CN101334270A - Laser line scanning feeler geometric transformation calibration and curved face interpolation correcting method and apparatus - Google Patents

Abstract

The invention discloses a method for the geometric transformation calibration and the curved surface interpolation modification of a laser line scanning probe, and a device; the noncontact measurement technology is adopted to be used for the accurate calibration and further modification of the probe of a laser line scanning three-dimensional profile measuring system; the method carries out calculation on the basis of the basic measuring principle of a probe system and completely eliminates calibration theory errors brought by the traditional calibration method (such as a polyfit calculation method, etc.). In the detailed parameter calculating process, the statistical analysis technologies of curve analysis and least squares, etc., are adopted by the algorithm; therefore, the requirement of multiscale calibration caused by the related factors of lens distortion, probe assembly and adjustment error, etc. to different calibration areas is fully considered on the basis that the basic principle is fully met, thus leading the probe to have excellent stability and operability.

Description

Laser line scanning feeler geometric transformation calibration and curved face interpolation correcting method and device

Technical field

The present invention relates to a kind of laser line scanning feeler geometric transformation calibration and curved face interpolation correcting method and device.

Background technology

What laser line scanning feeler adopted is the three-D profile scanning survey that ripe optical triangulation principle realizes testee.The linear laser device penetrates a laser plane, laser is mapped to the testee surface, the light that forms on testee surface produces reflection, and part images on the imaging plane of area array CCD by the CCD imaging lens, the laser-bounce image by image pick-up card by computer acquisition.Along with the variation of testee height, the image that collects also changes thereupon, and this that is to say, includes the elevation information of testee in the laser-bounce image that CCD collects.

Present traditional scaling method mainly contains:

1, standard component method.The master body that promptly has certain precision by processing and fabricating obtains the corresponding variation relation of image according to the gauge head and the relative motion of standard component then.Also can directly carry out computer acquisition by the dot matrix panel that makes, and the computed image dot matrix, to obtain corresponding relation.The general conversion situation of considering to demarcate two directions in the plane respectively of this method, can only adapt to ideal situation, and also require very strict for calibration process, and in actual conditions, because the restriction of multiple factor, CCD measures the visual field there is multiple dimensioned requirement in calibration result, and promptly desirable calibration result can't satisfy application requirements.Simultaneously owing to there is not initiatively light source, the Flame Image Process difficulty, this has also limited the raising of stated accuracy.

2, virtual grid method.Utilize point of fixity or set up corresponding relation by assisting linear laser instrument and the linear laser instrument of incident to intersect the image change that forms after the intersection point that generates moves in the position of gauge head.This method is owing to adopt the open loop campaign to realize, spatial positioning accuracy is not high, promptly can't form high precision planning space net point, simultaneously all asks to hand over according to straight line during two linear laser Flame Image Process and calculates, the intersection error of being asked is too big, so also can't obtain image lattice point accurately.

Mainly contain interpolation calculation method and Function Mapping method on the corresponding relation setting up.All there is the computational accuracy problem in these two kinds of computing method.The former constructs the interpolation difficulty, and calculates more complicated, and key is that interpolation precision is difficult to be guaranteed under the situation of limited calibration point; The latter is difficult to explain this relation by a suitable function owing to the relationship map complexity, so the way of Function Mapping, and as the polynomial expression mapping function, the precision difference is very big in the different measuring zone.

Summary of the invention

The objective of the invention is to overcome the shortcoming of above-mentioned prior art, a kind of complete scaling method based on line sweep triangle imaging geometry conversion principle is provided, calculate relevant geometric transformation parameter, simultaneously carry out inverse and extract gauge head and measure systematic error in the field depth entirely demarcating dot matrix based on calibration result, when measuring,, can well satisfy gauge head and measure in the field depth zones of different entirely laser line scanning feeler geometric transformation calibration and the curved face interpolation correcting method and the device of the multiple dimensioned requirement of calibration result in addition Real-time and Dynamic compensation of measurement result.

For achieving the above object, the inventive system comprises: comprise Machine basement and be arranged on the Machine basement mutually vertically by motor-driven along X, Y, Z direction rectilinear motion X-axis precise grating, Y-axis precise grating and Z axle precise grating, the Y-axis precise grating is provided with parallel and measurement board support the linear laser instrument vertical with the XOZ plane of incident angle, is positioned at the lower end of measuring linear laser instrument on the surface of Machine basement and is provided with rotary table.

Z axle precise grating of the present invention is parallel with the linear laser instrument of measurement, and vertical with the Y-axis precise grating; Also be provided with auxiliary linear laser instrument in addition, and the laser plane of auxiliary linear laser instrument intersects with the laser plane of measuring linear laser instrument generation; Also be provided with the area array CCD that has camera lens on the Y-axis precise grating, the front end of this area array CCD is provided with optical polarizer.

Demarcation of the present invention and curved face interpolation correcting method are as follows:

1) adjusts each caliberating device according to calibration request, guarantee to measure linear laser instrument incident laser perpendicular to XOY plane under the lathe coordinate system, and be parallel to X-axis; The image visual scope of area array CCD and the decision of camera lens optical axis direction can reach measures depth of field requirement; Adjust the position of area array CCD, guarantee ZOX plane parallel under camera lens optical axis and the lathe coordinate system; According to the auxiliary linear laser instrument orientation of calibration range adjustment guarantee with the intersection point of measuring linear laser instrument in visual range, the frosted scaling board with flatness＜5um places on the rotary table then, guarantees that laser drops on the scaling board;

2) move gauge head according to demarcating planned trajectory, whenever spatial value under an assigned address record lathe coordinate system simultaneously, is noted respectively by spline curve fitting two laser images, and with two intersections of complex curve;

3) move by measure gauge head that linear laser instrument, area array CCD and camera lens forms along X to Z to moving as rectangular node, obtain demarcating The Mesh Point Battle; Utilize the principle match respective straight of least square, and adjust the correction image dot matrix, generate one group of corrected grid chart picture point (N * M), write down gauge head simultaneously along the average gradient K of X to the image straight line lx of the generation of moving;

4) utilize formula $\frac{1}{\mathrm{\δ}}=\frac{1}{\mathrm{\Δ}}k+b,$ Any point p ' on the image _IjPut gauge head along the image straight line lx of X to playing location to the generation of moving ₀Distance be image displacement δ _IjCorresponding with gauge head is the homologue displacement along Z to move distance _IjSatisfy

$\frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b,1<i<M,0\&le;j<N.$

So all the 1＜i that satisfies condition＜M, the calibration point of 0≤j＜N can be set up above-mentioned equation, can set up an indeterminate system of equations like this:

$\left\{\begin{array}{cc}\frac{1}{{\mathrm{\&delta;}}_{20}}=\frac{1}{{\mathrm{\&Delta;}}_{20}}k+b& \\ \&CenterDot;\&CenterDot;\&CenterDot;& \\ \frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b& \\ \&CenterDot;\&CenterDot;\&CenterDot;& \\ \frac{1}{{\mathrm{\&delta;}}_{M-1N-1}}=\frac{1}{{\mathrm{\&Delta;}}_{M-1N-1}}k+b& \end{array}\right.,1<i<M,0\&le;j<N.$

Utilize least square thought,, try to achieve the system of equations least square solution, promptly solve parameter k, b by Household matrixing method.

5) for a certain measurement point, its image coordinate is p ' (x ', y '), calculates its volume coordinate p (Zp ', Xp ') according to the principle of geometric transformation.

At first on the correction image grid, cross some p ', make straight line lx with slope K _{P '}, hand over gauge head respectively along the image straight line lz of Z to mobile generation _jIn a pz ' _kAccording to the basic condition of each point image displacement, utilize calculating parameter k, b reaches $\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b$ Calculate the spatial value Zpz ' at this some place on each straight line _k, then at straight line lx _{P '}Go up corresponding image displacement is compared with the thing displacement, the mean value of asking for this ratio on this straight line then is designated as Rp ', then according to a p ' at straight line lx _{P '}On image displacement δ px ' and mean ratio Rp ' the volume coordinate Xp ' that calculates this point;

Next calculation level p ' puts gauge head along the straight line lx of X to the generation of moving to playing location ₀Distance, i.e. image displacement δ pz ' in the YOZ plane, then according to calculating parameter k, b and $\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b$ Calculate the spatial value Zp ' at this some place;

So fully calculate the corresponding volume coordinate p of picture point p ' (Zp ', Xp ') according to the geometric transformation relation of image-forming principle;

6) measure the multiple dimensioned requirement that calibration result is demarcated the zone in difference in order to satisfy, intact also need of calibrated and calculated further revised calibration result.

According to the 5th step computing method, inverse goes out uncalibrated image net point p ' _IjCorresponding spatial point p _IjCoordinate figure (Zp ' _Ij, Xp ' _Ij), and with corresponding demarcation space lattice point coordinate value (Zp _Ij, Xp _Ij) comparison, that is:

Δx(Zp _ij，Xp _ij)＝Xp _ij-Xp’ _ij，

Δy(Zp _ij，Xp _ij)＝Zp _ij-Zp’ _ij。

Form (as nurbs surface) that so just can the fairing curved surface is set up the error curved surface that is extracted by calibration result:

Δx＝f _x(z，x)，

Δz＝f _z(z，x)。

Like this, the method elder generation computer memory coordinate figure of describing according to the 6th step when concrete conversion the (Zp ', Xp ') calculates the error amount of measurement result then according to the error curved surface that extracts, error amount compensated on the measurement result just can obtain profile measurements more accurately, that is:

Zp＝Zp’+f _x(Zp’，Xp’)，

Xp＝Xp’+f _z(Zp’，Xp’)。

This method has very big lifting with device for classic method on technology and practicality, provide a kind of overcome in the standard component method fully based on the caliberating device of line sweep triangle imaging geometry conversion principle and method can only adapt to ideal situation.And in actual applications, overcome the restriction of multiple factor, make to demarcate to have very strong practicality and operability.This method promotes at the virtual grid method technically with device, utilization is moved along the good actual point of planning by the high-precision motion device of precise closed-loop motion control and the positioning control of software closed loop, and when measuring in addition Real-time and Dynamic compensation of measurement result, overcome the shortcoming that the virtual grid method can't form high precision planning space net point, easier extraction high-precision calibrating net point; The intersection point that calculates two laser rays images by spline curve fitting calculates find intersection than virtual grid method according to straight line and more meets and demarcate actually respectively, and it is also higher to try to achieve the precision of demarcating net point; Scaling method based on the principle of triangulation of gauge head, by asking for the correlating transforms parameter as calibration result, has overcome the deficiency of interpolation calculation method and Function Mapping method fully; Carry out inverse and extract gauge head and measure systematic error in the field depth entirely demarcating dot matrix based on calibration result, measuring accuracy and test specification all are improved.

Description of drawings

Fig. 1 is how much imaging schematic diagrams of laser line scanning feeler triangle

Fig. 2 demarcates corresponding dot matrix

Fig. 3 is an one-piece construction synoptic diagram of the present invention;

Fig. 4 is that gauge head of the present invention is demarcated the realization synoptic diagram;

Fig. 5 is that the present invention moves gauge head is done the formation of rectangular node when motion spotting point in being parallel to the plane of ZOX corresponding grid;

Fig. 6 is the corrected net point system of battle formations of the present invention.

Embodiment

Below in conjunction with accompanying drawing the present invention is described in further detail.

Referring to Fig. 1, two figure are XOY plane and YOZ plane inner laser line sweep geometric triangulation image-forming principle when principle has been showed laser vertical incident respectively.As shown in Figure 1a, ideally, in XOY plane, become similar triangles as trigonometric sum thing triangle, corresponding sides grow up to proportionate relationship, so the space displacement in the XOY plane of measured point moves into the geometric ratio relation with the image position of corresponding imaging point.

And in the YOZ plane, shown in Fig. 1 b, space displacement and image displacement then are a kind of nonlinear relations.According to the triangle sine principle, it is as follows to get the correlation parameter relation:

$\left\{\begin{array}{c}\frac{\mathrm{\&Delta;}}{\sin \mathrm{\&theta;}}=\frac{L}{\sin (\mathrm{\&alpha;}-\mathrm{\&theta;})}\\ \frac{\mathrm{\&delta;}}{\sin \mathrm{\&theta;}}=\frac{l}{\sin (\mathrm{\&beta;}-\mathrm{\&theta;})}\end{array}\right.---\left(1\right)$

In the formula, Δ, δ are respectively space displacement and the corresponding imaging point image displacement in the YOZ plane; L, l are respectively the object distance and the image distance of test reference position; α, β be respectively test reference position image line respectively with the angle of laser projections plane and CCD imaging plane; θ is the angle of test reference position and current measured position image line.

Can further release the space displacement Δ with the pass of corresponding imaging point image displacement δ is:

$\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}\frac{L}{l}\frac{\sin \mathrm{\&beta;}}{\sin \mathrm{\&alpha;}}+(\mathrm{ctg\&alpha;}-\mathrm{ctg\&beta;})\frac{\sin \mathrm{\&beta;}}{l}---\left(2\right)$

By formula (2) as can be seen, after gauge head basic structure was fixed and risen point position and determines, there was linear relationship in the inverse of space displacement Δ with the inverse of corresponding imaging point image displacement δ, formula (2) and can being reduced to:

$\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b---\left(3\right)$

In the formula, k, b are respectively

With

Between the slope and the intercept of linear relationship.

The purpose of laser scanning testing head is exactly to set up on the measuring height direction corresponding relation between the image coordinate point and volume coordinate point in the measurement range by setting up mathematical model, and as shown in Figure 2, wherein straight-line intersection is required demarcation dot matrix.This relation has been arranged, just calculative image coordinate can be scaled corresponding volume coordinate.

Referring to Fig. 3, the inventive system comprises Machine basement 10 and be arranged on the Machine basement 10 mutually vertically by motor-driven along X, Y, Z direction rectilinear motion X-axis precise grating 1, Y-axis precise grating 2 and Z axle precise grating 3, Y-axis precise grating 2 is provided with parallel and measurement board support 10 the linear laser instrument 5 vertical with the XOZ plane of incident angle, also be provided with the area array CCD 6 that has camera lens 7 on the Y-axis precise grating 2, the front end of this area array CCD 6 is provided with optical polarizer 8, area array CCD 6 and camera lens 7 optical axises have angle also to be provided with on (size is measured the depth of field according to the required Measurement Resolution of gauge head and gauge head and determined) Z axle precise grating 3 and the Y-axis precise grating of measuring between the linear laser instrument 52 and assist linear laser instrument 4 with measuring the laser plane that linear laser instrument 5 produces, and the laser plane of auxiliary linear laser instrument 4 intersects with the laser plane of measuring linear laser instrument 5 generations, auxiliary linear laser instrument 4 is mainly used to generate the demarcation impact point at timing signal, auxiliary linear laser instrument 4 positions remain unchanged in calibration process, therefore when in gauge head is being parallel to the plane of XOZ, moving, can think that the intersection point locus that two light beams form is constant on worktable.Be positioned at the lower end of measuring linear laser instrument 5 on the surface of Machine basement 10 and be provided with rotary table 9.In Fig. 4, measure the laser plane vertical table plane incident that linear laser instrument produces, and be parallel to XOY plane, incide the imaging of area array CCD target surface by optical lens through the light after the measured surface diffuse reflection.

Optical polarizer 8 of the present invention is convenient to obtain high-quality image to carry out effective Flame Image Process.Calibration process realizes that on the three-dimensional board of precision board X, Y, Z axle all have the precise grating location, can effectively guarantee the accuracy of timing signal measuring position.Three are moved by close-loop feedback control by motor, have very high bearing accuracy.Add the positioning control of software closed loop when moving simultaneously, can be so that the motion positions precision be more accurate.Auxiliary linear laser instrument 4 is mainly used to demarcating the crossing demarcation point of fixity that forms on the plane at the laser that timing signal generation laser and the linear laser instrument of measurement generate.

Referring to Fig. 4,5, to open and measure linear laser instrument 5 and auxiliary linear laser instrument 4, this moment, two laser rays intersected at a point on worktable 9.On CCD, become image by spline curve fitting two laser rays respectively, and write down two intersections of complex curve.Control X-axis, Z spindle motor make gauge head do the rectangular node motion in being parallel to the plane of ZOX, promptly move along the X axis fixed step size, whenever move a position and write down spatial value and the intersecting point coordinate of measuring linear laser instrument 5 and auxiliary 4 liang of laser rays of linear laser instrument under the lathe coordinate system simultaneously according to planned trajectory; At X after moving certain step number M, control Z axle motion motor, make gauge head move (can not wait to the motion step-length) once along the axial fixed step size of Z, move M step and record (, the word motion of work opposite to moving direction) along the X axis fixed step size then with a preceding X with X according to planned trajectory.Repeat above-mentioned motion process up to covering the rectangular node that sets, promptly at X to each mobile M time, at Z to mobile N time (M, it is definite with measuring accuracy that N measures the depth of field according to CCD).Because the spotting point maintains static, mobile gauge head promptly is equivalent to the spotting point and moves with respect to gauge head.At each grid position of intersecting point, the intersection point image coordinate of corresponding two laser rays forms the image lattice of a correspondence.

Laser line scanning feeler geometric transformation calibration and curved face interpolation correcting method specific implementation and operation steps are as follows:

1) at first utilize precision measurement equipment such as laser interferometer Machine basement 10 to be carried out the calibrating and the record of position degree, linearity, verticality correlated error, carry out the system mechanics accuracy compensation based on 21 error analyses of measurement of coordinates and compensation principle then, guarantee that Machine basement 10 has very high spatial positioning accuracy.The spatial positioning accuracy of Machine basement 10 determines the stated accuracy of gauge head to a great extent;

2) adjust each caliberating device as requested, make each device reach following requirement: guarantee to measure linear laser instrument 5 incident lasers perpendicular to XOY plane under the lathe coordinate system, and be parallel to X-axis; The image visual scope of area array CCD 6 and the decision of camera lens 7 optical axis directions can reach measures depth of field requirement; Adjust the position of area array CCD 6, guarantee ZOX plane parallel under camera lens 7 optical axises and the lathe coordinate system; According to auxiliary linear laser instrument 4 orientation of calibration range adjustment guarantee with the intersection point of measuring linear laser instrument 5 in visual range, the angle direction makes two laser images part as much as possible be arranged in images acquired in demarcating the zone as far as possible.Frosted scaling board with flatness＜5um places on the rotary table 9 then, guarantees that laser drops on the scaling board;

3) move gauge head according to planned trajectory, whenever write down spatial value under the lathe coordinate system at an assigned address (guaranteeing) by precise closed-loop motion control and the positioning control of software closed loop, simultaneously, note respectively by spline curve fitting two laser images, and with two intersections of complex curve.Since commonplace components, as hardware such as laser instrument, camera lens and CCD, the existence of defective own, the real linear laser device laser image distortion that obtains is very serious, and is especially all the more so around image.Therefore can obtain real image dot matrix more accurately by the spline fit curve find intersection;

4) mobile gauge head along X to Z to moving as rectangular node, the demarcation The Mesh Point Battle of similar Fig. 2 that record obtains.From according to the line sweep measuring principle, point-blank the image lattice of spatial point correspondence also should be a straight line, based on this, utilize the principle match respective straight of least square, and adjustment correction image dot matrix, generate one group of corrected grid chart picture point (N * M), write down gauge head simultaneously along the average gradient K of X to the image straight line lx of the generation of moving.Specifically as shown in Figure 6.

5) based on the scaling method of gauge head imaging geometry conversion fully based on the principle of triangulation of gauge head, by asking for the correlating transforms parameter as calibration result.Utilize Fig. 1 (b) and formula (3), any point p ' on the image _IjPut gauge head along the image straight line lx of X to playing location to the generation of moving ₀Distance be image displacement δ _IjCorresponding with gauge head is the homologue displacement along Z to move distance _IjSatisfy

$\frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b,1<i<M,0\&le;j<N.$

So all the 1＜i that satisfies condition＜M, the calibration point of 0≤j＜N can be set up above-mentioned equation.Can set up an indeterminate system of equations like this:

$\left\{\begin{array}{cc}\frac{1}{{\mathrm{\&delta;}}_{20}}=\frac{1}{{\mathrm{\&Delta;}}_{20}}k+b& \\ \&CenterDot;\&CenterDot;\&CenterDot;& \\ \frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b& \\ \&CenterDot;\&CenterDot;\&CenterDot;& \\ \frac{1}{{\mathrm{\&delta;}}_{M-1N-1}}=\frac{1}{{\mathrm{\&Delta;}}_{M-1N-1}}k+b& \end{array}\right.,1<i<M,0\&le;j<N.$

Utilize least square thought,, try to achieve the system of equations least square solution, promptly solve parameter k, b by Household matrixing method.

6) for a certain measurement point, its image coordinate is p ' (x ', y '), calculates its volume coordinate p (Zp ', Xp ') according to the principle of geometric transformation.

At first on the correction image grid, cross some p ', make straight line lx with slope K _{P '}, hand over gauge head respectively along the image straight line lz of Z to mobile generation _jIn a pz ' _kAccording to the basic condition of each point image displacement, utilize calculating parameter k, b and formula (3) are calculated the spatial value Zpz ' at this some place on each straight line _k, then at straight line lx _{P '}Go up corresponding image displacement is compared with the thing displacement, the mean value of asking for this ratio on this straight line then is designated as Rp '.Then according to a p ' at straight line lx _{P '}On phase shift δ px ' and mean ratio Rp ' the volume coordinate Xp ' that calculates this point.

Next calculation level p ' puts gauge head along the straight line lx of X to the generation of moving to playing location ₀Distance, i.e. phase shift δ pz ' in the YOZ plane, then according to calculating parameter k, b and formula (3) are calculated the spatial value Zp ' at this some place.

So fully calculate the corresponding volume coordinate p of picture point p ' (Zp ', Xp ') according to the geometric transformation relation of image-forming principle.

7) measure the multiple dimensioned requirement that calibration result is demarcated the zone in difference in order to satisfy, intact also need of calibrated and calculated further revised calibration result.

According to the 6th step computing method, inverse goes out uncalibrated image net point p ' _IjCorresponding spatial point p _IjCoordinate figure (Zp ' _Ij, Xp ' _Ij), and with corresponding demarcation space lattice point coordinate value (Zp _Ij, Xp _Ij) comparison, that is:

Δx(Zp _ij，Xp _ij)＝Xp _ij-Xp’ _ij，

Δy(Zp _ij，Xp _ij)＝Zp _ij-Zp’ _ij。

Form (as nurbs surface) that so just can the fairing curved surface is set up the error curved surface that is extracted by calibration result:

Δx＝f _x(z，x)，

Δz＝f _z(z，x)。

Like this, the method elder generation computer memory coordinate figure of describing according to the 7th step when concrete conversion the (Zp ', Xp ') calculates the error amount of measurement result then according to the error curved surface that extracts.Error amount compensated on the measurement result just can obtain profile measurements more accurately, that is:

Zp＝Zp’+f _x(Zp’，Xp’)，

Xp＝Xp’+f _z(Zp’，Xp’)。

Claims (4)

1, laser line scanning feeler geometric transformation calibration and surface interpolation correcting device, comprise Machine basement (10) and be arranged on the Machine basement (10) mutually vertically by motor-driven along X, Y, Z direction rectilinear motion X-axis precise grating (1), Y-axis precise grating (2) and Z axle precise grating (3), it is characterized in that: Y-axis precise grating (2) is provided with the linear laser instrument of measurement (5) of parallel and board support (10) vertical with the XOZ plane of incident angle, is positioned at the lower end of measuring linear laser instrument (5) on the surface of Machine basement (10) and is provided with rotary table (9).

2, laser line scanning feeler geometric transformation calibration according to claim 1 and surface interpolation correcting device, it is characterized in that: said Z axle precise grating (3) is parallel with the linear laser instrument of measurement (5), and it is vertical with Y-axis precise grating (2), also be provided with auxiliary linear laser instrument (4) in addition, and the laser plane of auxiliary linear laser instrument (4) intersects with the laser plane of measuring linear laser instrument (5) generation.

3, laser line scanning feeler geometric transformation calibration according to claim 1 and surface interpolation are repaiied device, it is characterized in that: also be provided with the area array CCD (6) that has camera lens (7) on the said Y-axis precise grating (2), the front end of this area array CCD (6) is provided with optical polarizer (8).

4, laser line scanning feeler geometric transformation calibration and curved face interpolation correcting method is characterized in that:

1) adjusts each caliberating device according to calibration request, guarantee to measure linear laser instrument (5) incident laser perpendicular to XOY plane under the lathe coordinate system, and be parallel to X-axis; The image visual scope of area array CCD (6) and the decision of camera lens (7) optical axis direction can reach measures depth of field requirement; Adjust the position of area array CCD 6, guarantee ZOX plane parallel under camera lens (7) optical axis and the lathe coordinate system; According to auxiliary linear laser instrument (4) orientation of calibration range adjustment guarantee with the intersection point of measuring linear laser instrument (5) in visual range, the frosted scaling board with flatness＜5um places on the rotary table (9) then, guarantees that laser drops on the scaling board;

2) move gauge head according to demarcating planned trajectory, whenever spatial value under an assigned address record lathe coordinate system simultaneously, is noted respectively by spline curve fitting two laser images, and with two intersections of complex curve;

3) move by measure gauge head that linear laser instrument (5), area array CCD (6) and camera lens (7) forms along X to Z to moving as rectangular node, obtain demarcating The Mesh Point Battle; Utilize the principle match respective straight of least square, and adjust the correction image dot matrix, generate one group of corrected grid chart picture point (N * M), write down gauge head simultaneously along the average gradient K of X to the image straight line lx of the generation of moving;

4) utilize formula $\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b,$ Any point p ' on the image _IjPut gauge head along the image straight line lx of X to playing location to the generation of moving ₀Distance be image displacement δ _IjCorresponding with gauge head is the homologue displacement along Z to move distance _IjSatisfy

$\frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b,1<i<M,0\&le;j<N.$

So all the 1＜i that satisfies condition＜M, the calibration point of 0≤j＜N can be set up above-mentioned equation, can set up an indeterminate system of equations like this:

$\left\{\begin{array}{c}\frac{1}{{\mathrm{\&delta;}}_{20}}=\frac{1}{{\mathrm{\&Delta;}}_{20}}k+b\\ ...\\ \frac{1}{{\mathrm{\&delta;}}_{\mathrm{ij}}}=\frac{1}{{\mathrm{\&Delta;}}_{\mathrm{ij}}}k+b\\ ...\\ \frac{1}{{\mathrm{\&delta;}}_{M-1N-1}}=\frac{1}{{\mathrm{\&Delta;}}_{M-1N-1}}k+b\end{array}\right.,1<i<M,0\&le;j<N.$

Utilize least square thought,, try to achieve the system of equations least square solution, promptly solve parameter k, b by Household matrixing method.

5) for a certain measurement point, its image coordinate is p ' (x ', y '), calculates its volume coordinate p (Zp ', Xp ') according to the principle of geometric transformation.

At first on the correction image grid, cross some p ', make straight line lx with slope K _{P '}, hand over gauge head respectively along the image straight line lz of Z to mobile generation _jIn a pz ' _kAccording to the basic condition of each point image displacement, utilize calculating parameter k, b reaches $\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b$ Calculate the spatial value Zpz ' at this some place on each straight line _k, then at straight line lx _{P '}Go up corresponding image displacement is compared with the thing displacement, the mean value of asking for this ratio on this straight line then is designated as Rp ', then according to a p ' at straight line lx _{P '}On image displacement δ px ' and mean ratio Rp ' the volume coordinate Xp ' that calculates this point;

Next calculation level p ' puts gauge head along the straight line lx of X to the generation of moving to playing location ₀Distance, i.e. image displacement δ pz ' in the YOZ plane, then according to calculating parameter k, b and $\frac{1}{\mathrm{\&delta;}}=\frac{1}{\mathrm{\&Delta;}}k+b$ Calculate the spatial value Zp ' at this some place;

So fully calculate the corresponding volume coordinate p of picture point p ' (Zp ', Xp ') according to the geometric transformation relation of image-forming principle;

6) measure the multiple dimensioned requirement that calibration result is demarcated the zone in difference in order to satisfy, intact also need of calibrated and calculated further revised calibration result.

According to the 5th step computing method, inverse goes out uncalibrated image net point p ' _IjCorresponding spatial point p _IjCoordinate figure (Zp ' _Ij, Xp ' _Ij), and with corresponding demarcation space lattice point coordinate value (Zp _Ij, Xp _Ij) comparison, that is:

Δx(Zp _ij，Xp _ij)＝Xp _ij-Xp’ _ij，

Δy(Zp _ij，Xp _ij)＝Zp _ij-Zp’ _ij。

Form (as nurbs surface) that so just can the fairing curved surface is set up the error curved surface that is extracted by calibration result:

Δx＝f _x(z，x)，

Δz＝f _z(z，x)。

Like this, the method elder generation computer memory coordinate figure of describing according to the 6th step when concrete conversion the (Zp ', Xp ') calculates the error amount of measurement result then according to the error curved surface that extracts, error amount compensated on the measurement result just can obtain profile measurements more accurately, that is:

Zp＝Zp’+f _x(Zp’，Xp’)，

Xp＝Xp’+f _z(Zp’，Xp’)。

Images