CN1508512A - Overall calibrating method for multi-vision sensor detecting system - Google Patents

Abstract

The invention refers to the improvement of sensor scaling method in structured light three dimension vision measurement. The steps are: system constructing, collecting scaling spot, selecting coordinate system, setting target, collecting target picture, calculating the three dimension object coordinate of the special spot, extracting the two-dimension image coordinate, repeating to extract the coordinate, scaling sensor, storing the scaling results.

Description

A kind of multiple vision sensor detection system global calibration method

Technical field

The invention belongs to field of measuring technique, relate to improvement transducer calibration method in the structured light three-dimensional vision detection.

Background technology

In many visible detection methods, structured light three-dimensional vision detects and is widely used in the integrality of workpiece, the measurement of surface smoothness; The automatic detection of microelectronic component (IC chip, PC plate, BGA) etc.; The detection of soft, frangible parts; The detection of various mould 3D shapes; Robotic vision guiding etc.Vision detection technology with its wide range, big visual field, measuring speed is fast, optical strip image is easy to extract and characteristics such as degree of precision have obtained using more and more widely in industrial environment.The large-sized object of forming by the vision sensor array (as aviation component etc.) space three-dimensional size multiple vision sensor detection system the most attractively.This system flexibility is good, is measured as contactlessly, and dynamic response is fast, can satisfy to produce " beat " short requirements in a large number, and whole measuring process is increasingly automated.Make whole multiple vision sensor detection system become an independently measuring system, must demarcate total system, the relative position relation of each sensor and measured workpiece coordinate system in promptly definite system, this work becomes global calibration.Detect in the application at structured light three-dimensional vision, the global calibration method is the research contents of a key always.At present, the global calibration of multiple vision sensor detection system mainly is " electronic theodolite is to method ".This method is to utilize two electronic theodolites to form the three-dimensional three-dimensional coordinates measurement of binocular system, and space high accuracy three-dimensional calibration point is provided.Because electronic theodolite moves flexibly, the precision height, measurement range is big, is well suited for on-site proving, thereby becomes the main means of present multiple vision sensor detection system global calibration.In fact the global calibration of multiple vision sensor detection system is exactly to determine the position relation of each sensor unit with respect to the workpiece calibration coordinate system, and like this, the data that systematic survey goes out are only the data with engineering significance.Utilize electronic theodolite that the method for carrying out global calibration specifically is divided into " coordinate united law of the same name " and " intermediate coordinates united law " again at present.

The demarcating steps of coordinate united law of the same name is: the inner parameter of at first in the laboratory, finishing video camera demarcate and system in the part demarcation of single-sensor.At this moment, each position component of video camera and sensor must can not be changed by good fixing, to guarantee the validity of institute's calibrating parameters.Then, to measure field, all the sensors is fixed, build total system after, by transit system and each sensor are measured the calibration point overall measurement coordinate system under simultaneously again, realize that the overall situation of sensor is unified.

The demarcating steps of intermediate coordinates united law is: the inner parameter of at first finishing video camera in the laboratory is demarcated.At this moment, each position component of video camera and sensor must can not be changed by good fixing, to guarantee the validity of institute's calibrating parameters.Secondly,, all the sensors is fixed to measure field, build total system after, determine position and the direction of video camera under a certain demarcation target coordinate system of each sensor.The 3rd, by transit system is determined position and the direction of target coordinate system under workpiece coordinate system, realize the unification of the overall situation.

The deficiency that " coordinate united law of the same name " and " intermediate coordinates united law " exists is: the first, partial parameters is after the laboratory is demarcated, in the process of on-the-spot transportation, because the influence of the factor that vibrations etc. can't overcome, the relative position of each parts of sensor may change, thereby makes the parameter distortion of having demarcated.The second, pass through the repeatedly conversion of coordinate system in the unified process of the overall situation, caused the loss of precision inevitably.The deficiency that the global calibration method of existing multiple vision sensor detection system exists makes the accuracy of detection of total system be difficult to the level that reaches higher.

Summary of the invention

Technical matters to be solved by this invention is: overcome the deficiency that existing method exists, provide a kind of precision higher, be fit to the global calibration method of the multisensor vision-based detection system of on-site proving, further improve multiple vision sensor detection system measuring accuracy.

Technical solution of the present invention is: a kind of multiple vision sensor detection system global calibration method is characterized in that demarcating according to following steps:

(1) in measure field whole multiple vision sensor detection system is built, determine that promptly the position of each vision sensor is also fixed, determine that simultaneously the position of transit is also fixed;

(2) in measure field, place the standard target chi that is used to demarcate transit in position, gather the calibration point of sufficient amount, transit is built up measuring system;

(3) according to the concrete measurement requirement of different measured workpieces, selected workpiece coordinate system uses transit that system has been set up workpiece coordinate system;

(4) a demarcation target with distinct characteristic point is laid in place, should be made each vision sensor and transit in the system can both normally use this target; Target is made up of two planes, and its angle theta is 45 °≤θ≤135 °, on two target faces the unique point that sets in advance is arranged, and the target face is one of following structure:

A, target first have the rectangular block of the black of two row triplex row projectioies on the target face, the quantity of rectangular block is 2～100 in every row, and the spacing of rectangular block is 10～200mm, and four summits of each rectangular block are unique point;

B, target second have into the circular hole of arranged on the target face, the line number of circular hole and columns are 3～100, and the center of adopting circular hole is as unique point;

C, target the third have into the crosshair of arranged on the target face, crosshair line number and columns are 3～100, with the point of crossing of crosshair as unique point;

(5) open the power supply of multiple vision sensor detection system, gather the image of target, store in the computing machine by the vision sensor of multiple vision sensor detection system;

(6) by transit the some unique points on the target are aimed at, read value and the record of its level and angle of pitch α and β, calculate then this unique point under workpiece coordinate system the three-dimensional article coordinate and store in the computing machine;

(7) the feature point extraction algorithm that employing is relevant obtains the two-dimensional image coordinate of this unique point correspondence on the target image and stores in the computing machine;

(8) needs of demarcating all vision sensors are satisfied in the work of repetition (6), (7) up to the calibration point that is obtained;

(9) utilize the calibration point that obtains that each vision sensor of whole multiple vision sensor detection system is demarcated, obtain the position relation of each vision sensor, finish global calibration with respect to workpiece coordinate system;

(10) the storage calibration result uses in order to follow-up measurement.

The advantage of the inventive method is: the first, realized the unification at demarcation of measure field intrinsic parameters of the camera and sensor global calibration, the loss of significance of having avoided the variation of the sensor parameters that may cause in sensor handling process and the installation process and having produced, guarantee the consistance of demarcation state and duty, helped improving the precision of whole vision detection system.The second, reduce the coordinate conversion number of times, thereby reduced the loss of significance that is brought by the coordinate system conversion.

Description of drawings

Fig. 1 is a kind of synoptic diagram of demarcating target.

Fig. 2 is the pictorial diagram of structured light vision sensor.Among Fig. 2, the 1st, housing, the 2nd, CCD form, the 3rd, laser emitting mouth.

Embodiment

Below the inventive method is described in further details.The inventive method claims " the unique method of world coordinates " again, the basic ideas that it carries out multiple vision sensor detection system global calibration are: in measure field, directly measuring system is obtained world's workpiece coordinate system unique point down as calibration point, be used for the inner parameter of while calibrating camera and the external parameter of sensor by transit.Like this, after each transducer calibration finishes in the total system, just obtain the inner parameter of each video camera, and realized that simultaneously the overall situation of each sensor is unified, promptly finished the global calibration of system.

The concrete steps that use the present invention carries out multiple vision sensor detection system global calibration are as follows:

(1) in measure field whole multiple vision sensor detection system is built, determine that promptly the position of each vision sensor is also fixed, determine that simultaneously the position of transit is also fixed.

(2) in measure field, place the standard target chi that is used to demarcate transit in position, gather the calibration point of sufficient amount, transit is built up measuring system; The concrete grammar of setting up transit is a known technology, can be referring to Luo Mingzhu " multisensor vision detection system and applied research thereof ", University Of Tianjin's PhD dissertation, 1996.

(3) according to the concrete measurement requirement of different measured workpieces, selected workpiece coordinate system uses transit that system has been set up workpiece coordinate system.The foundation of workpiece coordinate system is generally carried out according to its design drawing, promptly workpiece on the following various sizes of the workpiece coordinate system of being set up and tolerance and drawing, indicated in full accord.

(4) a demarcation target with distinct characteristic point is laid in place, should be made each vision sensor and transit in the system can both normally use this target.Generally, make each vision sensor and transit in the system can both " see " target, can obtain from target and demarcate needed abundant unique point.Target is made up of two planes, and its angle theta is 45 °≤θ≤135 °, on two target faces the unique point that sets in advance is arranged, and the target face is one of following structure:

A, target first have the rectangular block of the black of two row triplex row projectioies on the target face, the quantity of rectangular block is 2～100 in every row, and the spacing of rectangular block is 10～200mm, and four summits of each rectangular block are unique point.

B, target second have into the circular hole of arranged on the target face, the line number of circular hole and columns are 3～100, and the center of adopting circular hole is as unique point.

C, target the third have into the crosshair of arranged on the target face, crosshair line number and columns are 3～100, with the point of crossing of crosshair as unique point.

As shown in Figure 1, this is a kind of embodiment of target first, and the angle point of black square is a unique point on the target face, θ=90 °.

(5) open the power supply of multiple vision sensor detection system, gather the image of target, store in the computing machine by the vision sensor of multiple vision sensor detection system.

(6) by transit the some unique points on the target are aimed at, read value and the record of its level and angle of pitch α and β, calculate then this unique point under workpiece coordinate system the three-dimensional article coordinate and store in the computing machine.Concrete technical method can be referring to Luo Mingzhu " multisensor vision detection system and applied research thereof ", University Of Tianjin's PhD dissertation, 1996.

(7) the feature point extraction algorithm that employing is relevant obtains the two-dimensional image coordinate of this unique point correspondence on the target image and stores in the computing machine.The feature point extraction algorithm is referring to the article " a kind of comprehensive angle point and edge finder " [A combined corner andedge detector] of C.G.Harris and M.J.Stephens, the 4th Alvey visual conference collection of thesis, Manchester, the 147th～151 page, 1988 years.[Proceedings Fourth Alvey Vision Conference, Manchester.pp 147～151,1988] and the article " straightedge limit characteristic points automatic extraction method in the structured light vision detection " of Wei Zhenzhong, Zhang Guangjun, 2002 national photoelectric technology academic exchange meeting paper collection, the 196th～199 page.

(8) needs of demarcating all vision sensors are satisfied in the work of repetition (6), (7) up to the calibration point that is obtained.

(9) utilize the calibration point that obtains that each vision sensor of whole multiple vision sensor detection system is demarcated, obtain the position relation of each vision sensor, finish global calibration with respect to workpiece coordinate system.

(10) the storage calibration result uses in order to follow-up measurement.

Embodiment

Utilize method of the present invention, the multiple vision sensor detection system that is made of as shown in Figure 2 five structured light vision sensors is demarcated.The parameter that needs to demarcate is suc as formula shown in [1] and the formula [2]:

$\left[\begin{array}{c}\mathrm{\ω}\·X\\ \mathrm{\ω}\·Y\\ \mathrm{\ω}\end{array}\right]=\left[\begin{array}{cccc}f& 0& 0& 0\\ 0& f& 0& 0\\ 0& 0& 1& 0\end{array}\right]\left[\begin{array}{cc}R& t\\ 0^T& 1\end{array}\right]\·\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]---\left[1\right]$

Wherein $R=\left[\begin{array}{ccc}{r}_1& r_2& r_3\\ r_4& r_5& r_6\\ r_7& r_8& r_9\end{array}\right]$ Be rotation matrix, $t=\left[\begin{array}{c}{t}_x\\ t_y\\ t_z\end{array}\right]$ Be translation vector, ω is a scale factor, is a constant.

The equation of optic plane equations can be described as: ax _w+ by _w+ cz _w+ d=0 [2]

The result who demarcates is as follows:

(1) sensor one:

$\·R=\left[\begin{array}{ccc}0.347027& 0.244612& -0.905393\\ -0.627354& 0.778147& -0.030224\\ 0.697136& 0.578491& 0.423497\end{array}\right]t=\left[\begin{array}{c}55.099585\\ 9.309396\\ 491.199576\end{array}\right]$

The structured light plane equation coefficient

[a?b?c?d]＝[0.003167?0.001147?-0.014804?1.390404]

(2) sensor two:

$\·R=\left[\begin{array}{ccc}0.180767& 0.352066& -0.918353\\ -0.869409& 0.493760& 0.018158\\ 0.459839& 0.795142& 0.395344\end{array}\right]t=\left[\begin{array}{c}11.624459\\ 22.440675\\ 544.071467\end{array}\right]$

The structured light plane equation coefficient

[a?b?c?d]＝[-0002036?0.000763?-0.018162?0.598640]

(3) sensor three:

$\·R=\left[\begin{array}{ccc}0.183649& 0.330879& -0.925631\\ -0.820275& 0.570485& 0.41182\\ 0.541685& 0.751708& 0.376180\end{array}\right]t=\left[\begin{array}{c}67.876501\\ 8.940593\\ 515.987783\end{array}\right]$

The structured light plane equation coefficient

[a?b?c?d]＝[0.171965?-0.134121?2.392007?-228.244697]

(4) sensor four:

$\·R=\left[\begin{array}{ccc}0.251127& 0.301105& -0.919930\\ -0.776509& 0.630079& -0.005742\\ 0.577900& 0.715776& 0.392041\end{array}\right]t=\left[\begin{array}{c}56.021709\\ 14.672523\\ 509.689664\end{array}\right]$

The structured light plane equation coefficient

[a?b?c?d]＝[-0.049510?0.175063?-3.77078035?0.999684]

(5) sensor five:

$\·R=\left[\begin{array}{ccc}0.289718& 0.302576& -0.908026\\ -0.750725& 0.660327& -0.019492\\ 0.593696& 0.687325& 0.418460\end{array}\right]t=\left[\begin{array}{c}47.350343\\ -1.241391\\ 494.465622\end{array}\right]$

The structured light plane equation coefficient

[a?b?c?d]＝[-0.053936?-0.153525?2.731544?-248.999779]

Utilizing the above detection system of being made up of five structured light vision sensors of demarcating, is that the ruler of 1021.413mm is measured to standard length, and measuring accuracy has reached 0.373mm.

Claims (1)

1, a kind of multiple vision sensor detection system global calibration method is characterized in that demarcating according to following steps:

(1) in measure field whole multiple vision sensor detection system is built, determine that promptly the position of each vision sensor is also fixed, determine that simultaneously the position of transit is also fixed;

(2) in measure field, place the standard target chi that is used to demarcate transit in position, gather the calibration point of sufficient amount, transit is built up measuring system;

(3) according to the concrete measurement requirement of different measured workpieces, selected workpiece coordinate system uses transit that system has been set up workpiece coordinate system;

(4) a demarcation target with distinct characteristic point is laid in place, should be made each vision sensor and transit in the system can both normally use this target; Target is made up of two planes, and its angle theta is 45 °≤θ≤135 °, on two target faces the unique point that sets in advance is arranged, and the target face is one of following structure:

A, target first have the rectangular block of the black of two row triplex row projectioies on the target face, the quantity of rectangular block is 2～100 in every row, and the spacing of rectangular block is 10～200mm, and four summits of each rectangular block are unique point;

B, target second have into the circular hole of arranged on the target face, the line number of circular hole and columns are 3～100, and the center of adopting circular hole is as unique point;

C, target the third have into the crosshair of arranged on the target face, crosshair line number and columns are 3～100, with the point of crossing of crosshair as unique point;

(5) open the power supply of multiple vision sensor detection system, gather the image of target, store in the computing machine by the vision sensor of multiple vision sensor detection system;

(6) by transit the some unique points on the target are aimed at, read value and the record of its level and angle of pitch α and β, calculate then this unique point under workpiece coordinate system the three-dimensional article coordinate and store in the computing machine;

(7) the feature point extraction algorithm that employing is relevant obtains the two-dimensional image coordinate of this unique point correspondence on the target image and stores in the computing machine;

(8) needs of demarcating all vision sensors are satisfied in the work of repetition (6), (7) up to the calibration point that is obtained;

(9) utilize the calibration point that obtains that each vision sensor of whole multiple vision sensor detection system is demarcated, obtain the position relation of each vision sensor, finish global calibration with respect to workpiece coordinate system;

(10) the storage calibration result uses in order to follow-up measurement.

Images