CN102944191B - Method and device for three-dimensional vision measurement data registration based on planar circle target - Google Patents

Abstract

The invention discloses a method for three-dimensional vision measurement data registration based on a planar circle target. The method includes fixing and placing an object to be tested to a reasonable position capable of being observed by a marked binocular vision system, and building a global coordinate system; placing the planar circle target before the object to be tested, moving the marked binocular vision system, shooting elliptic images formed by measurement positions before and after the movement; extracting ellipses from the shot planar circle target images, fitting elliptic equations, and rebuilding circle characteristics of the circle target under the two local measurement coordinates; and constructing and optimizing objective functions according to the circle characteristics and solving a registration matrix. The invention further discloses a device for the three-dimensional vision measurement data registration based on the planar circle target. According to the method and the device, the problem of failure of three-dimensional data registration under the condition that a target portion is shielded in the prior art can be solved, the accuracy of three-dimensional measurement data registration is guaranteed, and reliability of the three-dimensional measurement data registration is improved.

Description

A kind of dimensional visual measurement data joining method based on flat circle target and device

Technical field

The present invention relates to dimensional visual measurement splicing technology, particularly relate to a kind of dimensional visual measurement data joining method based on flat circle target and device.

Background technology

For realizing the measurement to large scale measured object three-dimensional appearance, generally large scale measured object surface being divided into multiple subregion, respectively all subregion being measured from multiple visual angle, and under each local measurement data are spliced to global coordinate system.The precision of three-dimensional data splicing determines the precision that can reach large scale measured object three-dimensional appearance vision measurement, therefore, has important realistic meaning to the research of three-dimensional data joining method.

At present, the method for three-dimensional data splicing mainly contains three kinds:

The first, be expand measurement range by large-scale plants such as precision surface plate, transit, laser trackers, measurement mechanism used in this method is expensive, and measurement range is limited;

The second, by binding mark point in the public view field of adjacent twice measurement of measurement mechanism, utilize wherein non-colinear three points to ask for splicing matrix, it is comparatively loaded down with trivial details that this method also exists the work of pasting and removing mark, and can damage the shortcoming on measured object surface;

The third is iterative closest point method (ICP, Iterative Closest Point) algorithm, but this method to there is interative computation amount large, the problems such as long operational time, and the measured object being not suitable for that surface curvature change do not enrich.

Overcome the deficiency of above three kinds of methods based on flat target calibration method, the unique point that the method utilizes the plane target drone in adjacent twice measurement public view field to provide is to calculate splicing matrix, not only easy to operate, and precision is high, has prospect of the application widely.At present, the plane target drone that conventional is based on checkerboard features, utilizes the coupling of grid angle point to ask for splicing matrix, and then realizes three-dimensional data splicing.But in actual applications, when plane target drone part is blocked, grid angle point often there will be the situation of error hiding, thus causes three-dimensional data to be spliced unsuccessfully.

Summary of the invention

In view of this, fundamental purpose of the present invention is to provide a kind of dimensional visual measurement data joining method based on flat circle target and device, solve the prior art problem that three-dimensional data splicing is failed under plane target drone part is blocked situation, ensure that the precision that 3 d measurement data splices, improve 3 d measurement data splicing reliability.

For achieving the above object, technical scheme of the present invention is achieved in that

The invention provides a kind of dimensional visual measurement data joining method based on flat circle target, the method comprises:

Measured object fixed and is placed on the rational position that the binocular vision system demarcated can observe, and setting up global coordinate system;

Holding plane circle target, before measured object, moves the binocular vision system demarcated, camera plane circle target elliptical image formed by mobile fore-and-aft survey position;

Extract oval from the flat circle target image of shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

According to described round latent structure optimization object function, solve splicing matrix.

In such scheme, described round feature comprises the normal vector of central coordinate of circle and disk.

In such scheme, described fitted ellipse equation, for: the Direct Least Square method in conjunction with RANSAC algorithm (RANSAC, Random Sample Consensus) simulates elliptic equation.

In such scheme, the round feature of described rebuilding plane circle target under the local measurement coordinate system of twice, front and back, for:

Based on the radius value of circle feature on elliptic equation, camera parameters matrix and flat circle target that the Direct Least Square method in conjunction with RANSAC simulates, utilize the round characteristic 3 D method for reconstructing of geometrical intersection, be reconstituted in the round feature under local measurement coordinate system.

In such scheme, describedly solve splicing matrix according to described round latent structure optimization object function, for:

Utilize the round feature homogeneity in space of fore-and-aft survey position of rebuilding, constitution optimization objective function, and according to exterior point ratio in obtaining during RANSAC fitted ellipse equation; The weight that the marginal point set pair arranging coupling is answered, utilizes civilian Burger-Ma Kuaertefa (Levenberg-Marquardt) nonlinear optimization method of row to calculate high-precision splicing matrix.

Present invention also offers a kind of dimensional visual measurement data splicing apparatus realized based on flat circle target, this device comprises elliptical image acquisition module, rebuilds circle characteristic module and splicing matrix computations module; Wherein,

Elliptical image acquisition module, for obtaining flat circle target elliptical image formed by mobile fore-and-aft survey position;

Rebuild circle characteristic module, oval for extracting in the flat circle target elliptical image from shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

Splicing matrix computations module, for according to described round latent structure optimization object function, solves splicing matrix.

In such scheme, described elliptical image acquisition module, reconstruction circle characteristic module, splicing matrix computations module, be arranged in binocular vision system.

Three-dimensional data joining method based on flat circle target provided by the present invention and device, utilize the marginal information that the homogeneity of circle feature and circle enrich; In the process obtaining splicing matrix, by high-precision detection elliptical edge point set, improve the precision of fitted ellipse equation, so improve splicing matrix ask for precision.And, the present invention utilizes round feature to be blocked in part also can by the advantage accurately extracted and mate in situation, precision and the reliability of three-dimensional data splicing can be improved, thus solve in prior art and adopt plane grid Bar Method to be blocked the problem that three-dimensional data splicing is failed in situation in part, ensure that the precision that 3 d measurement data splices, improve 3 d measurement data splicing reliability.

In addition, because circle target is not vulnerable to the impact of partial occlusion, so be more suitable for the application under various complicated site environment.

Accompanying drawing explanation

Fig. 1 is the dimensional visual measurement data split-join model schematic diagram that the present invention is based on flat circle target;

Fig. 2 is the dimensional visual measurement data joining method schematic flow sheet that the present invention is based on flat circle target;

Fig. 3 is the structural representation of flat circle target of the present invention;

Fig. 4 is the composition structural representation that the present invention realizes the dimensional visual measurement data splicing apparatus based on flat circle target;

Fig. 5 is the present invention one instantiation experimental system figure;

Fig. 6 is the three-dimensional point cloud atlas of local measurement position of the present invention;

Fig. 7 is the comparison diagram that the present invention is spliced acquired results and ICP method and spliced acquired results;

Fig. 8 is the scene schematic diagram of right video camera photographing section shielded image;

Fig. 9 is the three-dimensional point set schematic diagram of grid target in partial occlusion situation;

Figure 10 is partial occlusion situation upper/lower positions 1 Point set matching figure three-dimensional with position 2 grid target.

Embodiment

For a better understanding of the present invention, first the ultimate principle of dimensional visual measurement data splicing is introduced, Fig. 1 is the dimensional visual measurement data split-join model schematic diagram based on flat circle target, as shown in Figure 1, sets up camera coordinate system O respectively at local measurement position k and position k+1 _ckx _cky _ckz _ckand O _ck+1x _ck+1y _ck+1z _ck+1, global coordinate system is based upon the camera coordinate system O of position 1 _c1x _c1y _c1z _c1under; In the video camera moving process of twice local measurement in front and back, flat circle target 11 keeps motionless, that is: the round feature on flat circle target 11 is constant under global coordinate system; Here, described round feature comprises the normal vector of central coordinate of circle and disk;

Suppose that circle is characterized as at camera coordinate system O _ckx _cky _ckz _ckand O _ck+1x _ck+1y _ck+1z _ck+1result be respectively with suppose camera coordinate system O _ckx _cky _ckz _ckand O _ck+1x _ck+1y _ck+1z _ck+1splicing matrix be M _{k+1, k}, M _{k+1, k}by 3 × 3 orthogonal rotation matrix R _{k+1, k}with 3 × 1 translation vector t _{k+1, k}composition, its expression formula is as shown in formula (1):

$M_{k+1,k}=\left[\begin{array}{cc}{R}_{k+1,k}& t_{k+1,k}\\ O_{1\×3}& 1\end{array}\right]---\left(1\right)$

In the measuring process of local measurement position k to position k+1, according to the homogeneity of round feature under global coordinate system of flat circle target, the following relation of known existence:

${\stackrel{\→}{v}}_{k+1}=R_{k+1,k}\·{\stackrel{\→}{v}}_k---\left(2\right)$

p _k+1＝R _k+1，k·p _k+t _k+1，k(3)

R can be calculated by formula (2) and formula (3) _{k+1, k}, t _{k+1, k}, and then camera coordinate system transform matrix M can be calculated according to formula (1) _{k+1, k}.

Be global measuring coordinate system with the camera coordinates at position 1 place, in k+1 place, position measurement data through conversion M _2,1m _3,2m _{k+1, k}, global coordinate system O can be unified _c1x _c1y _c1z _c1under.In like manner, 3-D scanning gauge head all can unify global coordinate system O in the local measurement data of all positions _c1x _c1y _c1z _c1under, thus obtain the three-dimensional appearance data of whole measured object, complete the splicing of dimensional visual measurement data; Wherein, 3-D scanning gauge head comprises above-mentioned two video cameras and a projector, for obtaining measured object three-dimensional appearance data.Usually, binocular vision system 12 comprises two video cameras, and described two video cameras are for taking stitching image.

According to the ultimate principle of dimensional visual measurement data splicing, basic thought of the present invention is: fixed by measured object and be placed on the rational position that the binocular vision system demarcated can observe, and setting up global coordinate system; The binocular vision system demarcated also is moved, camera plane circle target elliptical image formed by mobile fore-and-aft survey position before flat circle target is placed on measured object; Extract oval from the flat circle target image of shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back; Splicing matrix is solved according to described round latent structure optimization object function.

Below in conjunction with the drawings and specific embodiments, the technical solution of the present invention is further elaborated.

Fig. 2 is the dimensional visual measurement data joining method schematic flow sheet that the present invention is based on flat circle target, and as shown in Figure 2, described method comprises:

Step 201, fixes measured object and is placed on the rational position that the binocular vision system demarcated can observe, and setting up global coordinate system;

Concrete, the demarcation mode that demarcation binocular vision system adopts can see " AFlexible New Technique for Camera Calibration [J] (IEEE Trans.Pattern Analysisand Machine Intelligence, the Nov.2000) " of Z.Y.ZHANG;

Describedly measured object is placed on the rational position that the binocular vision system demarcated can observe and is: measured object is divided into some subregions, the binocular vision system that placement has been demarcated is at position 1 place, under global coordinate system being arranged on a camera coordinate system in binocular vision system, be designated as O simultaneously _c1x _c1y _c1z _c1; Here, binocular vision system comprises two video cameras;

Concrete, measured object being divided into how many sub regions is determine according to the pattern of measured object, and final purpose the overall pattern of measured object can be stitched together by binocular vision system; Preferably, 1/3rd overlaps are had between every two sub regions.

Step 202, holding plane circle target, before measured object, moves the binocular vision system demarcated, camera plane circle target elliptical image formed by mobile fore-and-aft survey position;

Here, be before flat circle target is placed on measured object subregion;

Concrete, Fig. 3 is the flat circle target schematic diagram adopted in the embodiment of the present invention, as shown in Figure 3, the plane of this flat circle target is a square, concentric circles in target is made up of three circles, and diameter is 95mm, 65mm, 35mm from big to small successively, and foursquare length of side size is 125mm; Wherein, concentric circles, for the accurate extraction of the normal vector of space central coordinate of circle and disk; The square of outside, calculates the correspondence of round unique point for the determination of target co-ordinates system and nonlinear optimization.

Here, due to the globality of circle feature on flat circle target, flat circle target is made to have in the sightless situation of circle characteristic, the advantage still can accurately extracted, therefore, only need when mobile biocular systems shooting the round target feature photographing part, substantially increase the convenience of binocular vision system movement.Because circle is oval at the image of space projection, so camera plane circle target can obtain elliptical image.

Step 203, extracts oval from the flat circle target image of shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

Wherein, described round feature comprises the normal vector of central coordinate of circle and disk;

Concrete, calculate in Fig. 1 with for image formed by the video camera imaging plane of flat circle target in binocular vision system, by obtaining oval point set after the method process such as the sub-pixel edge detection based on the canny operator improved, and by elliptical shape feature filtering disordered point, then in conjunction with RANSAC, Direct Least Square method simulate elliptic equation; When partial occlusion, the shelter marginal point that edge extracting goes out does not belong to oval circular arc.

Here, the point on described oval circular arc is called interior point; The marginal point of shelter is called exterior point; Existence due to these exterior points can cause the precise decreasing of Direct Least Square method fitted ellipse equation, therefore, utilizes RANSAC to distinguish interior point and the exterior point of edge ellipse, thus removes the impact of exterior point, improves the precision of ellipse fitting.

Concrete, the circle of described rebuilding plane circle target under the local measurement coordinate system of twice, front and back is characterized as: based on the radius value of circle feature on elliptic equation, camera parameters matrix and flat circle target that the Direct Least Square method in conjunction with RANSAC simulates, utilize the round characteristic 3 D method for reconstructing of geometrical intersection, rebuild the round feature under local measurement coordinate system, that is: ask under the k local measurement coordinate system of position, three-dimensional central coordinate of circle and the normal vector of flat circle target being justified feature are respectively in like manner, try to achieve

Step 204, according to described round latent structure optimization object function, solves splicing matrix;

Here, the round feature homogeneity in space of fore-and-aft survey position is utilized, constitution optimization objective function;

Concrete, this step is exactly calculate the rotation matrix R shown in Fig. 1 and translation vector t; Because rotation matrix has three degree of freedom, therefore, at least need to choose two to vector, utilize this two couple vector and multiplication cross vectorial, calculate rotation matrix R; But this linear computational method precision is lower, is generally used as the initial value of nonlinear computation and global optimization.

When known more than three to vector time, rotation matrix R can try to achieve by least-squares linear regression, and computing formula is as shown in formula (4):

R＝N _k+1·N _k ^T·(N _k·N _k ^T) ^-1(4)

In formula (4), $N_k={({\stackrel{\→}{v}}_{k,1},{\stackrel{\→}{v}}_{k,2},....,{\stackrel{\→}{v}}_{k,n})}^T;$ $N_{k+1}={({\stackrel{\→}{v}}_{k+1,1},{\stackrel{\→}{v}}_{k+1,2},....,{\stackrel{\→}{v}}_{k+1,n})}^T;$ N is the right number of the normal vector of space circular plane of coupling, and n > 3.

After trying to achieve rotation matrix R by formula (4), translation vector t can be tried to achieve by formula (5):

t＝p _k+1-R·p _k(5)

Based on the normal vector pair of the space circular plane of coupling, rotation matrix R Rodrigues vector form is represented, and obtains the ratio value of interior exterior point when going out ellipse according to RANSAC the Fitting Calculation, it is right that vector is set with corresponding weight w _i, constitution optimization calculates the least square problem of rotation matrix R as shown in formula (6):

$\min F=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{\left|\right|{\stackrel{\→}{v}}_{k+1,i}-R\·{\stackrel{\→}{v}}_{k,i}\left|\right|}^2---\left(6\right)$

Levenberg-Marquardt nonlinear optimization method is used to formula (6), calculates the rotation matrix R after optimization.

On the basis of the rotation matrix R that the translation vector t calculated at formula (5) and formula (6) calculate, utilize the space circle edge point set of coupling and the central coordinate of circle of space circle, constitution optimization objective function, calculates splicing matrix, as shown in formula (7):

$\min F=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{\left|\right|p_{k+1}-\left[\begin{array}{cc}R& t\\ O_{1\×3}& 1\end{array}\right]\·p_k\left|\right|}^2---\left(7\right)$

Same, rotation matrix R Rodrigues vector form is represented, and according to exterior point ratio in obtaining during RANSAC fitted ellipse equation, the edge point set p of coupling is set _kand p _k+1corresponding weight w _iutilize Levenberg-Marquardt nonlinear optimization method (M.Galassi, J.Davies, J.Theiler, G.Jungman, M.Booth, and F.Rossi.GNU Scientific Library Reference Manual, Network Theory, Aug.2006), calculate high-precision rotation matrix R and translation vector t; And then to calculate according to formula (1) and splice matrix accurately, complete the splicing of dimensional visual measurement data.

Fig. 4 is the composition structural representation that the present invention realizes the dimensional visual measurement data splicing apparatus based on flat circle target, and as shown in Figure 4, this splicing apparatus comprises: elliptical image acquisition module 40, reconstruction circle characteristic module 41 and splicing matrix computations module 42; Wherein,

Elliptical image acquisition module 40, for obtaining flat circle target elliptical image formed by mobile fore-and-aft survey position;

Rebuild circle characteristic module 41, oval for extracting in the flat circle target elliptical image from shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

Splicing matrix computations module 42, for according to described round latent structure optimization object function, solves splicing matrix.

Wherein, described round feature comprises the normal vector of central coordinate of circle and disk.

Concrete, described reconstruction circle characteristic module 41 can be used for realizing step 203; Described splicing matrix computations module 42 can be used for realizing step 204, does not repeat them here.

Further, elliptical image acquisition module 40, comprising at acquisition flat circle target before elliptical image formed by mobile fore-and-aft survey position:

Measured object fixed and is placed on the rational position that the binocular vision system demarcated can observe, setting up global coordinate system;

Before flat circle target is placed on measured object, when moving the binocular vision system demarcated, camera plane circle target elliptical image formed by mobile fore-and-aft survey position.

Described realization, based on elliptical image acquisition module 40, reconstruction circle characteristic module 41, the splicing matrix computations module 42 of the dimensional visual measurement data splicing apparatus of flat circle target, can be arranged in binocular vision system.

Realize effect in order to what joining method of the present invention was described better, flat circle target method of the present invention can be carried out Experimental comparison with ICP method and plane grid target method, respectively to evaluate the actual effect that flat circle target method is spliced for three-dimensional data.Build experimental system as shown in Figure 5, adopt two AVT F302b type ccd video cameras, each video camera target surface is of a size of 2/3inch, and resolution is 1028pixel × 960pixel, in conjunction with two schnider 12mm camera lens composition binocular vision systems, form 3-D scanning gauge head with projector.

A. the contrast of flat circle target method and ICP method:

Take guided missile model as measured object, 3-D scanning gauge head is utilized to carry out twice measurement to it respectively, as shown in Figure 6, Fig. 6 (a) and Fig. 6 (b) are respectively the point cloud chart obtained twice measuring position, wherein, Fig. 6 (a) is position 2 for position 1, Fig. 6 (b).

Fig. 7 (a) splices based on flat circle target method the point cloud chart obtained for what adopt the present invention to propose, Fig. 7 (b) splices for adopting ICP method the point cloud chart obtained, as can be seen from Figure 7, adopt the splicing of flat circle target method to obtain good splicing effect, the three dimensional point cloud that two measuring positions obtain obtains correct splicing; And adopt ICP method, then there is the situation of splicing mistake; In addition, 213.6 seconds consuming time of ICP method splicing, and flat circle target method of the present invention is spliced 6.8 seconds consuming time.

ICP method occurs that the reason of splicing mistake is: measured object superficial makings is abundant, feature is few, and as can be seen here, the measured object that ICP method is not enriched for superficial makingss such as being similar to cylindrical surface is also inapplicable.As can be seen from splicing experimental result contrast, the present invention propose based on flat circle target calibration method in splicing effect and on the algorithm process time, be all better than ICP method.

B. the contrast of flat circle target method and plane grid target method:

In unscreened situation, under adopting experiment condition as shown in Figure 5, the splicing precision of three-dimensional data joining method on x, y, z direction of principal axis based on flat circle target is respectively 0.067mm, 0.035mm, 0.134mm; Utilize plane grid target method to calculate the camera coordinate system splicing matrix of same position relationship, the splicing precision obtained on x, y, z direction of principal axis is respectively 0.057mm, 0.032mm, 0.103mm.As can be seen here, this experimental result illustrates, the flat circle target method that the present invention proposes is in unscreened situation, suitable with plane grid target method precision.

Under flat circle target moiety is blocked situation, here, can utilize and place shelter to realize partial occlusion before splicing target, under same employing experiment condition as shown in Figure 5, the image that right video camera in binocular vision system is taken in position 1 and position 2 as shown in Figure 8, Fig. 8 (a) is for right video camera is at the shooting image of position 1, and Fig. 8 (b) is for right video camera is at the shooting image of position 2.

Under flat circle target moiety is blocked situation, based on plane grid target method in two measuring positions (position 1 and position 2), the three-dimensional reconstruction result of grid angle point as shown in Figure 9, Fig. 9 (a) is the three-dimensional point set of position 1 grid target, and Fig. 9 (b) is the three-dimensional point set of position 2 grid target.Due to the uncertainty of circumstance of occlusion, coupling between the three-dimensional point reconstructed also exists the multiple possibility as shown in Figure 10 (a) ~ Figure 10 (d), cause the uncertain of coupling because block, plane grid Bar Method causes three-dimensional data cannot realize correct splicing; Under kindred circumstances, what the present invention proposed does not affect by partial occlusion based on flat circle target calibration method, and the splicing precision on x, y, z direction of principal axis is respectively 0.062mm, 0.037mm, 0.168mm; Experimental result shows, the flat circle target method that the present invention proposes, when partial occlusion, is better than plane grid target method, is suitable for the three-dimensional data splicing under complicated site environment.

The above, be only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention.

Claims (6)

1., based on a dimensional visual measurement data joining method for flat circle target, it is characterized in that, described method comprises:

Measured object fixed and is placed on the rational position that the binocular vision system demarcated can observe, and setting up global coordinate system;

Holding plane circle target, before measured object, moves the binocular vision system demarcated, camera plane circle target elliptical image formed by mobile fore-and-aft survey position;

Extract oval from the flat circle target image of shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

According to described round latent structure optimization object function, solve splicing matrix;

Wherein, described fitted ellipse equation, for:

Direct Least Square method in conjunction with RANSAC algorithm RANSAC simulates elliptic equation.

2. method according to claim 1, is characterized in that, described round feature comprises the normal vector of central coordinate of circle and disk.

3. method according to claim 1, is characterized in that, the round feature of described rebuilding plane circle target under the local measurement coordinate system of twice, front and back, for:

Based on the radius value of circle feature on elliptic equation, camera parameters matrix and flat circle target that the Direct Least Square method in conjunction with RANSAC simulates, utilize the round characteristic 3 D method for reconstructing of geometrical intersection, be reconstituted in the round feature under local measurement coordinate system.

4. method according to claim 1, is characterized in that, describedly solves splicing matrix according to described round latent structure optimization object function, for:

Utilize the round feature homogeneity in space of fore-and-aft survey position of rebuilding, constitution optimization objective function, and according to exterior point ratio in obtaining during RANSAC fitted ellipse equation; The weight that the marginal point set pair arranging coupling is answered, utilizes the civilian Burger-Ma Kuaertefa Levenberg-Marquardt nonlinear optimization method of row to calculate high-precision splicing matrix.

5. realize the dimensional visual measurement data splicing apparatus based on flat circle target, it is characterized in that, this device comprises elliptical image acquisition module, rebuilds circle characteristic module and splicing matrix computations module; Wherein,

Elliptical image acquisition module, for obtaining flat circle target elliptical image formed by mobile fore-and-aft survey position;

Rebuild circle characteristic module, oval for extracting in the flat circle target elliptical image from shooting, and fitted ellipse equation, the round feature of rebuilding plane circle target under the local measurement coordinate system of twice, front and back;

Splicing matrix computations module, for according to described round latent structure optimization object function, solves splicing matrix;

Wherein, described fitted ellipse equation, for:

Direct Least Square method in conjunction with RANSAC simulates elliptic equation.

6. device according to claim 5, is characterized in that, described elliptical image acquisition module, reconstruction circle characteristic module, splicing matrix computations module, be arranged in binocular vision system.