constructed Lighting Vision System parameter of above-mentioned Coordinate Conversion, structured-light system is applied to actual measurement it Before, it needs to be respectively completed camera calibration and Light-plane calibration by the method demarcated.Wherein, camera calibration is to obtain Take intrinsic parameters of the camera, and Light-plane calibration is the space equation for asking for laser plane.

At present, generally use Zhang Zhengyou at its open source literature Zhang Z. (2000) .A in technical field of visual measurement flexible new technique for camera calibration.IEEE Trans.on Pattern Analysis And Machine Intelligence, a kind of camera marking method proposed in 22 (11): 1330-1334. completes to take the photograph This process of camera calibration.

The present inventor is at AFlexible Calibration Method Using the Planar Targetwith a Square Pattern for Line Structured Light VisionSystem.Sun, Qiucheng,et al.A Flexible Calibration Method Using the Planar Target with a Square Pattern for Line Structured Light Vision System.PloS one 9.9:e106911. (SCI) a kind of line-structured light vision system calibration method utilizing square pattern plane template is disclosed, in the document in 2014. In, use the camera marking method of traditional Zhang Zhengyou to achieve camera calibration, and disclose a kind of by extracting such as Fig. 3 Tetra-angular coordinate Corner points of ABCD to the square pattern plane template square pattern shown in Fig. 4 Extraction determines the plane space equation The target under its camera coordinate system of the plane template as shown in Figure 5 The method of plane in camera coordinate frame, in order to determine structure optical plane as shown in Figure 1 Structured light plane space plane equation in the camera coordinate system shown in this patent Fig. 2.Additionally, at this A kind of line-structured light scaling method utilizing square pattern plane template also discloses the line that utilizes as shown in Figure 6 to 8 tie The imaging in cameras photograph of the structure laser projection Rhizoma Dioscoreae (peeled) resolves further and determines structure optical plane Structured The throwing that light plane is formed with the square pattern plane template square pattern under different spaces attitude inclination angle respectively The method for solving of the space equation of shadow intersection line segment EF or E ' F ', finally, by the space equation of EF line segment and E ' F ' line segment Space equation, can try to achieve structure optical plane Structured light plane space equation under camera coordinate system.

But, including a kind of line-structured light utilizing square pattern plane template disclosed in aforementioned the present inventor Including scaling method, the academic discussion relevant to line-structured light vision system calibration is the most all concentrated mainly on structure optical plane Space plane equation determines and in the problem such as the arithmetic accuracy relevant to camera calibration link and efficiency, it is generally only completed The demarcation of video camera and Light-plane calibration two processes.But, calibrated line-structured light visual system will be completed and be applied to During workpiece configurations profile scan and identification, video camera and linear laser diode project and photographic unit as an entirety, Also need to determine that the structure optical plane Structured light plane that this projection is projected with photographic unit is relative separately Produced direction of relative movement when measured three-dimensional body makes scanning motion, to this end, also need to put down laser further The direction of Surface scan is demarcated, to obtain structured-light system scanning direction vector equation under camera coordinate system, thus Just can complete the integral calibrating process of line-structured light visual system.But at present, about determining that line-structured light visual system scans The method in direction is less, conventional method as shown in Figure 8: old sliding by the straight line being approximately parallel to scanning object long axis of body direction It is to utilize the overall projection with line slide rail system and photographic unit 1 and square that rail realizes approximating the method for given scanning direction Shape projection target target 2, wherein, overall projection based on line slide rail system and photographic unit 1 include line slide rail 1-1, slide block 1-2, video camera 1-3 and linear laser diode 1-4；Video camera 1-3 and linear laser diode 1-4 is all fixed on slide block 1-2, slide block 1- 2 are slidably connected with line slide rail 1-1；Rectangular projection target plate 2 is placed in the dynamic coverage of video camera 1-3；Straight line swashs Light device 1-4 and video camera 1-3 towards identical and keep geo-stationary and transport along the x-axis direction being parallel to camera coordinate system Dynamic, the movement angle in this direction is ensured by line slide rail, ensures line slide rail side in scanning process as far as possible Parallel to the x-axis with camera coordinate system, described less parallel, refer between line slide rail 1-1 and scanning object long axis of body Angular range is less than 2 °.But, the linear rail system auxiliary mentioned in aforementioned documents even with the present inventor guarantees The collimation of relative motion, scanning direction is still difficult to Efficient Support with the depth of parallelism of target plane, and bigger systematic error is in institute Unavoidably.And on the other hand, if re-using other householder methods after the calibration process having been completed structured-light system Demarcate and obtained calibrated line-structured light visual system scanning direction vector under camera coordinate system separately Then certainly will be added significantly to the complexity of calibration process, may not only cannot cut down meanwhile, the most even introduce new bigger Error deviation factor.

At present, the relevant discussion in this area not yet can explore a kind of line comprising scanning direction information with regard to the problems referred to above The complete scaling method of structure light vision measuring systems so that it is can synchronize to realize Constructed Lighting Vision System calibration process and sweep with determining Retouch direction, the stated accuracy of vision measurement system can be greatly improved again and simplify calibration process.

Summary of the invention

There is no can synchronize to determine knot to solve the complete scaling method of existing Vision Measuring System With Structured Light Stripe Structure light visual system parameter and determine scanning direction vector, can be greatly improved again vision measurement system stated accuracy and simplify mark Determining the technical problem of process, the present invention provides a kind of Vision Measuring System With Structured Light Stripe comprising scanning direction information completely to demarcate Method.

It is as follows that the present invention solves the technical scheme that technical problem taked:

Comprising the complete scaling method of Vision Measuring System With Structured Light Stripe of scanning direction information, it comprises the steps:

Step one: the relative motion based on line slide rail system that foundation is approximately parallel to scanning object long axis of body direction is auxiliary Help caliberating device:

Set up the sliding based on straight line of overall projection based on line slide rail system and photographic unit and rectangular projection target plate The relative motion auxiliary calibration device of rail system, wherein, overall projection based on line slide rail system and photographic unit include directly Line slide rail, slide block, video camera and linear laser diode；Video camera and linear laser diode are all fixed on slide block, and slide block is sliding with straight line Rail is slidably connected；Rectangular projection target plate is placed in the dynamic coverage of video camera；Linear laser diode and the court of video camera To identical and holding geo-stationary；

Step 2: the parameter of video camera is carried out initial alignment, it specifically includes following sub-step:

Step 2.1: establish conversion coordinate system, it may be assumed that set up world coordinate system, camera coordinate system, image physical coordinates system With image pixel plane coordinate system；

Three-dimensional coordinate point coordinates (x in world coordinate system^w,y^w,z^w) be converted to video camera point coordinates (x successively^c,y^c, z^c), ideal image point coordinates (x^u,y^u), true picture pixel coordinate (x^d,y^d) and actual pixels point coordinates (x^p,y^p)；Separately Outward, the camera marking method of reference Zhang Zhengyou is by the z of world coordinate system^wAxial coordinate is set to 0, i.e. z^w=0；Now, world coordinates In system, the coordinate representation of arbitrfary point is (x^w,y^w, 0), or it is only meant as the two-dimensional points coordinate (x of world coordinate system XY axial plane^w, y^w)；

Step 2.2: traditional gridiron pattern camera calibration plate is done in the coverage of video camera repeatedly different attitude Under change of pitch angle and shoot the three above photos of width, then the inner parameter and external parameter to video camera is demarcated；

I.e. by obtaining inner parameter matrixWith distortion factor K=(k₁,k₂,p₁,p₂)；

Wherein α and β represents U axle and the scale factor of V axle in pixel planes respectively, and γ represents pixel planes Two coordinate axle The out of plumb factor, (u₀,v₀) represent the coordinate that the intersection point of camera optical axis and the plane of delineation is fastened at pixel coordinate；k₁,k₂Represent The coefficient of radially distortion function in image physical coordinates system；p₁,p₂Tangential distortion function in expression image physical coordinates system respectively Coefficient；

Step 3: utilize the relative motion auxiliary calibration device based on line slide rail system described in step one, obtains and swashs Optical projection Rhizoma Dioscoreae (peeled) space plane equation, it specifically includes following sub-step:

Step 3.1: make the slide block edge in relative motion auxiliary calibration device based on line slide rail system described in step one Line slide rail sequentially moves (k-1) individual fixing step-length unit (9 >=k >=1, k takes natural number), and allows linear laser diode by laser Projection Rhizoma Dioscoreae (peeled) EF of structure optical plane S projects on static rectangular projection target plate all the time, uses video camera to divide during this Other to a width calibrating template photo corresponding under each position of rectangular projection target plate shooting, until obtaining k width to comprise target plate Calibrating template photo with projection Rhizoma Dioscoreae (peeled)；

Step 3.2: the method solving according to linear model and demarcating, comprises target plate and throwing to the k width described in step 3.1 The calibrating template photo of shadow Rhizoma Dioscoreae (peeled) carries out rectangular block Corner Detection and Light stripes center extraction detection, and it specifically includes following sub-step Rapid:

Step 3.2.1: four angle point ABCD on the rectangular projection target plate after translational moving system Coordinate Conversion are carried out Corner Detection:

Step 3.2.1.1: carry out translational moving system Coordinate Conversion: the k width described in step 3.1 is comprised target plate and projection The calibrating template photo of Rhizoma Dioscoreae (peeled) regards as under identical shooting condition the rectangular projection target by Constructed Lighting Vision System transfixion Plate in camera coordinate system along the step-length unit moving fixing with the rightabout of scanning direction time shoot acquisition respectively K width scaling board photo, equivalence can obtain scaling board under relative motion reverse state space occur in camera coordinate system Heterodromous i width new images (9 >=i >=1, i takes natural number)；

Step 3.2.1.2: for described in step 3.2.1.1 any one heterodromous i width new images, utilize angle point to examine Method of determining and calculating, respectively the pixel detection coordinate of four angle points of black box in detection template imageWherein i= 1 ..., 9 represent the width number sequence number of images, j=1 ..., 4 represent in current single images four angle points sequence number in the figure: Utilize the inner parameter described in step 2.2 and calibrating external parameters model can set up following expression:

$\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\\ 1\end{array}\right]={\left(\begin{array}{ccc}\mathrm{\α}& \mathrm{\γ}& u_0\\ 0& \mathrm{\β}& v_0\\ 0& 0& 1\end{array}\right)}^{-1}\left[\begin{array}{c}{x}_{ij}^p\\ y_{ij}^p\\ 1\end{array}\right]$

$\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\end{array}\right]=(1+k_1{r}^2+k_2{r}^4)\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\end{array}\right]+\left[\begin{array}{c}2p_1{x}_{ij}^d{y}_{ij}^d+p_2(r^2+2{\left(x_{ij}^d\right)}^2)\\ p_1(r^2+2{\left(y_{ij}^d\right)}^2)+2p_2{x}_{ij}^d{y}_{ij}^d\end{array}\right]$

Pixel detection by four angle point ABCD in described in step 3.2.1.1 any one heterodromous i width new images CoordinateAll correspondences are converted into ideal image coordinate system coordinateWherein Represent the angle point image coordinate system coordinate comprising distortion,Represent the ideal image coordinate system coordinate of angle point；

Utilize corner pixels detection coordinatesTwo-dimensional points coordinate figure corresponding in its world coordinate systemAccording to the model of Zhang Zhengyou, following equation can be obtained

$s\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\\ 1\end{array}\right]=\[\begin{array}{ccc}{r}_1& r_2& T\end{array}\]\left[\begin{array}{c}{x}_{ij}^w\\ y_{ij}^w\\ 1\end{array}\right]\mathrm{......}\left(1\right)$

In formula (1), S represents that scale factor, every photos image i external parameter corresponding to target that hit all includes R_iAnd T_i Two parameters；R_i=[r₁,r₂,r₃] it is spin matrix, r_i(i=1,2,3) spin matrix R is represented_iI-th row, T_iRepresent translation Vector；If

$\tilde{m}={\[x_{ij}^u,y_{ij}^u,1\]}^T$

$\tilde{M}={\[x_{ij}^w,y_{ij}^w,1\]}^T$

H=[r₁ r₂ T]

Then equation (1) can be to be expressed as form:

$s\tilde{m}=H\tilde{M}\mathrm{......}\left(2\right)$

Divide out in equation (2) s, can obtain

WhereinI-th row of representing matrix H, (3) formula also may indicate that into following form

(4) formula is rewritten as matrix form obtain

Step 3.2.1.3: setThen when i sequentially takes 1 to 4, can be by step according to formula (5) The ideal image coordinate system coordinate of four angle point ABCD in the most described every heterodromous i width new images (j=sequentially takes 1 to 4) all can obtain two equations according to formula (5) is corresponding and can be expressed as homogeneous equation group Lx= 0；Therefore, when i sequentially takes 1 to 4, add up to and obtain eight equations altogether；

Wherein L is the coefficient matrix of 8 × 9；Matrix L is carried out singular value decomposition, the then right side that matrix L minimum singular value is corresponding Singular vector is the solution of equation group；Such that it is able to solve spin matrix R_iWith translation vector T_i, thus obtain step The space equation of the most described any one heterodromous i width new images place plane, and it is right to obtain its institute further Answer each width of step 3.2 comprise target plate and projection Rhizoma Dioscoreae (peeled) calibrating template photo k sequentially take in space equation i of plane 2 ... 9, the spin matrix R of (i takes natural number) and translation vector T the two external parameter；

Step 3.2.1.4: the rectangle included in described in step 3.2.1.1 every heterodromous i width new images is thrown Laser projection Rhizoma Dioscoreae (peeled) on shadow target plate carries out Rhizoma Dioscoreae (peeled) widthwise central detection, utilizes optical losses detection algorithm to detect and is incident upon The pixel coordinate point of optical losses in every width target plate imageWherein i=1 ..., 9, (i takes natural number) represents The width number sequence number of image, j=1 ..., N (N takes natural number) represents the number of optical losses pixel coordinate point；

Step 3.2.2: utilize every photos described in the inner parameter of video camera described in step 2.2 and step 3.2.1.3 Image hits spin matrix R corresponding to target and the pixel coordinate of optical losses is thrown by translation vector T the two external parameter Shadow, in camera coordinate system, i.e. obtains the three-dimensional camera coordinate system coordinate points of correspondenceThus obtain with light Bar centrage is the laser projection Rhizoma Dioscoreae (peeled) point set characterized；

Step 3.3: repetition step 3.2.1 is to step 3.2.2, until sequentially obtaining i=1 ..., nine groups of 9 difference correspondences Target plate and the calibrating template photo laser projection Rhizoma Dioscoreae (peeled) spatial point set one to one of projection Rhizoma Dioscoreae (peeled) is comprised with k width；

Step 4: the whole three-dimensional point in nine groups of laser projection Rhizoma Dioscoreae (peeled) step 3 obtained all project to unified taking the photograph In camera coordinate system, and utilize these point set matchings to obtain a unique space plane, thus solve the sky of structure optical plane Between equation, it is achieved the demarcation of optical plane；

Step 5: utilize spin matrix R and translation that intrinsic parameters of the camera described in step 2.2 is corresponding with scaling board image Vector T the two external parameter, four the corner pixels coordinates that will obtain in the step 3.1 in each imageProjection In unified camera coordinate system, it is possible to obtain four groups of three-dimensional point setWherein i=1 ..., 9 represent image Width number sequence number, j=1 ..., 4；Utilize this four groups of three-dimensional point set, four space parallel lines vector equations of matching:

$\left\{\begin{array}{c}\frac{x-x_0^1}{X_1}=\frac{y-y_0^1}{Y_1}=\frac{z-z_0^1}{Z_1}\\ \frac{x-x_0^2}{X_2}=\frac{y-y_0^2}{Y_2}=\frac{z-z_0^2}{Z_2}\\ \frac{x-x_0^3}{X_3}=\frac{y-y_0^3}{Y_3}=\frac{z-z_0^3}{Z_3}\\ \frac{x-x_0^4}{X_4}=\frac{y-y_0^4}{Y_4}=\frac{z-z_0^4}{Z_4}\end{array}\right.\mathrm{.......}\left(6\right)$

Step 6: the average of four space parallel lines vector equations described in calculation procedure five, it is thus achieved that equation:

$\left\{\begin{array}{c}X=\frac{X_1+X_2+X_3+X_4}{4}\\ Y=\frac{Y+Y_2+Y_3+Y_4}{4}\\ Z=\frac{Z_1+Z_2+Z_3+Z_4}{4}\end{array}\right.\mathrm{......}\left(7\right)$

And then may determine that the vector of the scanning direction of system

Beneficial effects of the present invention is as follows:

The present invention, under the auxiliary of relative motion auxiliary calibration device based on line slide rail system, innovatively incorporates relatively Kinetic coordinate system conversion thinking, thus the Light-plane calibration process directly utilizing routine disposably obtains several and includes light simultaneously Bar positional information and the reverse relative motion uncalibrated image of scanning direction information, thus while being greatly simplified demarcating steps, Decrease the operand calculated needed for system solution structure optical plane space equation the most very significantly.Additionally, utilize several same Time include Rhizoma Dioscoreae (peeled) positional information and the uncalibrated image of scanning direction information, the method can also accurately solve and simulate scanning side To vector equation, and scanning direction can be any direction, thoroughly releases and has abandoned camera coordinate system and swept really Retouch direction coordinate system to need the artificial error precision keeping high accuracy to overlap to ensure to limit, and final whole new definition and establish bag Containing determining that the step of scan vector is at the most complete interior Vision Measuring System With Structured Light Stripe scaling method.

The present invention has both the double effects improving demarcation efficiency with reducing calibrated error, and its logic links up rigorous, fully Utilize and play on the basis of prior art, its elite is come down in a continuous line, it is simple to understanding, in the field of business have higher amalgamation and Acceptance, and for the shortcoming and defect of old technology, the present invention looks for another way, solves preferably, and therefore, this technology is determined scheme tool There is higher promotion and use value.

Accompanying drawing explanation

Fig. 1 is the figure Figure 1 in paper recited in background technology；

Fig. 2 is the relative transformational relation schematic diagram in Fig. 1 between four kinds of camera calibration coordinate systems；

Fig. 3 is the figure Figure 3 in paper recited in background technology；

Fig. 4 is the left figure in the figure Figure 4 in paper recited in background technology；

Fig. 5 is the figure Figure 6 in paper recited in background technology；

Fig. 6 is the close-up schematic view of I extracting section optical losses algorithm in Fig. 4；

Fig. 7 is the figure Figure 10 in paper recited in background technology；

Fig. 8 is old by being approximately parallel to the line slide rail in scanning object long axis of body direction to approximate given scanning direction Principle schematic；

Fig. 9 is the structure chart of present invention relative motion based on line slide rail system auxiliary calibration device and realizes auxiliary Determine the schematic diagram approximating given scanning direction；

Figure 10 is the present invention to be closed and collective fits to the space of only one structure optical plane by organizing laser projection Rhizoma Dioscoreae (peeled) point set more The schematic diagram of equation；

Figure 11 is that the present invention utilizes the angle point ABCD of many group rectangular target blocks to simulate the former of target block relative displacement direction Reason figure.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is described in further details.

As shown in Figures 9 to 11, the Vision Measuring System With Structured Light Stripe comprising scanning direction information of the present invention is completely marked The method of determining comprises the steps:

Step one: the relative motion based on line slide rail system that foundation is approximately parallel to scanning object long axis of body direction is auxiliary Help caliberating device, by this system, incorporate to innovation formula the relative motion coordinate system conversion thinking that the present invention is exclusive, auxiliary to realize Help the scanning direction vector determining that approximation is given

Set up overall projection based on line slide rail system and photographic unit 1 and rectangular projection target plate 2 based on straight line The relative motion auxiliary calibration device of slide type rail system, wherein, overall projection based on line slide rail system and photographic unit 1 include Line slide rail 1-1, slide block 1-2, video camera 1-3 and linear laser diode 1-4；Video camera 1-3 and linear laser diode 1-4 is all fixed on On slide block 1-2, slide block 1-2 is slidably connected with line slide rail 1-1；Rectangular projection target plate 2 is placed on the dynamic bat of video camera 1-3 In the range of taking the photograph；Linear laser diode 1-4 and video camera 1-3 towards identical and keep geo-stationary；

Step 2: the parameter of video camera 1-3 is carried out initial alignment, it specifically includes following sub-step:

Step 2.1: according to known Tsai R Y at A versatile camera calibrationtechnique for high-accuracy 3D machine vision metrology using off-the-shelf TV cameras And lenses " [J] .IEEE Journal ofRobotics and Automation, 1987,3 (4): 323-344 literary compositions carry The coordinate transformation method that goes out establishes conversion coordinate system, it may be assumed that set up world coordinate system, camera coordinate system, image physical coordinates system With image pixel plane coordinate system；

Three-dimensional coordinate point coordinates (x in world coordinate system^w,y^w,z^w) be converted to video camera point coordinates (x successively^c,y^c, z^c), ideal image point coordinates (x^u,y^u), true picture pixel coordinate (x^d,y^d) and actual pixels point coordinates (x^p,y^p)；Separately Outward, the camera marking method of reference Zhang Zhengyou is by the z of world coordinate system^wAxial coordinate is set to 0, i.e. z^w=0；Now, world coordinates In system, the coordinate representation of arbitrfary point is (x^w,y^w, 0), or it is only meant as the two-dimensional points coordinate (x of world coordinate system XY axial plane^w, y^w)；

Step 2.2: traditional gridiron pattern camera calibration plate is done in the coverage of video camera 1-3 repeatedly different appearance Change of pitch angle under state also shoots the three above photos of width, according still further to known Zhang Zhengyou at Zhang Z. (2000) .A flexible new technique for camera calibration.IEEE Trans.on Pattern Analysis And Machine Intelligence, proposed in 22 (11): 1330-1334., camera marking method is to video camera 1-3's Inner parameter and external parameter are demarcated；

I.e. by obtaining inner parameter matrixWith distortion factor K=(k₁,k₂,p₁,p₂)；

Wherein α and β represents U axle and the scale factor of V axle in pixel planes respectively, and γ represents pixel planes Two coordinate axle The out of plumb factor, (u₀,v₀) represent the coordinate that the intersection point of camera optical axis and the plane of delineation is fastened at pixel coordinate；k₁,k₂Represent The coefficient of radially distortion function in image physical coordinates system；p₁,p₂Tangential distortion function in expression image physical coordinates system respectively Coefficient；

Step 3: utilize the relative motion auxiliary calibration device based on line slide rail system described in step one, obtains and swashs Optical projection Rhizoma Dioscoreae (peeled) space plane equation, it specifically includes following sub-step:

Step 3.1: make the slide block 1-2 in relative motion auxiliary calibration device based on line slide rail system described in step one Linearly slide rail 1-1 sequentially moves step-length unit 9 >=k >=1 fixing for k-1, and k takes natural number, and allows the linear laser diode 1-4 will Projection Rhizoma Dioscoreae (peeled) EF of laser structure optical plane (Structured light plane) S projects to static rectangular projection target all the time On target 2, during this, use video camera 1-3 that rectangular projection target plate 2 shoots a width mark respectively corresponding under each position Solid plate photo, until obtaining k width to comprise target plate and the calibrating template photo of projection Rhizoma Dioscoreae (peeled)；

Step 3.2: according to the present inventor at known A Flexible Calibration Method Using the Planar Targetwith a Square Pattern for Line Structured Light VisionSystem.Sun,Qiucheng,et al.A Flexible Calibration Method Using the Planar Target with a Square Pattern for Line Structured Light Vision Linear model disclosed in System.PloS one 9.9:e106911. (SCI) 2014. solves and scaling method, to step The calibrating template photo that k width described in 3.1 comprises target plate and projection Rhizoma Dioscoreae (peeled) carries out rectangular block Corner Detection and optical losses carries Taking detection, it specifically includes following sub-step:

Step 3.2.1: according to motion principle of relativity, shoot the K width scaling board photo of acquisition in step 3.1, with identical bat Taking the photograph under environment, Constructed Lighting Vision System is fixed, and rectangular projection target plate in camera coordinate system along with scanning direction When rightabout motion, the captured scaling board photo obtained is equivalent, accordingly, to translational moving system Coordinate Conversion After rectangular projection target plate 2 on four angle point ABCD carry out Corner Detection:

Step 3.2.1.1: carry out translational moving system Coordinate Conversion: the k width described in step 3.1 is comprised target plate and projection The calibrating template photo of Rhizoma Dioscoreae (peeled) regards as under identical shooting condition the rectangular projection target by Constructed Lighting Vision System transfixion Plate in camera coordinate system along the step-length unit moving fixing with the rightabout of scanning direction time shoot acquisition respectively K width scaling board photo, equivalence can obtain scaling board under relative motion reverse state space occur in camera coordinate system Heterodromous i width new images 9 >=i >=1, i takes natural number；

Step 3.2.1.2: for described in step 3.2.1.1 any one heterodromous i width new images, utilizes known JeanY B. is at Pyramidal implementation of the Lucas Kanade feature tracker Corner Detection Algorithm disclosed in description of the algorithm. document is black in detection template image respectively The pixel detection coordinate of four angle point ABCD of color gridWherein i=1 ..., 9 represent the width number sequence number of image, j =1 ..., 4 represent in current single images four angle points sequence number in the figure: utilize the inner parameter described in step 2.2 and Calibrating external parameters model can set up following expression:

$\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\\ 1\end{array}\right]={\left(\begin{array}{ccc}\mathrm{\α}& \mathrm{\γ}& u_0\\ 0& \mathrm{\β}& v_0\\ 0& 0& 1\end{array}\right)}^{-1}\left[\begin{array}{c}{x}_{ij}^p\\ y_{ij}^p\\ 1\end{array}\right]$

$\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\end{array}\right]=(1+k_1{r}^2+k_2{r}^4)\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\end{array}\right]+\left[\begin{array}{c}2p_1{x}_{ij}^d{y}_{ij}^d+p_2(r^2+2{\left(x_{ij}^d\right)}^2)\\ p_1(r^2+2{\left(y_{ij}^d\right)}^2)+2p_2{x}_{ij}^d{y}_{ij}^d\end{array}\right]$

The pixel detection of four angle point ABCD in described in step 3.2.1.1 any one heterodromous i width new images is sat MarkAll correspondences are converted into ideal image coordinate system coordinateWherein Represent the angle point image coordinate system coordinate comprising distortion,Represent the ideal image coordinate system coordinate of angle point；

Utilize corner pixels detection coordinatesTwo-dimensional points coordinate figure corresponding in its world coordinate systemAccording to the model of Zhang Zhengyou, following equation can be obtained

$s\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\\ 1\end{array}\right]=\[\begin{array}{ccc}{r}_1& r_2& T\end{array}\]\left[\begin{array}{c}{x}_{ij}^w\\ y_{ij}^w\\ 1\end{array}\right]\mathrm{......}\left(1\right)$

In formula (1), S represents that scale factor, every photos image i external parameter corresponding to target that hit all includes R_iAnd T_i Two parameters；R_i=[r₁,r₂,r₃] it is spin matrix, r_iI=1,2,3 represents spin matrix R_iI-th row, T_iExpression is translated towards Amount；If

$\tilde{m}={\[x_{ij}^u,y_{ij}^u,1\]}^T$

$\tilde{M}={\[x_{ij}^w,y_{ij}^w,1\]}^T$

H=[r₁ r₂ T]

Then equation (1) can be to be expressed as form:

$s\tilde{m}=H\tilde{M}\mathrm{......}\left(2\right)$

Divide out in equation (2) s, can obtain

WhereinI-th row of representing matrix H, (3) formula also may indicate that into following form

(4) formula is rewritten as matrix form obtain

Step 3.2.1.3: setThen when i sequentially takes 1 to 4, can be by step according to formula (5) The ideal image coordinate system coordinate of four angle point ABCD in the most described every heterodromous i width new imagesJ=sequentially takes 1 to 4 and all can obtain two equations according to formula (5) is corresponding and can be expressed as homogeneous equation Group Lx=0；Therefore, when i sequentially takes 1 to 4, total can obtain eight equations altogether；

Wherein L is the coefficient matrix of 8 × 9；Matrix L is carried out singular value decomposition, the then right side that matrix L minimum singular value is corresponding Singular vector is the solution of equation group；Such that it is able to solve spin matrix R_iWith translation vector T_i, thus obtain step The space equation of the most described any one heterodromous i width new images place plane, and it is right to obtain its institute further Answer each width of step 3.2 comprise target plate and projection Rhizoma Dioscoreae (peeled) calibrating template photo k sequentially take in space equation i of plane 2 ... 9, i take natural spin matrix R and translation vector T the two external parameter；

Step 3.2.1.4: according to the present inventor at Chinese patent CN103400399A mono-kind line structure based on spatial moment Method disclosed in light center extracting method to described in step 3.2.1.1 every heterodromous i width new images included in Rectangular projection target plate on laser projection Rhizoma Dioscoreae (peeled) carry out Rhizoma Dioscoreae (peeled) widthwise central detection, utilize optical losses detection algorithm to detect Go out to be incident upon the pixel coordinate point of optical losses in every width target plate imageWherein i=1 ..., 9, i take from so The width number sequence number of number expression image, j=1 ..., N_NTake natural number and represent the number of optical losses pixel coordinate point；

Step 3.2.2: utilize every photos described in the inner parameter of video camera described in step 2.2 and step 3.2.1.3 Image hits spin matrix R corresponding to target and the pixel coordinate of optical losses is thrown by translation vector T the two external parameter Shadow, in camera coordinate system, i.e. obtains the three-dimensional camera coordinate system coordinate points of correspondenceThus obtain with Rhizoma Dioscoreae (peeled) Centrage is the laser projection Rhizoma Dioscoreae (peeled) point set characterized；

Step 3.3: repetition step 3.2.1 is to step 3.2.2, until sequentially obtaining i=1 ..., nine groups of 9 difference correspondences Target plate and the calibrating template photo laser projection Rhizoma Dioscoreae (peeled) spatial point set one to one of projection Rhizoma Dioscoreae (peeled) is comprised with k width；

Step 4: the whole three-dimensional point in nine groups of laser projection Rhizoma Dioscoreae (peeled) step 3 obtained all project to unified taking the photograph In camera coordinate system, and utilize these point set matchings to obtain a unique space plane, thus solve the sky of structure optical plane Between equation, it is achieved the demarcation of optical plane；

Step 5: utilize spin matrix R and translation that intrinsic parameters of the camera described in step 2.2 is corresponding with scaling board image The most important external parameter of vector T the two, sits the pixel of four the angle point ABCD obtained in the step 3.1 in each image MarkProject in unified camera coordinate system, it is possible to obtain four groups of three-dimensional point setWherein i= 1 ..., the width number sequence number of 9 expression images, j=1 ..., 4；Utilize this four groups of three-dimensional point set, four space straight parallel of matching Line vector equation:

$\left\{\begin{array}{c}\frac{x-x_0^1}{X_1}=\frac{y-y_0^1}{Y_1}=\frac{z-z_0^1}{Z_1}\\ \frac{x-x_0^2}{X_2}=\frac{y-y_0^2}{Y_2}=\frac{z-z_0^2}{Z_2}\\ \frac{x-x_0^3}{X_3}=\frac{y-y_0^3}{Y_3}=\frac{z-z_0^3}{Z_3}\\ \frac{x-x_0^4}{X_4}=\frac{y-y_0^4}{Y_4}=\frac{z-z_0^4}{Z_4}\end{array}\right.\mathrm{......}\left(6\right)$

Step 6: the average of four space parallel lines vector equations described in calculation procedure five, it is thus achieved that equation:

$\left\{\begin{array}{c}X=\frac{X_1+X_2+X_3+X_4}{4}\\ Y=\frac{Y+Y_2+Y_3+Y_4}{4}\\ Z=\frac{Z_1+Z_2+Z_3+Z_4}{4}\end{array}\right.\mathrm{......}\left(7\right)$

And then may determine that the vector of the scanning direction of system

The present invention has both the double effects improving demarcation efficiency with reducing calibrated error, has higher popularization and use valency Value.

Claims (1)

1. comprise the complete scaling method of Vision Measuring System With Structured Light Stripe of scanning direction information, it is characterised in that: the method bag Include following steps:

Step one: set up relative motion based on the line slide rail system auxiliary mark being approximately parallel to scanning object long axis of body direction Determine device:

Set up overall projection based on line slide rail system and photographic unit (1) and rectangular projection target plate (2) based on straight line The relative motion auxiliary calibration device of slide type rail system, wherein, overall projection based on line slide rail system and photographic unit (1) bag Include line slide rail (1-1), slide block (1-2), video camera (1-3) and linear laser diode (1-4)；Video camera (1-3) and linear laser Device (1-4) is all fixed on slide block (1-2), and slide block (1-2) is slidably connected with line slide rail (1-1)；Rectangular projection target plate (2) It is placed in the dynamic coverage of video camera (1-3)；Linear laser diode (1-4) and video camera (1-3) towards identical and protect Hold geo-stationary；

Step 2: the parameter of video camera (1-3) is carried out initial alignment, it specifically includes following sub-step:

Step 2.1: establish conversion coordinate system, it may be assumed that set up world coordinate system, camera coordinate system, image physical coordinates system and figure As pixel planes coordinate system；

Three-dimensional coordinate point coordinates (x in world coordinate system^w,y^w,z^w) be converted to video camera point coordinates (x successively^c,y^c,z^c), reason Think picture point coordinate (x^u,y^u), true picture pixel coordinate (x^d,y^d) and actual pixels point coordinates (x^p,y^p)；It addition, reference The camera marking method of Zhang Zhengyou is by the z of world coordinate system^wAxial coordinate is set to 0, i.e. z^w=0；Now, appoint in world coordinate system The coordinate representation of meaning point is (x^w,y^w, 0), or it is only meant as the two-dimensional points coordinate (x of world coordinate system XY axial plane^w,y^w)；

Step 2.2: traditional gridiron pattern camera calibration plate is done in the coverage of video camera (1-3) repeatedly different attitude Under change of pitch angle and shoot the three above photos of width, then the inner parameter and external parameter to video camera (1-3) is demarcated；

I.e. by obtaining inner parameter matrixWith distortion factor K=(k₁,k₂,p₁,p₂)；

Wherein α and β represents U axle and the scale factor of V axle in pixel planes respectively, and γ represents not hanging down of pixel planes Two coordinate axle The straight factor, (u₀,v₀) represent the coordinate that the intersection point of camera optical axis and the plane of delineation is fastened at pixel coordinate；k₁,k₂Represent image The coefficient of radial direction distortion function in physical coordinates system；p₁,p₂Represent tangential distortion function in image physical coordinates system respectively is Number；

Step 3: utilize the relative motion auxiliary calibration device based on line slide rail system described in step one, obtains laser and throws Shadow Rhizoma Dioscoreae (peeled) space plane equation, it specifically includes following sub-step:

Step 3.1: make slide block (1-2) edge in relative motion auxiliary calibration device based on line slide rail system described in step one Line slide rail (1-1) sequentially moves (k-1) individual fixing step-length unit (9 >=k >=1, k takes natural number), and allows linear laser diode (1-4) projection Rhizoma Dioscoreae (peeled) EF of laser structure optical plane S is projected to all the time on static rectangular projection target plate (2), this process Middle use video camera (1-3) shoots a width calibrating template photo corresponding under each position respectively to rectangular projection target plate (2), Until obtaining k width to comprise target plate and the calibrating template photo of projection Rhizoma Dioscoreae (peeled)；

Step 3.2: the method solving according to linear model and demarcating, comprises target plate and projected light to the k width described in step 3.1 The calibrating template photo of bar carries out rectangular block Corner Detection and Light stripes center extraction detection, and it specifically includes following sub-step:

Step 3.2.1: four angle point ABCD on rectangular projection target plate (2) after translational moving system Coordinate Conversion are carried out angle Point detection:

Step 3.2.1.1: carry out translational moving system Coordinate Conversion: the k width described in step 3.1 is comprised target plate and projection Rhizoma Dioscoreae (peeled) Calibrating template photo regard as under identical shooting condition that rectangular projection target plate exists by Constructed Lighting Vision System transfixion The k width of acquisition is shot respectively time in camera coordinate system along the step-length unit moving fixing with the rightabout of scanning direction Scaling board photo, of equal value can obtain scaling board under relative motion reverse state and occur space reversely to transport in camera coordinate system Dynamic i width new images (9 >=i >=1, i takes natural number)；

Step 3.2.1.2: for described in step 3.2.1.1 any one heterodromous i width new images, utilize Corner Detection to calculate Method, respectively the pixel detection coordinate of four angle points of black box in detection template imageWherein i=1 ..., 9 Represent the width number sequence number of image, j=1 ..., 4 represent in current single images four angle points sequence number in the figure: utilize step Inner parameter described in rapid 2.2 and calibrating external parameters model can set up following expression:

$\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\\ 1\end{array}\right]={\left(\begin{array}{ccc}\mathrm{\α}& \mathrm{\γ}& u_0\\ 0& \mathrm{\β}& v_0\\ 0& 0& 1\end{array}\right)}^{-1}\left[\begin{array}{c}{x}_{ij}^p\\ y_{ij}^p\\ 1\end{array}\right]$

$\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\end{array}\right]=(1+k_1{r}^2+k_2{r}^4)\left[\begin{array}{c}{x}_{ij}^d\\ y_{ij}^d\end{array}\right]+\left[\begin{array}{c}2p_1{x}_{ij}^d{y}_{ij}^d+p_2(r^2+2{\left(x_{ij}^d\right)}^2)\\ p_1(r^2+2{\left(y_{ij}^d\right)}^2)+2p_2{x}_{ij}^d{y}_{ij}^d\end{array}\right]$

The pixel detection of four angle point ABCD in described in step 3.2.1.1 any one heterodromous i width new images is sat MarkAll correspondences are converted into ideal image coordinate system coordinateWherein Represent the angle point image coordinate system coordinate comprising distortion,Represent the ideal image coordinate system coordinate of angle point；

Utilize corner pixels detection coordinatesTwo-dimensional points coordinate figure corresponding in its world coordinate systemRoot According to the model of Zhang Zhengyou, following equation can be obtained

$s\left[\begin{array}{c}{x}_{ij}^u\\ y_{ij}^u\\ 1\end{array}\right]=\[\begin{array}{ccc}{r}_1& r_2& T\end{array}\]\left[\begin{array}{c}{x}_{ij}^w\\ y_{ij}^w\\ 1\end{array}\right]\mathrm{......}\left(1\right)$

In formula (1), S represents that scale factor, every photos image i external parameter corresponding to target that hit all includes R_iAnd T_iTwo Parameter；R_i=[r₁,r₂,r₃] it is spin matrix, r_i(i=1,2,3) spin matrix R is represented_iI-th row, T_iRepresent translation vector； If

$\tilde{m}={\[x_{ij}^u,y_{ij}^u,1\]}^T$

$\tilde{M}={\[x_{ij}^w,y_{ij}^w,1\]}^T$

H=[r₁ r₂ T]

Then equation (1) can be to be expressed as form:

$s\tilde{m}=H\tilde{M}\mathrm{......}\left(2\right)$

Divide out in equation (2) s, can obtain

WhereinI-th row of representing matrix H, (3) formula also may indicate that into following form

(4) formula is rewritten as matrix form obtain

Step 3.2.1.3: setThen when i sequentially takes 1 to 4, can be by step according to formula (5) The ideal image coordinate system coordinate of four angle point ABCD in the most described every heterodromous i width new images(j=sequentially takes 1 to 4) all can obtain two equations according to formula (5) is corresponding and can be expressed as homogeneous side Journey group Lx=0；Therefore, when i sequentially takes 1 to 4, add up to and obtain eight equations altogether；

Wherein L is the coefficient matrix of 8 × 9；Matrix L carries out singular value decomposition, then the right side that matrix L minimum singular value is corresponding is unusual Vector is the solution of equation group；Such that it is able to solve spin matrix R_iWith translation vector T_i, thus obtain step 3.2.1.1 institute State the space equation of any one heterodromous i width new images place plane, and it is corresponding in step to obtain it further The calibrating template photo k that rapid 3.2 each width comprise target plate and projection Rhizoma Dioscoreae (peeled) sequentially takes 2 in space equation i of plane ... 9, The spin matrix R of (i takes natural number) and translation vector T the two external parameter；

Step 3.2.1.4: to the rectangular projection target included in described in step 3.2.1.1 every heterodromous i width new images Laser projection Rhizoma Dioscoreae (peeled) on target carries out Rhizoma Dioscoreae (peeled) widthwise central detection, utilizes optical losses detection algorithm to detect and is incident upon every width The pixel coordinate point of optical losses in target plate imageWherein i=1 ..., 9, (i takes natural number) represents image Width number sequence number, j=1 ..., N (N takes natural number) represent optical losses pixel coordinate point number；

Step 3.2.2: utilize every photos image described in the inner parameter of video camera described in step 2.2 and step 3.2.1.3 The pixel coordinate projection of optical losses is arrived by spin matrix R and translation vector T the two external parameter that the target that hits is corresponding In camera coordinate system, i.e. obtain the three-dimensional camera coordinate system coordinate points of correspondenceThus obtain with in Rhizoma Dioscoreae (peeled) Heart line is the laser projection Rhizoma Dioscoreae (peeled) point set characterized；

Step 3.3: repetition step 3.2.1 is to step 3.2.2, until sequentially obtaining i=1 ..., nine groups and k of 9 difference correspondences Width comprises target plate and the calibrating template photo laser projection Rhizoma Dioscoreae (peeled) spatial point set one to one of projection Rhizoma Dioscoreae (peeled)；

Step 4: the whole three-dimensional point in nine groups of laser projection Rhizoma Dioscoreae (peeled) step 3 obtained all project to unified video camera In coordinate system, and utilize these point set matchings to obtain a unique space plane, thus solve the space side of structure optical plane Journey, it is achieved the demarcation of optical plane；

Step 5: utilize spin matrix R and translation vector that intrinsic parameters of the camera described in step 2.2 is corresponding with scaling board image T the two external parameter, four the corner pixels coordinates that will obtain in the step 3.1 in each imageProject to system In the camera coordinate system of, it is possible to obtain four groups of three-dimensional point setWherein i=1 ..., 9 represent the width of image Number sequence number, j=1 ..., 4；Utilize this four groups of three-dimensional point set, four space parallel lines vector equations of matching:

$\left\{\begin{array}{c}\frac{x-x_0^1}{X_1}=\frac{y-y_0^1}{Y_1}=\frac{z-z_0^1}{Z_1}\\ \frac{x-x_0^2}{X_2}=\frac{y-y_0^2}{Y_2}=\frac{z-z_0^2}{Z_2}\\ \frac{x-x_0^3}{X_3}=\frac{y-y_0^3}{Y_3}=\frac{z-z_0^3}{Z_3}\\ \frac{x-x_0^4}{X_4}=\frac{y-y_0^4}{Y_4}=\frac{z-z_0^4}{Z_4}\end{array}\right.\mathrm{......}\left(6\right)$

Step 6: the average of four space parallel lines vector equations described in calculation procedure five, it is thus achieved that equation:

$\left\{\begin{array}{c}X=\frac{X_1+X_2+X_3+X_4}{4}\\ Y=\frac{Y+Y_2+Y_3+Y_4}{4}\\ Z=\frac{Z_1+Z_2+Z_3+Z_4}{4}\end{array}\right.\mathrm{......}\left(7\right)$

And then may determine that the vector of the scanning direction of system